Corgi Labs

Overview Dashboard

2026-04-23T15:15:34.987Z

The Overview Dashboard is your home base in Corgi Intelligence. It provides an at-a-glance summary of your business health, powered by AI-generated Key Insights that surface the most important trends and opportunities in your payment data.

Dashboard Modes: Retail vs. Subscription

Corgi Intelligence automatically tailors your Overview Dashboard based on your business model:

Retail/e-commerce merchants see metrics focused on Gross Merchandise Value (GMV), transaction counts, refund rates, average order value, customer lifetime value, and top products by revenue.
Subscription/usage-based merchants see metrics focused on Monthly Recurring Revenue (MRR), net revenue retention, gross revenue retention, logo churn rate, dunning recovery rate, deferred revenue balance, and revenue at risk.

Corgi Labs auto-detects the best dashboard mode for your business during onboarding. If a different view would be more appropriate, reach out to us and we will update your configuration.

Dashboard Metrics

Retail Dashboard

If your business primarily sells one-time products or services (e-commerce, marketplaces, physical goods), you will see the Retail overview.

Metric	Definition
Gross Merchandise Value (GMV)	Total value of all transactions processed in the selected period.
Number of Transactions	Total count of payment attempts.
Refund Rate	Percentage of transactions that were refunded, with trend comparison.
Number of Unique Customers	Distinct customers who made at least one purchase.
Repeat Purchase Rate	Percentage of customers making 2 or more purchases.
Customer Lifetime Value (CLV)	Average revenue per customer over their relationship with your business.
Average Order Value (AOV)	Mean transaction amount per order.
Top 10 Products by GMV Contribution	A ranked table showing your best-performing products by revenue, including purchase count, repeat customers, and percentage of total revenue.

Subscription / Usage-Based Dashboard

If your business operates on a recurring or usage-based billing model (SaaS, subscriptions, AI token usage), you will see the Subscription overview.

Metric	Definition
Monthly Recurring Revenue (MRR)	Your predictable monthly revenue from active subscriptions.
Net Revenue Retention (NRR)	Measures expansion, contraction, and churn relative to your starting revenue. An NRR above 100% indicates growth from existing customers.
Gross Revenue Retention (GRR)	Revenue retained from existing customers, excluding expansion. Reflects how well you hold onto revenue before upsells.
Logo Churn Rate	Percentage of customers who canceled during the period.
Dunning Recovery Rate	Percentage of failed subscription payments that were successfully recovered through retry and dunning flows.
Deferred Revenue Balance	Revenue collected but not yet recognized, important for accounting and forecasting.
Revenue at Risk	Revenue associated with customers showing churn signals or failed payments.

Key Insights

At the top of the Overview page, Corgi surfaces AI-generated Key Insights: automated observations about your most important payment trends. These insights are generated from your latest data and highlight items such as:

Approval Upside: Estimated additional revenue you could unlock by improving your approval rate, with specific dollar estimates.
Disputes Down / Up: Notable changes in your dispute rate compared to the previous period, including breakdowns by primary vs. comparison range.
Churn Spike: Alerts when churned revenue exceeds historical norms, with recommended actions (for example, aligning dunning and card updater efforts).
Rule Bias: Warnings when your fraud rules may be generating too many generic declines, with suggestions to refine rules and avoid suppressing legitimate customers.

Key Insights update automatically. They are designed to draw your attention to the most impactful changes without requiring you to dig through every chart.

Payments & Fraud Overview

Both dashboard modes include a Payments & Fraud Overview section at the bottom of the page, showing:

Metric	Definition
Authorization Rate	Percentage of payment attempts approved by the issuing bank.
Dispute Rate	Percentage of transactions that resulted in a customer dispute.
Block Rate	Percentage of transactions blocked by your fraud rules or risk screening.
Abandonment Rate	Percentage of checkout sessions that were abandoned before payment completion.

Each metric includes a trend comparison against the previous period so you can spot changes at a glance. For deep dives, refer to Payment Analytics and Dispute & Fraud metrics, respectively.

Payment Analytics

2026-04-23T15:14:57.045Z

The Payment Analytics page provides a comprehensive view of your payment performance from checkout to final outcome. Use it to understand your payment funnel, identify where transactions fail, and find opportunities to improve your authorization rate.

Payment Funnel

The Payment Funnel visualizes the complete transaction flow from checkout through risk screening, 3D Secure authentication, and final outcome. Each stage shows both the count and the dollar volume, so you can see both the frequency and financial impact at every step. The funnel has four sequential stages:

Checkout: Total checkout sessions initiated by customers, and how many lead to attempted payments.
Risk Model: Transactions evaluated by fraud prevention services, typically offered by your payment provider or a 3rd party, including by Corgi Labs. Shows how many transactions were approved vs. blocked.
3DS: Transactions routed through 3D Secure authentication, performed by the card issuer. Shows No 3DS (not required), 3DS Passed, and 3DS Failed counts.
Outcome: Final result of the payment: succeeded or declined by the issuer.

Core Payment Metrics

Metric	Definition
Payment Success Rate	The overall percentage of payment attempts that resulted in a successful charge. This metric accounts for all failure points, including fraud blocks, 3DS failures, and issuer declines. Trend comparison against the previous period is shown.
Authorization Rate	The percentage of payment attempts that were authorized (approved) by the issuing bank, after passing through risk screening and 3DS. This isolates network-level acceptance from fraud blocks performed by your payment and fraud prevention providers.

Block, Decline & Abandonment Rates

Metric	Definition
Block Rate	Percentage of transactions blocked by your fraud rules or risk model before reaching the payment network.
Decline Rate	Percentage of transactions declined by the card network or issuing bank.
Abandonment Rate	Percentage of checkout sessions abandoned by the customer.

Block Reasons

This section breaks down why transactions were blocked during risk screening. A stacked bar chart shows monthly volumes by block source.

3D Secure Authentication

This section examines 3DS challenge rates and authentication performance.

Metric	Definition
3DS Rate	Percentage of transactions routed through 3D Secure.
3DS Success Rate	Percentage of 3DS that were successful.
Challenge Rate	Percentage of 3DS transactions that triggered an active challenge (as opposed to frictionless authentication).
Challenge Success Rate	Percentage of 3DS challenges that were successful.

A monthly breakdown chart shows Successful Challenges, Successful Frictionless, and Failed Challenges over time. A summary table shows the count, share, and request volume for each outcome.

Decline Reasons Distribution

A stacked bar chart shows decline counts by reason category over time. Use this to identify whether declines are driven by preventable issues (like expired cards) or systemic problems (like issuer-side fraud flags).

Geographic Payment Distribution

A world map visualization shows payment volume by card-issuing country. This helps you understand the geographic distribution of your customers and identify regions with unusually high decline or fraud rates.

Deduplication Setting

Payment metrics can be viewed as deduplicated or raw:

Deduplicated: Count only the final outcome per payment, which removes duplicate retries. Multiple attempts are considered the same payment when they have:
- Same invoice_id for subscription payments, or
- Same Customer + close-in-time + same amount, or
- Same card number + close-in-time + same amount
Raw: Count all payment attempts.

Tracking the same metrics across the same setting allows a fair evaluation of the metrics' performance over time. Depending on your business model, a significant difference between deduplicated metrics and raw metrics may indicate increased customer checkout friction.

Customer Clusters

2026-04-23T15:14:22.818Z

The Customer Clusters page uses AI-powered segmentation to group your customers by spending behavior and purchase cadence. It answers three practical questions for payment ops and marketing teams: who are your customers, how do they behave, and where should you concentrate retention and acquisition effort. Because the clustering is refreshed as new transactions flow in, the segments reflect current behavior rather than a static snapshot.

Customer Segmentation Clusters

Corgi automatically identifies distinct customer segments based on two dimensions:

Average Order Value (x-axis): how much the customer spends per transaction.
Purchase Frequency (y-axis): how often the customer makes purchases, measured per month.

A scatter plot visualizes all customers, color-coded by their assigned cluster. This makes it easy to see the natural groupings in your customer base at a glance, and to spot outliers who may warrant individual attention.

Cluster Definitions

High-Value Customers

Premium customers with high spending and frequent purchases. These are your most valuable customers.

Characterized by above-average order values and high purchase frequency.
Strategy: prioritize retention, offer loyalty rewards, and ensure premium support.

Loyal Frequent Buyers

Regular customers with moderate spending but high engagement. They purchase often, even if individual order values are modest.

Strategy: encourage upselling and cross-selling to increase AOV.

Price-Conscious Buyers

Budget-focused customers who respond well to discounts and promotions. They tend to have lower order values and less frequent purchases.

Strategy: use targeted promotions and discount campaigns to drive repeat purchases.

New Explorers

Recent customers who are still evaluating your products. They have low frequency and moderate order values.

Strategy: focus on onboarding, first-purchase follow-ups, and welcome campaigns to convert them into repeat buyers.

Dormant Customers

Previously active customers who have not purchased recently. They represent potential win-back opportunities.

Strategy: launch re-engagement campaigns, win-back offers, or surveys to understand why they left.

Cluster Summary Cards

Each cluster is summarized in a card with four at-a-glance metrics:

Metric	What it shows
Total Revenue	Aggregate revenue from all customers in the segment.
Avg Order Value	Mean transaction amount for the segment.
Customers	Number of customers in the segment.
Frequency	Average purchases per month.

Together these four numbers give you a quick read on both the size and the economic weight of each segment, so you can tell at a glance whether a cluster is a small group of heavy spenders or a large group of light spenders.

Segment Performance

A combined bar and line chart shows the number of customers and total revenue for each segment side by side. This helps you quickly compare which segments contribute the most revenue and which have the most customers. The gap between the two series is often the most revealing part: a segment with many customers but little revenue points to an upsell opportunity, while a segment with few customers but outsized revenue points to a retention priority.

Purchase Behavior by Segment

A chart comparing average purchases per month and average order value across segments. This makes it easy to see which segments buy more often vs. which spend more per order, and to decide whether a given segment is best influenced through frequency plays (subscriptions, reminders, replenishment nudges) or basket-size plays (bundles, cross-sells, tier thresholds).

Customer Overview

2026-04-23T15:13:34.682Z

The Customer Overview page helps you analyze customer behavior, retention patterns, and identify your most valuable customers.

Customer Acquisition, Retention & Churn Analysis

The main chart provides a combined view of customer trends over time.

Metric	Definition
Total Customers	The total number of unique customers in the selected period.
New Customers	First-time buyers during the period.
Returning Customers	Customers who made at least one prior purchase.
Churn Rate	Percentage of customers who purchased in the previous period who did not return in the next period.

This visualization helps you understand the balance between acquisition and retention over time. A rising churn rate alongside flat new customer growth is a signal that retention strategies need attention.

Repeat Purchase Rate

The Repeat Purchase Rate shows the percentage of customers making 2 or more purchases. This is a key loyalty metric. A higher repeat rate generally indicates strong product-market fit and effective retention.

Customer Lifetime Value (CLV)

The Customer Lifetime Value metric shows the average total revenue per customer across their entire relationship with your business. This helps you understand how much each customer is worth and informs decisions about acquisition spend and retention investment.

Top Frequent Customers

A detailed table lists your most active customers, filtered to show those with strong repeat purchase volume and fewer disputes.

Column	Definition
Customer name and email	Identifies the customer.
Purchases	Total number of transactions.
Disputes	Number of disputes filed by this customer.
Total Spent	Cumulative spend.
Avg. Order	Average transaction value.
Last Purchase	Date of most recent transaction.

This table is useful for identifying VIP customers, spotting potential fraud patterns (high purchase volume combined with high dispute counts), or building targeted retention campaigns.

Product Performance

2026-04-23T15:13:02.006Z

The Product Performance page helps you understand which products drive the most purchases, revenue, and customer engagement.

Top Products by Purchase Volume

A bar chart ranks your products by total number of purchases during the selected period. This helps you identify your best sellers and spot any underperformers. The ranking reflects only purchases during the selected period, so adjust the period filter to change the ranking window.

Revenue Distribution by Category

A donut chart shows how revenue is distributed across your product categories. This gives you a high-level view of category concentration. For example, if 100% of your revenue comes from a single category, this signals an opportunity (or risk) depending on your business strategy. Use this view to gauge category concentration risk at a glance.

Product Performance Details

A paginated table provides a complete breakdown of every product. The table is sortable by any column and paginated to support businesses with large product catalogs.

Column	Definition
Product Name	The name or description of the product.
Category	The product category.
Purchases	Total number of transactions for this product.
Repeat Customers	Number of customers who purchased this product more than once.
Revenue	Total revenue generated by this product.
% of Total	This product's share of your total revenue.

How to Use This Data

Identify top performers. Focus marketing spend on products already generating strong revenue and repeat purchases.
Spot products with high repeat rates. Products with high repeat customer counts may be good candidates for subscription or bundle offers.
Find underperformers. Products with low purchase counts or revenue share may need pricing adjustments, better descriptions, or promotional support.

Disputes & Fraud

2026-04-23T15:12:23.094Z

The Disputes and Fraud page gives payment operations teams a single place to monitor dispute activity, fraud patterns, and the financial drag both create. Use it to spot emerging trends, trace fraud back to its source, and act on the levers that reduce chargebacks before they compound.

Dispute, Fraud, and Chargeback Rate & Count

The top of the page surfaces two headline metrics side by side. The Dispute, Fraud, and Chargeback Rate expresses the share of transactions that ended in a dispute, fraud case, or chargeback as a percentage. The Count shows the raw number of cases behind that percentage, so you can tell whether a rate shift reflects a handful of incidents or a meaningful volume change.

Both metrics include a trend comparison against the previous period. A rising rate paired with a rising count points to a real increase in dispute pressure, while a rising rate on a falling count usually reflects lower transaction volume rather than worsening fraud exposure.

Dispute Win Rate

The Dispute Win Rate is the percentage of disputes you have won. A low win rate is a signal worth investigating. It often points to gaps in the evidence you submit, delays in your response workflow, or documentation that does not map cleanly to the reason codes issuers expect. Treat a sustained dip as a prompt to audit your representment process rather than a one-off outcome.

Estimated Loss Analytics

This section quantifies the financial impact of disputes and fraud. The estimated loss is calculated as three times the disputed amount, a multiplier that captures the operational and financial costs that sit alongside the face value of the dispute itself: fees, recovery labor, merchandise loss, and downstream risk exposure.

Watching this figure next to raw dispute counts helps you prioritize. A small number of high-value disputes can outweigh a larger volume of low-value ones, and the loss view makes that tradeoff explicit.

Fraud Breakdowns

Dispute Reason Breakdown

A stacked bar chart plots dispute counts by reason category over time. The categories are Fraud, Product Not Received, Product Unacceptable, Duplicated Charge, and All Other Reasons. The shape of the stack tells you whether your disputes are primarily fraud-driven or operational, which in turn tells you whether to lean on fraud tooling or on fulfillment and billing fixes.

Payment Method Distribution

This view breaks disputes down by payment method, including JCB, Visa, Mastercard, and American Express. Three callouts sit above the chart:

Highest Fraud Rate by method
Most Disputes by method
Safest Payment Method

A bar chart beneath the callouts compares Fraud versus Other Disputes for each payment method, so you can see whether a method's risk profile is driven by fraud specifically or by a broader mix of dispute types.

Payment Method Details Table

The details table gives you the full numeric picture for each payment method:

Column	What it shows
Dispute Rate	Share of transactions on this method that became disputes
Dispute Count	Number of disputes on this method
% of Total Disputes	This method's share of all disputes
Fraud Rate	Share of transactions on this method flagged as fraud
Fraud Count	Number of fraud cases on this method
% of Total Fraud	This method's share of all fraud cases

Country Breakdown

Switch to the Countries tab to see the same patterns mapped by geography. Country-level concentration often surfaces risk signals that payment-method slicing hides, especially for cross-border volume.

Rate Basis Setting

You can toggle the basis used to calculate the rates on this page:

Dispute date. Rates are calculated based on when the dispute was filed.
Transaction date. Rates are calculated based on when the original transaction occurred.

Dispute-date basis reflects what your operations team is handling right now. Transaction-date basis reflects the quality of the transactions you accepted in a given period, which is the right lens for evaluating acceptance and fraud-screening changes.

AI Insights & Reports

2026-04-23T15:07:20.138Z

The AI Insights & Reports page provides access to AI-generated payment insights reports that analyze your payment data and deliver actionable recommendations.

How It Works

Corgi Labs' AI engine periodically analyzes your payment data and generates comprehensive reports that summarize key trends, anomalies, and optimization opportunities. These reports are designed to save your team hours of manual analysis by surfacing the most important findings automatically.

Report Contents

Each AI Insights report typically covers the following areas:

Section	What it covers
Approval Rate Analysis	How your approval rate changed compared to the previous period, including daily trends and contributing factors.
Decline & Block Trends	Changes in decline rates, block rates, and the reasons driving them.
Transaction Volume	Number of approvals out of total transactions and how this compares historically.
Anomaly Detection	Unusual patterns or spikes that may require attention.
Optimization Recommendations	Specific, actionable suggestions to improve payment outcomes.

Report List

The main page shows a table of all generated reports with the following columns:

Column	Description
Period	The date range the report covers (e.g., Apr 6, 2026 to Apr 12, 2026).
Insights Preview	A brief summary of the key finding.
Generated	When the report was created.
Actions	View the full report or Download it.

Report Cadence by Plan

Report frequency depends on your Corgi Intelligence plan:

Plan	Report Frequency
Core	1 report per month
Pro	1 report per week (on-demand coming soon)
Enterprise	On-demand generation

Using Insights Effectively

AI Insights reports are most valuable when paired with the operational tools in Corgi Intelligence. Follow this sequence to turn a report into action:

Read the report to understand what changed and why.
Navigate to the relevant analytics page (e.g., Payment Analytics, Disputes & Fraud) for deeper investigation.
Take action using Payment Tools: adjust fraud rules, update block/allow lists, or respond to disputes based on the report's recommendations.

Dispute Management

2026-04-23T14:59:38.745Z

The Dispute Management page is your central hub for managing active disputes, tracking outcomes, and preventing chargebacks with automated early fraud warning responses.

Overview Metrics

At the top of the page, you'll see key metrics for the selected period:

Dispute Rate & Count. Current dispute rate and total number of disputes, with trend comparison.
Dispute Win Rate. Percentage of disputes you have won, helping you assess the effectiveness of your response strategy.

Both metrics include trend arrows and can be toggled between dispute date and transaction date basis.

Outcome Breakdown

A stacked bar chart shows monthly dispute outcomes, categorized as:

Needs Response. Disputes awaiting your action.
Under Review. Disputes with submitted evidence being reviewed.
Lost. Disputes decided in the customer's favor.
Won. Disputes decided in your favor.

A summary table below shows each outcome's count and share of total disputes. Use this to track whether your response rate and win rate are improving over time.

Active Disputes Requiring Action

This table lists all disputes that currently require your attention. Each dispute row carries six fields:

Column	What it shows
Dispute ID	Unique identifier for the dispute.
Customer	Customer name.
Amount	Disputed amount.
Reason	The dispute reason (e.g., fraudulent, product not received).
Days Left	Remaining time to respond. 0 days means the deadline is imminent.
Priority	Color-coded priority (High, Medium, Low) based on urgency and amount.

The table is paginated and sortable. Focus on High-priority disputes with 0 days remaining first.

Historical Disputes

Below the active disputes, a Historical Disputes section lets you review past disputes, their outcomes, and resolution dates. Use this for reporting, identifying patterns, and improving your dispute response process.

List Management

2026-04-23T14:51:00.000Z

The List Management page is where payment operators build and maintain the reusable value lists that power Corgi's fraud detection rules. Every allow list, block list, and attribute collection lives here, whether it ships as a Corgi default or you author it yourself. Spend a few minutes on this page and the rules you write elsewhere in Intelligence become dramatically easier to reason about, because each rule can simply ask "is this value in that list?" instead of hard-coding individual values inline.

What Are Lists?

Lists are collections of values (emails, card fingerprints, IP addresses, customer IDs, etc.) that can be referenced in your fraud rules. For example, you might create an email block list containing known fraudulent email addresses, and then create a rule that blocks any transaction matching an entry in that list.

This indirection is the whole point. One list can feed many rules, and updating the list propagates everywhere it is referenced. When a new bad actor shows up, you add the value once rather than editing every rule that needs to know about it.

Default Lists

Corgi provides several default block lists out of the box, pre-populated with known bad actors. Default lists are managed by the system and updated automatically, so you inherit ongoing threat intelligence without operational overhead.

List	Purpose
Card Fingerprint Block List	Known fraudulent card fingerprints.
Email Block List	Known fraudulent email addresses.
Email Domain Block List	High-risk email domains.
Client IP Country Block List	Countries associated with high fraud rates.
Customer ID Block List	Known fraudulent customer identifiers.
Charge Description Block List	Suspicious charge descriptions.
Card Country Block List	Card-issuing countries to block.
Client IP Address Block List	Specific IP addresses to block.

Because these lists are system-managed, you cannot edit their entries directly, but you can reference them from your own rules exactly like any custom list.

Custom Lists

You can create custom allow and block lists for any supported attribute type. Custom lists are where your team's institutional fraud knowledge accumulates: the email domain that keeps chargebacks, the BIN range tied to a gift-card abuse pattern, the customer ID you want to always approve.

Attribute Type	Description
Card Fingerprint	Unique card identifier.
Email	Customer email address.
Email Domain	Domain portion of the email (case-insensitive matching).
Customer ID	Your internal customer identifier.
Country	Country code.
IP Address	Client IP address.
Charge Description	Transaction description string (case-sensitive or case-insensitive matching).
Card BIN	Bank Identification Number (first 6 to 8 digits of card number).
String	Flexible string matching for custom attributes.

Choose the type that matches the attribute your rules will evaluate. Email Domain and Charge Description are worth calling out because they offer case-insensitive (and for Charge Description, case-sensitive) matching options, which matters when fraudsters vary capitalization to evade exact-match checks.

List Details

The main table shows all lists with:

Column	Description
Name	List name and whether it's a Default list.
Alias	The programmatic reference used in rule conditions.
Type	The attribute type the list matches against.
Items	Number of entries in the list.
Created By	Who created the list (System or a team member).
Created	When the list was created.

The Alias column is the one you will use most often when authoring rules, because it is the short identifier your rule conditions actually reference. Pick aliases that read well inline, because they will appear in every rule that consumes the list.

Creating a New List

Click + New List to create a custom list. You'll need to specify:

Field	Description
Name	A descriptive name.
Alias	A short identifier for use in rule conditions.
Type	The attribute type.
Entries	Add values manually or import them in bulk.

Bulk import is the faster path when you are seeding a list from an existing spreadsheet, CSV export, or fraud analyst's investigation notes. Manual entry is better for one-off additions discovered in day-to-day review. Once the list exists, you can add or remove entries at any time, and every rule referencing the list picks up the change on the next evaluation.

Sales Tax

2026-04-23T14:48:38.915Z

The Sales Tax page helps you track sales tax thresholds and estimated tax due by jurisdiction, covering US states and international markets. Use it to see where your revenue and transaction counts stand against each jurisdiction's economic nexus rules, and to understand when collection obligations are triggered.

Jurisdiction Tax Calculation

The main table provides a comprehensive view of your sales tax obligations across all relevant jurisdictions. Each row corresponds to one jurisdiction, and the columns describe the data Corgi surfaces against it.

Column	What it shows
Jurisdiction	The state, territory, or country name.
Tax Type	The type of tax applicable (Sales Tax, Local Sales Tax, GST, GET, GRT, JCT, TPT, IVU, etc.).
Tax Rate	The applicable tax rate for the jurisdiction.
Revenue	Your total revenue in that jurisdiction during the selected period.
Transactions	Number of transactions in the jurisdiction.
Threshold	The economic nexus threshold that determines when you are required to collect and remit tax. Thresholds are typically based on revenue amount, transaction count, or both (for example, "$100,000 OR 200 transactions").
Threshold Status	Whether you have met the threshold. Threshold Met means you need to collect tax; Not Met means you are not yet obligated.

Read each row as a snapshot of your current exposure in that jurisdiction for the selected period. The Threshold Status column is the fastest signal for where action may be required.

Covered Jurisdictions

Corgi tracks thresholds and obligations for:

All US states with sales tax obligations, including states with no sales tax (Delaware, Montana, New Hampshire, Oregon) which are shown with a "No nexus threshold" indicator.
US territories such as Puerto Rico and the District of Columbia.
International markets including Japan (JCT), Singapore (GST), and others depending on where your customers are located.

Coverage expands as your customer footprint grows, so jurisdictions appear in the table once you have qualifying activity in them.

Understanding Thresholds

Economic nexus thresholds vary by jurisdiction. Common formats include:

Revenue-only. For example, $100,000 in annual sales.
Revenue OR transaction count. For example, $100,000 OR 200 transactions (meeting either triggers the obligation).
Revenue AND transaction count. For example, $100,000 AND 200 transactions (both must be met).
No threshold. Some jurisdictions have no economic nexus threshold, meaning any sale may trigger obligations.

Corgi automatically calculates your revenue and transaction counts against each jurisdiction's thresholds and indicates which ones you have reached. This lets you monitor progress toward nexus before you cross it, rather than discovering the obligation after the fact.

Important Notes

Sales tax calculations are estimates based on your transaction data. Consult a tax professional for compliance decisions.
Corgi tracks thresholds based on approved transactions. Refunded or disputed amounts may affect your actual obligations.
Threshold rules change periodically. Corgi updates threshold information regularly, but you should verify current rules with the relevant tax authority.

You Can't See Your Agent Channel Yet. Here's How to Fix That.

2026-04-16T21:45:00.000Z

Every agent-initiated purchase on Stripe looks identical to a human-initiated one after settlement. That's becoming an expensive blind spot.

Stripe has shipped the primitives you need to sell through AI agents: Shared Payment Tokens (SPTs), the Agentic Commerce Protocol, agentic network tokens from Visa and Mastercard, BNPL over SPT, and the Agentic Commerce Suite. The first merchants to go live (Coach, URBN, Ashley Furniture, and Kate Spade) are processing agent traffic from confirmed integrations like ChatGPT Instant Checkout and Microsoft Copilot Checkout, with additional partners like Perplexity in earlier stages.

Ask a payments lead at one of those merchants how their agent channel is performing. They can't tell you. The data isn't there.

The Persistence Problem

Here's the technical reality most teams haven't caught up to yet.

When an agent initiates a purchase, Stripe creates a Shared Payment Token: an spt_-prefixed object scoped to a single transaction, time-limited, and carrying Radar risk signals from the agent side. The merchant confirms a PaymentIntent with the SPT.

But at confirmation time, Stripe clones the underlying payment method and sets the PI's payment_method field to the clone. The SPT is consumed. The resulting charge object that lands in your Sigma tables, your BigQuery sync, or your data warehouse looks structurally identical to a normal card-on-file transaction.

As of April 2026, there is no is_agentic field on the charge. No channel = "agent" enum on the PaymentIntent. No agent platform attribution. The distinction between a checkout inside ChatGPT and one on your own website disappears in most standard reporting surfaces the moment the PI settles.

This is a defensible engineering choice. It keeps the downstream reporting surface consistent, refunds behave the same way, and legacy integrations don't break. But it leaves every merchant on the Agentic Commerce Suite unable to answer a basic question: how is my agent channel performing?

Why "Just Tag the Metadata" Isn't Enough

The obvious workaround is to tag the PaymentIntent metadata yourself, setting metadata.channel = "agentic" at the moment of PI creation. Your backend knows this is an SPT flow because it's running a different code path. One extra line.

In practice, this fails for three reasons:

It requires engineering effort. Agent traffic is still a small enough share of revenue that it rarely makes the sprint.
Tagging conventions drift. Some teams use agentic, some use agent, some use ai_channel. Reconciling across them becomes its own cleanup project.
It only covers new integrations. Every merchant already live on the Suite has a back catalog of agent transactions that were never tagged. You can't retroactively fix that.

The better answer is to detect agent transactions passively, without asking the merchant to change anything.

What We Built

Corgi's agent payments intelligence layer sits on top of the webhook subscriptions we already maintain on merchant Stripe accounts.

When a merchant confirms a PaymentIntent with an SPT, Stripe fires webhook events related to the shared payment token lifecycle. Our pipeline correlates these events with the resulting PaymentIntent to identify SPT-originated transactions. We write the association to a dedicated table (agentic_transactions) with the SPT ID, PaymentIntent ID, merchant account ID, and the event timestamp.

From there, every downstream metric in our pipeline gets an is_agentic flag through a join on PaymentIntent ID.

No merchant code changes. No metadata discipline. No retroactive cleanup. If we have webhook access to your Stripe account, we detect agentic transactions from the moment we connect. We can't recover pre-integration history, but from that point forward, no agent transaction goes untagged.

The same pipeline captures SPT revocation and expiry events, which matter for a specific class of agent fraud covered below.

What the Analytics Show You

Once the detection layer is running, the real question becomes: what do agent transactions actually look like compared to human ones?

Authorization Rates Diverge

Early agent traffic on SPTs tends to authorize higher than card-on-file for two reasons. Many SPT transactions use network tokens, which issuers trust more than raw PANs. Mastercard reports a 2.1% auth rate lift for tokenized transactions, and Visa cites 4.6%.

The risk signals Stripe forwards with the SPT (card testing likelihood, stolen card indicators, chargeback history) are also stronger than what issuers typically see on a standard ecommerce transaction. If your auth rate on agent traffic is lower than your card-on-file baseline, something is likely wrong with your integration, not with the channel.

Decline Distributions Are Expected to Shift

Issuers are rolling out new authorization logic for Visa Intelligent Commerce and Mastercard Agent Pay. Based on early patterns, we expect the mix of do_not_honor, insufficient_funds, invalid_account, and network-declined reasons to look different for agent traffic.

If your retry logic is tuned to human-channel declines, it will likely make the wrong call a meaningful fraction of the time.

Dispute Rates Are the Headline Metric

Stripe has publicly reported near-zero fraud on the first wave of Suite merchants. That's accurate for the early cohort: enterprise brands with mature fraud programs and a filtered customer base (buyers logged into Copilot or ChatGPT with saved payment methods).

As agent traffic scales past early adopters, expect the chargeback rate to move in ways that differ from human patterns. Two scenarios illustrate why:

Scope disputes. "I didn't authorize this purchase" claims get murkier when a buyer delegated authority to an agent but says the agent exceeded scope.
Signal loss. Traditional fraud signals vanish: no browser fingerprint, no mouse movement, no device telemetry. Fraud models tuned on human traffic will produce false declines on legitimate agents and miss adversarial ones at the same time.

Platform Attribution Matters for GTM

Consider a merchant processing $50 million annually through the Suite. If 80% of their agent volume comes from ChatGPT and 5% from Copilot, that's a very different partnerships conversation than a 40/40 split. Without channel-level data, these decisions get made on gut feel.

The Fraud Model Gap

Every fraud model in production today was trained on human payment traffic. Those models rely on signals that don't exist in agentic flows:

How the user navigated to checkout
What device they used
How they typed their card number
What else they did on the page

In an agentic world, the "user" is an API call. There is no device, no session, no typing cadence.

This creates two failure modes at once. Legitimate agent traffic gets scored as suspicious because it doesn't look human, leading to block rates that cost real revenue. Adversarial agents that do look human enough (scripted to mimic human behavior) slip past models never designed to catch them.

The answer isn't to throw out your fraud model. It's to know which transactions need to be scored by which model. That requires three things: a reliable channel flag, a drift monitor that alerts you to distribution shifts on agent traffic, and a retraining trigger indexed on agent volume crossing a threshold. Corgi Intelligence provides all three.

Why We Built This First

Our merchants started asking. Not in the abstract, but with specific questions.

"Can you tell me what my chargeback rate looks like on ChatGPT orders versus direct checkout?"

That question had no answer in Stripe's dashboard. As far as we've been able to determine, it had no answer in other third-party tooling either. We shipped the first version in a week because the lift was small and the need was obvious.

We're among the first fraud intelligence platforms with a production detection layer for agentic commerce transactions on Stripe. Not because the idea was hidden, but because the infrastructure we already had (live webhook ingestion across merchant Stripe accounts) turned out to be the right foundation. We didn't build new plumbing. We recognized that one additional event type closed the analytics gap.

Where This Goes Next

Agent payments intelligence is the starting point. The longer-term product is a fraud model explicitly trained on agentic traffic, one that uses SPT risk signals, agent platform attribution, and behavior patterns absent from any human-channel dataset.

The detection layer makes that training set possible. Every merchant we onboard adds labeled agent transactions to a growing corpus of agentic commerce data, the foundation for models purpose-built for a channel that barely existed 18 months ago.

If you're running a Stripe integration and you're live on the Agentic Commerce Suite (or about to be), the question worth asking your team this week is simple. When the board asks how our agent channel is performing, what exactly will we show them?

If the answer is "we'll figure it out later," we should talk.

Be part of our CLOSED BETA

Corgi Labs is a Y Combinator-backed payments intelligence platform built natively on Stripe. We help merchants detect fraud, recover false declines, and see their agent channel clearly.

Sources

Stripe, "Introducing the Agentic Commerce Suite" — https://stripe.com/blog/agentic-commerce-suite
Stripe, "Agentic Commerce Suite" (newsroom announcement) — https://stripe.com/newsroom/news/agentic-commerce-suite
Stripe, "Shared Payment Tokens" (documentation) — https://docs.stripe.com/agentic-commerce/concepts/shared-payment-tokens
Stripe, "Developing an Open Standard for Agentic Commerce" — https://stripe.com/blog/developing-an-open-standard-for-agentic-commerce
Stripe, "10 Things We Learned Building for the First Generation of Agentic Commerce" — https://stripe.com/blog/10-lessons
Stripe, "Stripe Powers Instant Checkout in ChatGPT" — https://stripe.com/newsroom/news/stripe-openai-instant-checkout
Stripe, "Microsoft Copilot and Stripe" — https://stripe.com/newsroom/news/microsoft-copilot-and-stripe
Stripe, "Supporting Additional Payment Methods for Agentic Commerce" — https://stripe.com/blog/supporting-additional-payment-methods-for-agentic-commerce
Solidgate, "Network Tokenization and Authorization Rates" — https://solidgate.com/blog/network-tokenization-authorization-rates/
Mastercard, "Agent Pay: Pioneering Agentic Payments Technology" — https://www.mastercard.com/global/en/news-and-trends/press/2025/april/mastercard-unveils-agent-pay-pioneering-agentic-payments-technology-to-power-commerce-in-the-age-of-ai.html
Y Combinator, "Corgi Labs" — https://www.ycombinator.com/companies/corgi-labs

Corgi Labs VS Pagos

2026-04-14T21:21:15.311Z

COMPARISON

If you’re evaluating payment analytics platforms, Pagos is probably on your shortlist. Their founding team includes former Braintree and PayPal leaders, and they’ve become a go-to for cross-PSP data normalization.

But here’s the question most payment leaders eventually ask: once you can see the problem, who actually fixes it?

That’s where Corgi and Pagos take fundamentally different paths.

What Pagos Does Well

Pagos is a payments intelligence platform built around three product pillars: Insights and Benchmarking, Cost Optimization, and BIN Data. They also offer Card Network APIs for account updating, network tokenization, and BIN-based routing.

Their Insights product consolidates payments data across processors and surfaces analytics through dashboards, anomaly alerts, and Pagos AI (a conversational interface for querying your data). Cost Optimization breaks down fees, effective rates, and penalties at the transaction level. Their BIN Data product supports eight and nine-digit ranges with global coverage and weekly updates.

If your primary need is consolidating data from several PSPs into one place, Pagos delivers. Their benchmarking compares your performance against anonymized industry averages. Their data export pipeline pipes normalized payments data into your warehouse. And their Card Network APIs handle card lifecycle tasks like keeping card-on-file data current and managing network tokens.

Where Pagos Stops

Pagos gives you intelligence, card lifecycle tools, and AI-powered analytics. What it doesn’t do is make real-time fraud decisions. There are no custom ML models trained on your transaction data, no live approve/decline logic, and no per-decision explainability.

Pagos AI can surface insights about your chargebacks and flag anomalies. But it can’t look at an incoming transaction and decide in 100 milliseconds whether to approve or block it. That’s a different product category entirely.

So after Pagos helps you identify that your false decline rate is costing you revenue, you still need to:

Evaluate and purchase a separate fraud tool
Integrate it independently
Hope the data formats align
Manage two vendor relationships, two contracts, two support channels

Visibility matters. But without action, it’s an overhead cost that doesn’t move your numbers.

How Corgi Bridges the Gap

Corgi combines payment analytics and payment optimization on a single platform.

Corgi Intelligence gives you the analytics layer: revenue analytics, payment conversion funnels, customer clustering, dispute monitoring, AI-powered trend detection, and real-time alerts. Like Pagos, it includes AI that helps you explore your data and spot patterns.

Corgi Model is where the platforms diverge. It trains custom machine learning models on your transaction data (not industry aggregates) and makes real-time fraud decisions in under 100 milliseconds. Every decision comes with a full explanation of why a transaction was approved or declined. Your risk team isn’t operating on a black box.

The distinction matters: Pagos AI tells you what’s happening in your payments data. Corgi Model acts on live transactions in real time. One is a reporting tool. The other is a decisioning engine.

The two products share the same data layer. Your analytics inform your optimization. Your optimization results flow back into your analytics. No stitching together separate vendors.

Head-to-Head Comparison

Capability	Corgi	Pagos
Cross-PSP payment analytics	Yes	Yes
Real-time monitoring and alerts	Yes	Yes
Data normalization	Yes	Yes
AI-powered insights	Yes	Yes (Pagos AI)
Cost optimization analytics	Roadmap	Yes
Card Network APIs	No	Yes (Account Updater, Tokens, Routing)
Custom ML fraud models	Yes	No
Real-time fraud decisioning	Yes	No
Per-decision explainability	Yes	N/A
Anonymous benchmarking	No	Yes
Data warehouse export	Roadmap	Yes
Published pricing	Yes ($299/mo+)	Yes ($0 Free / $1,000/mo+)
Free tier or trial	30-day trial, full Pro	Free tier (25K txns/mo, one PSP)
Seats included	3 (Core) / 10 (Pro)	Unlimited

The Revenue Recovery Difference

False declines cost merchants an estimated $443 billion annually. Most of that isn’t fraud prevention gone right. It’s fraud prevention gone wrong, blocking legitimate customers who would have paid.

Corgi Model’s custom ML approach has demonstrated a 94.2% approval rate compared to 87.1% from generic fraud tools. At the same time, it cuts chargeback rates from 0.45% to 0.12%.

One Corgi customer, a mid-market eRetailer, recovered $2.4M in annual revenue after implementation. That came from fewer false declines, reduced disputes, and higher authorization rates.

The difference between “your approval rate is 3 points below industry average” and “here’s a model that lifts your approval rate by 7 points” is the difference between a report and a result.

How Pricing Compares

Both platforms publish transparent pricing. Here’s how they break down.

Pagos offers a free tier with one processor connection, 25,000 transactions per month, and one year of historical data. Their Growth plan is $1,000/month and adds a second processor connection, 100,000 transactions, two years of history, and cost optimization. Benchmarking is a $500/month add-on. Enterprise is custom. Every plan includes unlimited seats.

Corgi Intelligence starts at $299/month for Core (three seats, monthly reports, email support) and $999/month for Pro (10 seats, weekly reports, live chat with a three-hour SLA). Enterprise pricing is custom. There’s a 30-day free trial with full Pro features. No credit card required. No sales call required.

If you only need analytics and monitoring, Pagos’s free tier is a strong starting point. But if you need analytics and fraud optimization on the same platform, Corgi’s $299/month entry point gets you both in a single data layer. That’s a capability Pagos doesn’t offer at any price.

Who Should Choose Pagos

If your primary need is payments intelligence, cost visibility, and card lifecycle management (and you already have a fraud tool you’re happy with), Pagos is a strong choice. Their free tier is a low-risk way to consolidate PSP data. Cost Optimization gives you fee-level detail most platforms don’t. Their Card Network APIs handle account updating and tokenization. And unlimited seats means your entire payments team gets access from day one.

Who Should Choose Corgi

If you want analytics and optimization on one platform, especially if you’re a mid-market company doing $5M to $500M in annual payment volume, Corgi gives you both without the complexity of managing separate vendors.

Choose Corgi if:

You’re losing revenue to false declines and want custom ML that actually fixes it
You want to start with a $299/month analytics product and grow into fraud optimization
You need per-decision explainability for your risk team or compliance requirements
You want analytics and fraud optimization sharing one data layer, not two separate vendor integrations

Getting Started

Corgi Intelligence offers a 30-day free trial with full Pro features. No credit card, no sales call. Connect your PSP via OAuth and you’ll have your first dashboard in minutes.

Ready to dig up the revenue hiding in your payments data? Book a demo with our team.

Corgi Labs is a YC-backed, SOC 2 certified payment optimization platform.

Japan's Chargeback Rate Looks Perfect. That's the Problem.

2026-04-14T21:20:46.083Z

Your Japan PSP dashboard probably looks clean. Clearly Payments data shows Japan's chargeback rate at 0.18%, tied with China for the lowest in the world and less than half the US rate of 0.47%. That number is accurate, and it's also the wrong number to watch.

Behind that reassuring metric, Japan's National Police Agency reported ¥307.5 billion in tracked fraud losses for 2024, an 89.1% increase year over year. Credit card fraud alone set a new record. If you're managing authorization rates and fraud exposure across multiple processors in Asia-Pacific, Japan's chargeback rate is giving you a false signal of safety.

What Japan's Fraud Numbers Actually Show

The headline figure is striking, but the category breakdown tells you where the risk concentrates.

Japan's NPA tracked 57,324 fraud cases in 2024, up 24.6% from the prior year. The two fastest-growing categories are part of a broader tracked total, and they account for most of the surge: specialized fraud (phone scams and impersonation) at ¥71.8 billion, up 58.6%, and social media investment and romance fraud at ¥127.2 billion, which nearly tripled.

Credit card fraud, the category most directly relevant to ecommerce merchants, hit ¥55.5 billion in 2024. That's up from ¥54.1 billion in 2023 and ¥43.6 billion in 2022, according to the Japan Consumer Credit Association. The upward trend has been consistent and accelerating.

Here's the detail that matters most for card-not-present merchants: KOMOJU reports that 93.3% of Japan's credit card fraud in 2023 came from card number theft. The vast majority of card fraud is happening online, at checkout, where your fraud models make their decisions.

Why Chargebacks Stay Low While Fraud Climbs

How does a country post record fraud losses while maintaining the world's lowest chargeback rate? The gap isn't a data error. It's a structural feature of how Japan handles payment disputes.

Consider a merchant processing $10 million annually through a Japan-facing PSP. Your chargeback dashboard shows a rate well under 0.20%. You'd reasonably conclude your fraud exposure is under control. But three things are happening that chargebacks don't capture:

Japanese consumers are culturally less likely to initiate chargebacks. Stripe's Japan payments guide notes that Japanese issuers are slower to issue chargebacks compared to issuers in other countries, and that each one tends to receive more attention when it does occur. The low rate reflects consumer behavior, not low fraud.
Non-chargeback resolution paths absorb potential disputes. Japan's consumer protection infrastructure likely routes many fraud complaints through mediation and resolution channels outside the chargeback process, reducing the volume that reaches card networks.
The fastest-growing fraud categories rarely trigger a chargeback. Social engineering, account takeover, and investment scams involve victims who authorize the transaction themselves under false pretenses. Your fraud rules won't flag it. Your chargeback metrics won't register it.

The Criminal Infrastructure Has Changed

The fraud operators behind these numbers aren't who you'd expect. Japan's traditional organized crime groups, the yakuza, are no longer the primary fraud infrastructure.

The Japan Times reports that tokuryuu networks (anonymous, loosely organized criminal groups) have overtaken yakuza in fraud-related arrests. These groups recruit through social media "dark part-time job" listings, assemble for specific operations, and disband quickly. They drove more than 10,000 arrests in 2024.

For merchants, this matters because tokuryuu operations are fast, distributed, and harder to pattern-match. They run social media investment scams, phone fraud, and card fraud through rapidly assembled teams. The Japan Times and NPA data also link tokuryuu activity to card-not-present fraud schemes. The average loss per investment fraud case reached ¥13.6 million per victim. These are not low-sophistication attacks.

The Attack Surface Is Growing Fast

Japan's cashless transition is expanding the digital payment volume that fraud actors target. METI reported that cashless payments reached 42.8% of consumer spending in 2024, with the government pushing toward an 80% target.

The QR code payment market alone hit ¥21.5 trillion in FY2024, up 23.9% year over year according to Yano Research. PayPay, the dominant mobile wallet, reached 70 million registered users as of July 2025, commanding roughly 64% of QR and barcode payment volume.

More digital payment volume means more surface area for fraud. And the expansion isn't just domestic.

Japan welcomed a record 36.87 million inbound tourists in 2024, per the Japan National Tourism Organization. Cross-border card-not-present transactions introduce authentication patterns that generic fraud models handle with blunt, region-agnostic rules. Industry analysis consistently shows that region-specific risk patterns rarely generalize, which forces effective detection to rely on localized models.

The 3DS Mandate: Compliance That Creates a New Problem

Japan's April 2025 EMV 3DS mandate added urgency to all of this. The JCA Credit Card Security Guidelines now require 3D Secure authentication for all online credit card transactions. Merchants who haven't implemented 3DS bear full liability for fraud-related chargebacks.

That's the compliance side. The conversion side is the problem you need to watch.

The risk is real: 3DS can disrupt the checkout experience and suppress conversions if your implementation relies on challenge flows rather than risk-based authentication. Every transaction that hits a challenge screen is a chance for a legitimate buyer to abandon the purchase.

There's a precedent worth noting. When the EU implemented Strong Customer Authentication under PSD2, merchants initially saw conversion drops. But over time, many saw fraud decrease, trust grow, and acceptance rates improve after proper implementation. The outcome depends entirely on how well 3DS is configured, and that configuration depends on having accurate risk signals for Japan-specific transactions.

Why Generic Fraud Models Miss Japan

If your fraud detection runs on pooled data from a global processor, it wasn't trained on the patterns that define Japan's payment landscape.

JCB, Japan's domestic card network, illustrates the problem. In 2022, JCB launched the FARIS Joint Scoring Service with IWI and PKSHA Technology, the first shared fraud data scoring service built specifically for Japanese card issuers. JCB built FARIS because existing rule-based and generic detection tools were insufficient for Japanese transaction patterns.

Japan's payment mix includes behaviors that look anomalous to models trained on US or European data:

JCB-specific authentication flows that differ from Visa and Mastercard protocols
Konbini (convenience store) payments where customers complete online orders with cash at 7-Eleven or Lawson
Domestic mobile wallets like PayPay that process outside traditional card rails
Seasonal spending spikes during Golden Week, Obon, and year-end gift-giving periods that generic models read as velocity anomalies

A fraud model trained on pooled global data will either miss Japan-specific fraud patterns or over-flag legitimate Japanese buying behavior. Both outcomes cost you revenue.

What Merchant-Specific Modeling Changes

The alternative is a model trained on your transaction data, learning the patterns specific to your Japan-facing business.

As Ravelin's fraud research notes, it's best to use your own customer data for your business, since different business models can have very different customer order cycles and amounts. A merchant-specific model learns what normal looks like for your Japan customers, not for a global average.

This approach changes three things:

Fraud detection improves because the model recognizes Japan-specific attack patterns (card number theft at checkout, tokuryuu-driven account takeover) rather than applying global thresholds.
False declines drop because the model understands that a ¥150,000 order during Golden Week from a JCB cardholder is normal for your business, not a risk flag.
3DS configuration gets smarter because accurate risk scoring lets you route low-risk transactions through low-friction 3DS flows, maintaining conversion rates while staying compliant with the JCA mandate.

Your data also stays isolated to your business, which aligns with Japan's strict data handling expectations and financial regulatory standards.

Three Things to Audit Now

Japan's fraud numbers are climbing, chargebacks aren't reflecting it, and the 3DS mandate is changing the rules. Here's where to start:

Stop using chargeback rate as your Japan fraud indicator. Look at fraud-related declines, authorization rate gaps between Japan and your other markets, and transaction velocity anomalies that your current tooling might classify as normal.
Audit your 3DS implementation for conversion impact. If you're running challenge flows on all Japan transactions rather than risk-based authentication, you're likely losing real buyers. Measure your Japan authorization rate before and after the April 2025 mandate.
Evaluate whether your fraud model understands Japan. If your detection is processor-locked or trained on pooled global data, it's not seeing JCB-specific patterns, konbini payment flows, or domestic wallet behavior.

Getting visibility into what's actually happening in your Japan payments data is the first step. Corgi Labs builds custom fraud models trained on merchant-specific data and cross-processor analytics that surface the patterns generic tools miss. If your Japan exposure is growing and your chargeback rate looks too clean, it's worth digging into what that number isn't telling you.

Book a demo

Sources

Clearly Payments, "Chargeback Rate by Country in Payments" (April 2024) — https://www.clearlypayments.com/blog/chargeback-rate-by-country-in-payments/
Nippon.com, "Crime Figures in Japan Rise Again in 2024" — https://www.nippon.com/en/japan-data/h02649/
Japan Consumer Credit Association (JCA), Credit Card Fraud Survey 2024 — via Statista: https://www.statista.com/statistics/1232728/japan-credit-card-fraud-losses/
KOMOJU, "E-Commerce Fraud Protection Guide 2025" — https://en.komoju.com/blog/payment-method/e-commerce-fraud-protection/
Stripe, "How to Accept Payments in Japan" — https://stripe.com/resources/more/payments-in-japan-an-in-depth-guide
Japan Times, "New Tokuryuu Crime Groups Outpace Yakuza in Arrests" — https://www.japantimes.co.jp/news/2025/04/03/japan/crime-legal/npa-organized-crime/
METI, "2024 Ratio of Cashless Payment" (March 2025) — https://www.meti.go.jp/english/press/2025/0331_001.html
Yano Research, "QR Code Payment Market Reached 21 Trillion Yen in FY2024" — https://www.yanoresearch.com/en/press-release/show/press_id/3896
PayPay Corporation, "70 Million Registered Users" (July 2025) — https://about.paypay.ne.jp/en/pr/20250715/01/
Japan National Tourism Organization, 2024 Visitor Statistics — https://www.jnto.go.jp/
Forter, "JCA EMV 3DS Mandate" — https://www.forter.com/blog/new-guidance-from-japan-credit-associate-jca-on-emv-3ds-mandate/
JCB / IWI / PKSHA Technology, FARIS Joint Scoring Service Announcement — https://www.iwi.co.jp/news/2022/11/iwipkshafaris-powered-by-pksha-security.html
Ravelin, "Machine Learning for Fraud Detection" — https://www.ravelin.com/insights/machine-learning-for-fraud-detection
Nippon.com, "Fraud and Crime Data in Japan 2024" — https://www.nippon.com/en/japan-data/h02424/

Visa VAMP 2026: Your Compliance Math Changed on April 1

2026-04-02T13:00:00.000Z

TL;DR
Visa’s Visa Acquirer Monitoring Program (VAMP) merchant threshold dropped from 2.20% to 1.50% on April 1st, 2026. If you were sitting just under the old line, you may already be in violation without changing a thing.
The new formula merges fraud reports and disputes into a single ratio and one transaction can count against you twice. It’s measured by count, not dollars, so high-volume merchants carry more exposure.
Over-blocking fraud doesn’t help. Declining legitimate transactions shrinks the denominator of the ratio without reducing the numerator, which can push you further out of compliance.
The path forward is dual optimization: approve more real transactions to grow the denominator while cutting fraud and disputes to shrink the numerator. That’s a different discipline than chargeback management alone.

On April 1st, 2026, the Visa Acquirer Monitoring Program (VAMP) merchant threshold for fraud-and-dispute ratio decreased from 2.20% to 1.50%. That’s a 32% reduction, effective overnight. If your combined fraud-and-dispute ratio was sitting at 1.8% in March, you were compliant. So on April 2, with the exact same transaction volume and dispute count, you will be in violation and subject to penalties.

This isn’t an incremental policy tweak. The VAMP restructured how network-level risk enforcement works for card-not-present merchants. It unifies fraud reports and disputes into a single ratio, introduces a mechanism where one transaction can count twice against you, and creates cascading pressure from Visa to acquirers to merchants. For risk directors at high-volume eCommerce platforms, April 1st turned compliance from a backward-looking chargeback management task into a forward-looking optimization problem.

Here’s what the new math looks like, why it’s harder than the old math, and what you can do about it.

The Threshold Shift: 420 Fewer Allowable Incidents at the Same Volume

The numbers tell the story clearly. Under the prior threshold, a merchant processing 60,000 transactions per month could absorb up to 1,320 combined fraud-and-dispute events before entering Visa's "Excessive" monitoring tier. Under the April 1 threshold of 1.50%, that ceiling drops to 900 events. That's 420 fewer allowable incidents for the exact same transaction volume (Source: Coinflow, 2026).

A merchant operating at 1,050 incidents per month (a 1.75% ratio, comfortably compliant under the old rules) now exceeds the threshold without any operational change (Source: Coinflow, 2026). The new limits apply to merchants in the US, Canada, EU, and APAC. The CEMEA region remains at 2.20% (Source: Basis Theory, 2026; Forter, 2025). Latin America and the Caribbean was already at 1.50% before April 2026 (Source: Basis Theory, 2026).

The enforcement timeline matters too. Visa began enforcing Above Standard fines against acquirers in January 2026, adding pressure to thresholds that have been in place since VAMP launched in June 2025. Merchant thresholds tightened further on April 1. First-time violators get a three-month grace period before fines begin (Source: Chargebacks911, 2025; Forter, 2025; Ravelin, 2025). (This applies to merchants not enrolled in VAMP monitoring within the prior 12 months.) That grace period is the window you're in right now if your ratio crossed the new line.

What VAMP Actually Measures, and Why One Transaction Can Count Twice

The VAMP ratio formula is: (TC40 Fraud Reports + TC15 Disputes) / TC05 Settled Transactions (Source: Basis Theory, 2026; Chargebacks911, 2025; Forter, 2025). Three things about this formula deserve close attention.

It unifies fraud and disputes into one number. Prior programs tracked fraud monitoring and dispute monitoring separately. VAMP collapses them. A fraud alert (TC40) and a chargeback (TC15) from the same transaction both count against your ratio. One disputed transaction can generate two events, hitting your ratio twice. Forter's analysis notes this creates the potential for fines on a single disputed transaction from two separate penalty events (Source: Forter, 2025).

It's count-based, not dollar-based. Visa calculates VAMP ratios by transaction count, not transaction value. A marketplace with $200M in gross merchandise value across millions of low-price-point orders faces far more ratio exposure per dollar of revenue than a merchant with the same GMV concentrated in fewer high-value orders (Source: Basis Theory, 2026; Ravelin, 2025; MRC, 2025). The minimum monitoring threshold of 1,500 combined fraud-and-dispute events per month means high-volume operators are almost always in scope (Source: Basis Theory, 2026; Ravelin, 2025; Equifax, 2025).

It only counts card-not-present transactions. VAMP targets eCommerce specifically. Only CNP transactions enter the TC05 denominator (Source: Chargebacks911, 2025; MRC, 2025).

The double-counting mechanism is especially important for friendly fraud. Visa data indicates that approximately 75% of all disputes originate as friendly fraud (Source: Chargebacks911, 2025, citing Visa data). When a buyer receives an order, uses it, and files a fraud claim for an unauthorized transaction, both the TC40 fraud report and the TC15 dispute flow into the VAMP ratio. If you can't invoke Compelling Evidence 3.0 to exclude those counts (more on that below), both stand.

The Cascade You May Not See Coming: Acquirer-Level Pressure

Your acquirer has its own VAMP thresholds, and they're tighter than yours. Visa set acquirer "Above Standard" at 0.50% and "Excessive" at 0.70% for portfolio-level VAMP ratios (Source: Equifax, 2025; Basis Theory, 2026). Acquirers exceeding those limits face fines per affected transaction across their entire portfolio (Source: Chargebacks911, 2025).

This creates strong financial incentive for acquirers to proactively restrict or offboard merchants whose individual ratios threaten portfolio compliance (Source: Equifax, 2025; Forter, 2025; Chargebacks911, 2025). You may not hear about this pressure directly. The first signal could be tighter processing restrictions, account reserves, or a settlement pause.

The terminal consequence for persistent non-compliance is MATCH list placement, which effectively ends your ability to process Visa payments. Settlement pauses often occur before merchant communication, creating fund freezes lasting three to 14+ days before any formal notice arrives (Source: Coinflow, 2026).

If your acquirer's own ratio is under pressure, your individual compliance buffer shrinks further. The margin for error narrows from both directions.

Compliance Exclusions Exist, but They Require Infrastructure You May Not Have

VAMP does offer ratio relief through specific dispute resolution channels. Transactions resolved through Compelling Evidence 3.0 (CE3.0), Verifi CDRN, or Visa Rapid Dispute Resolution (RDR) can be excluded from the VAMP ratio calculation (Source: Forter, 2025; Chargebacks911, 2025; Basis Theory, 2026).

CE3.0 exclusions are the most valuable, and the hardest to qualify for. They require three things:

Device ID and IP address captured on all prior transactions. If you weren't collecting device fingerprints before the dispute, you can't retroactively produce matching data.
120 days of transaction history. You need at least four months of matching device and IP records for the disputed cardholder (Source: Forter, 2025).
Resolution within the same calendar month as the dispute. Both the dispute and its CE3.0 resolution must land in the same month to qualify for exclusion.

This creates an asymmetric advantage. Merchants who already have device fingerprinting infrastructure will systematically earn ratio exclusions that unprepared merchants simply cannot access (Source: Forter, 2025; Chargebacks911, 2025). The data capture has to be in place before the dispute happens. There's no way to build this retroactively once you're already in the monitoring window.

VAMP also introduced a separate Enumeration Ratio, tracking confirmed card-testing attempts as a share of all authorization attempts, including declined transactions. Merchants exceeding a 20% enumeration threshold face enrollment in a separate monitoring track (Source: Ravelin, 2025; Chargebacks911, 2025; Forter, 2025). Monitoring entry also requires a minimum of 300,000 enumerated authorization transactions per month (Source: Ravelin, 2025). For marketplace platforms where seller-side fraud or API-exposed checkout flows may invite card testing, this is a second compliance surface to monitor.

The False Decline Trap Hiding Inside VAMP Compliance

Here's where VAMP creates a genuinely new problem. Under prior monitoring programs, over-aggressive fraud blocking could drive down chargebacks without direct regulatory consequence. You'd lose revenue from false declines, but you'd stay compliant. Under VAMP, that trade-off breaks.

The false decline problem under VAMP is a denominator problem. Every legitimate transaction you decline removes one settled transaction from TC05 without removing a single fraud or dispute event from your numerator. Meanwhile, the fraudulent transactions you do approve still generate TC40 and TC15 events. Over-aggressive blocking shrinks the denominator without proportionally reducing the numerator, which makes the ratio worse, not better. The only way to improve both sides simultaneously is to approve more legitimate transactions (growing TC05) while preventing fraud and disputes at their source (shrinking TC40 + TC15).

The false decline cost across the industry is already substantial: approximately $50 billion annually in lost revenue. Global false decline losses are estimated at $443 billion annually, roughly nine times the $48 billion lost to actual fraud (Source: Riskified, 2025).

VAMP turns this from a revenue problem into a compliance problem. The optimization it imposes is genuinely dual-sided: minimize (TC40 + TC15) / Settled Transactions while simultaneously maximizing approved volume (Source: Corgi Labs internal memo, March 2026). Tightening your fraud rules without precision doesn't help the ratio. It can make it worse.

What Risk Directors Should Do Now

VAMP compliance is no longer about managing chargebacks in isolation. It's about optimizing a ratio where both the numerator and denominator matter, and where over-correction in either direction carries consequences.

Five concrete steps to take this month:

Calculate your current VAMP ratio. Pull your TC40, TC15, and TC05 counts from the last 90 days. Know exactly where you stand against the 1.50% threshold.
Audit your CE3.0 readiness. Confirm that your system captures device ID and IP address on every transaction. Verify you have 120 days of history. If you don't, start collecting now.
Quantify your friendly fraud exposure. With approximately 75% of disputes originating as friendly fraud (Source: Chargebacks911, 2025), this is likely the largest single contributor to your VAMP ratio.
Measure your false decline rate alongside your fraud rate. If your fraud rules are blocking legitimate transactions, you're shrinking your denominator without reducing your numerator.
Talk to your acquirer. Understand where their portfolio ratio stands and whether they're tightening restrictions on merchants in your category.

The underlying challenge is a dual optimization: approve more real buyers to grow your denominator while reducing fraud and disputes to shrink your numerator. These two goals used to live in separate operational silos. VAMP forces them into a single formula.

Corgi Labs builds tools for exactly this kind of problem. Corgi Intelligence surfaces your fraud, dispute, and decline data in one view so you can see your actual VAMP ratio exposure across processors. Corgi Model uses custom machine learning trained on your transaction data to dig into both sides of the equation, approving more legitimate orders while reducing chargebacks. For one eCommerce merchant, that meant +22% payments accepted, an 18% reduction in realized fraud rate, and more than $2 million in recovered revenue (Source: Corgi Labs product documentation, 2026).

If you want to see where your VAMP ratio stands and where the optimization opportunities are, book a demo.

Source Index

How a 15% Decline Rate Compounds Into an ~18% Revenue Loss

2026-03-01T17:07:00.000Z

Roughly 15% of all ecommerce orders are declined during authorization. About 70% of those declined orders belong to legitimate customers who were qualified to buy. And between 27% and 33% of those falsely declined customers never come back. Each of those numbers is a problem on its own. Together, they compound into something much larger: approximately 18% of your addressable revenue, lost before it ever reaches your books.

This is not a single-source headline statistic. It’s derived math, with each component independently documented. This article breaks down that math, traces it to the three friction mechanisms responsible for most avoidable declines, and shows you what recovery looks like in practice.

The Three-Part Equation: Where the ~18% Comes From

Start with the broadest layer. Across all markets, roughly 15% of ecommerce transactions fail to process successfully. That 15% includes both recoverable declines (false positives, routing failures, 3DS abandonment) and non-recoverable ones (insufficient funds, closed accounts, hard fraud). The 70% filter in the next layer isolates the recoverable portion. E-commerce baseline decline rates run 10–13%, with subscription and recurring billing models hitting 18–20% due to expired cards, changed billing details, and automated bank blocks (Wallid; Recurly, “Top Payment Decline Reasons for eCommerce”). The 15% figure represents a cross-market average.

Now apply the second layer. For the average merchant, issuers decline one in every 10 ecommerce dollars during payment authorization, and 70% of these declined orders are from good customers qualified to make the purchase.

The math at this point: 15% of orders declined, 70% of those from real customers. That gives you roughly 10.5% of all attempted orders from genuine buyers incorrectly rejected. This is immediate, same-day revenue loss.

Now add the third layer: the customers who don’t come back. Among customers who experience a false decline, 27%-41% never return to the merchant.

When you combine the immediate transaction loss (10.5% of revenue from real customers) with the permanent lifetime value destruction from non-return (27–33% of those customers gone forever), the total revenue impact across the customer lifecycle approaches 18% of addressable revenue.

What “Authorization Friction” Actually Means: Three Root Causes

Authorization friction is not a synonym for “low auth rate.” It refers to specific mechanisms within the authorization flow that cause legitimate transactions to fail. Three friction sources drive the bulk of avoidable declines.

Issuer over-restriction: the “do not honor” black box. Issuers see the basics: card details, balance, amount, location, and results from their standard fraud checks. They don’t see behavioral patterns, order history, or device details. This information gap leads to systematic over-flagging. Banks falsely decline approximately 15% of legitimate orders. The “do not honor” response code represents 10–60% of all refusals depending on geography. American Express codes over 90% of its declines as “do not honor,” while Visa in the US codes approximately 10% this way (Churnkey, “Do Not Honor Decline”). You can’t fix what you can’t diagnose.

3DS challenge abandonment. Ravelin found that 22% of payments sent through 3DS were lost (Ravelin, “One Fifth of Payments Sent to 3D Secure Are Lost”). That data is from 2019, before widespread 3DS 2.x adoption, and the landscape has improved since: 64% of 3DS transactions now go through a frictionless flow globally, and the UK achieves 93% 3DS success rates (Ravelin, 2025 Global Payments Report). But the friction remains real and ongoing: in Ravelin’s 2019 data, 91% of 3DS transactions took over five seconds to authenticate, with an average of 37 seconds (Ravelin, 2019). European merchants still see a 2–3.5% conversion drop from poorly applied 3DS. In the US, where 3DS success rates average only 41%, merchants may lose up to 15% (DECTA, “Why Your 3DS Authentication Has Low Approval Rates”).

Routing inefficiency. Merchants relying on a single acquirer without fallback logic leave revenue on the table when network rules or processor conditions change (Solidgate, “Intelligent Payment Routing”). A transaction that fails through one processor might succeed through another. Without cascading fallback routing, that revenue is simply abandoned. Intelligent routing optimization can improve approval rates by 10–15% (Solidgate; FlyCode).

The Customer Who Doesn’t Come Back: Why the Loss Keeps Growing

The immediate transaction loss is only the first impact. The compounding effect is what turns a $150 declined order into a multi-thousand-dollar problem.

Among loyal customers (those with three or more prior approved orders), a false decline triggers a 65% reduction in the number of future orders and a 16% drop in average order value. Consider what that means: a customer who spent $1,200 with you last year gets falsely declined on a $150 order. If they’re in the 27–33% who never return, you’ve lost $1,350 in year-one value alone, not $150. If they’re in the group that comes back but spends less, you’ve still lost hundreds in lifetime revenue from that single decline event.

The behavioral data reinforces this. Only 25% of declined customers try another payment card. Thirty-nine percent abandon the cart entirely (PYMNTS, November 2023). And up to 32% of falsely declined customers post negative feedback on social media (ClearSale / Sapio Research, 2020). That’s brand damage on top of revenue loss.

Fiserv’s research adds another dimension. Twenty percent of cardholders stop using their card entirely after experiencing two or more false declines within a six-month period, and average monthly spending drops 15% per card after two or more false positive denials (Fiserv, “Financial Institutions Ease Cardholder Frustration by Addressing Transaction False Declines”). The damage extends beyond a single merchant.

The financial scale is significant. US ecommerce merchants permanently lost $81 billion to false declines in 2023, with $157 billion in sales initially at risk (PYMNTS, November 2023). Riskified’s 2025 Ascend research, based on a practitioner survey of 130+ payment professionals, puts total ecommerce losses from false declines, fraud, and policy abuse at $448 billion annually. J.P. Morgan data shows that false positive losses (19% of total fraud cost) actually exceed actual fraud losses (7% of total fraud cost) (J.P. Morgan, “False Positives & Fraud Prevention Tools”). False declines cost merchants 13 times more than actual fraud (Fiserv Carat). One Aite Group / ClearSale estimate from 2019 puts the ratio at 75 times.

Why Your Dashboard Doesn’t Show Any of This

You might expect your payments dashboard to surface this problem. It doesn’t.

Eighty-two percent of executives cannot pinpoint why their payments fail due to fragmented data (PYMNTS, August 2024). Only one-third of ecommerce merchants know whether fraud caused a failed payment (PYMNTS, November 2023). Sixty percent say failed payments are expensive to track and resolve (PYMNTS, 2024).

Standard PSP dashboards report total authorization rate. They don’t report false decline rate. They don’t track whether a declined customer returns or churns permanently. They don’t measure lifetime value erosion from individual decline events. For multi-PSP merchants, the problem multiplies: three different dashboards with three different authorization rate figures and no unified view of which declines are recoverable false declines versus genuine fraud.

This is why the ~18% revenue impact stays invisible. The data exists, but it sits in separate systems that don’t connect.

What High-Performing Merchants Do About Authorization Friction

Recovery is structured around the three root causes.

Enriching transaction data for issuers. Sending additional context (device fingerprint, customer tenure, transaction history) gives issuers the confidence to approve transactions they would otherwise flag.

Smarter 3DS application. Applying 3DS selectively (exempting low-risk transactions, using risk-based authentication) reduces abandonment while maintaining compliance. The difference between blanket 3DS enforcement and intelligent exemption strategies can be the difference between a 2–3.5% conversion drop and minimal impact.

Network tokenization. Replacing stored card numbers with network-level tokens delivers a 4.6% global authorization rate lift and 26–30% fraud reduction (Visa Acceptance Solutions, “Tokens Are Key to Future Proofing Payments”).

The results at scale are real. Checkout.com’s Intelligent Acceptance raised merchants’ acceptance rates by an average of 3.8% in 2024, generating over $10 billion in additional merchant revenue since launch (Checkout.com newsroom). Stripe’s Adaptive Acceptance recovered $6 billion in falsely declined transactions in 2024, a 60% year-over-year improvement in retry success rate (Stripe, “AI Enhancements to Adaptive Acceptance”).

The Recovery Calculation: What This Is Worth for Your Business

Here’s the math applied to a specific scenario.

If your business processes $10 million per month at an 87% authorization rate, $1.3 million per month is declining. If 70% of those declines are false declines from real customers, that’s $910,000 per month in legitimate buyers being turned away. A 3–5 percentage point improvement in authorization rate recovers $300,000 to $500,000 per month, or $3.6 million to $6 million per year. That recovery doesn’t require new customers, additional marketing spend, or pricing changes. It comes from approving real buyers who are already at checkout.

The first step is visibility: understanding which of your declines are recoverable, which friction mechanisms are driving them, and which customers you’re losing permanently. That’s the analytical layer most merchants are missing.

Corgi Intelligence surfaces exactly this data, unifying decline analytics across processors and quantifying the revenue impact of each friction source. Corgi Model takes it a step further with custom machine learning trained on your transaction data, approving more real buyers while reducing chargebacks.

Sources

Aite Group / ClearSale, “False Decline Cost Ratios” (2019), via Greip.io

Checkout.com, “Checkout.com Surpasses $10 Billion in Revenue Unlocked”

Churnkey, “Do Not Honor Decline: Meaning, Stats, and How To Fix”

ClearSale / Sapio Research via Digital Commerce 360, “33% of US Consumers Drop Retailers After a False Decline” (2020)

DECTA, “Why Your 3DS Authentication Has Low Approval Rates: 5 Optimisation Tips”

Fiserv, “Financial Institutions Ease Cardholder Frustration by Addressing Transaction False Declines”

Fiserv Carat, “False Decline”

FlyCode, “Smart Payment Orchestration: From Simple Rules to AI”

GR4VY, “Approval Rates in Payments: Meaning and Deep Dive for 2025”

J.P. Morgan, “False Positives & Fraud Prevention Tools”

LexisNexis, 2018 True Cost of Fraud Survey, via Sherwen

Primer.io, “How to Recover Lost Revenue with Cascading Payments”

PYMNTS, “82% of Merchants Don’t Have the Data to Pinpoint Why Payments Fail” (August 2024)

PYMNTS, “eCommerce Firms Will Lose $81B to False Declines in 2023” (November 2023)

PYMNTS, “Nearly 60% of Firms Say Failed Payments Are Expensive to Track and Resolve” (2024)

Ravelin, “One Fifth of Payments Sent to 3D Secure Are Lost” (2019)

Ravelin, “New Data in Payments Authentication & 3DS Released” (Global Payments Report 2025)

Recurly, “Top Payment Decline Reasons for Subscription eCommerce”

Riskified, “Unlock Revenue by Optimizing Payment Authorization Rates”

Riskified, Lorna Jane Case Study

Riskified / StockTitan, “85% of Merchants Battle to Balance Customer Experience and Fraud Prevention” (Ascend 2025)

Sherwen, “How False Declines Hurt More Than Actual Ecommerce Fraud” (LexisNexis data)

Signifyd, “5 Strategies to Increase Bank Authorization Rates for Merchants”

Signifyd, “False Declines Explained”

Solidgate, “Intelligent Payment Routing: Boost Conversion”

Stripe, “AI Enhancements to Adaptive Acceptance”

Visa Acceptance Solutions, “Tokens Are Key to Future Proofing Payments”

Wallid, “Where Payments Fail: Industries with Highest Decline Rates 2025”

Payment Optimization ROI: How to Build the Business Case Your CFO Will Approve

2026-02-16T02:04:20.420Z

Most payment optimization proposals stall in budget conversations because they speak in percentages instead of dollars. You have the benchmarks. You know your acceptance rate could be higher and your chargebacks could be lower. But the business case your CFO approves is built on payback period, net revenue impact, and margin preservation. Not on authorization rate dashboards.

This article gives you the ROI formula, the industry benchmarks to fill it in, a fully worked example you can adapt to your own numbers, and the structure for a business case presentation that translates payment metrics into P&L language.

Why Payment Optimization Proposals Stall (and What CFOs Actually Need)

Payment teams tend to pitch optimization in operational language: authorization rates, decline codes, chargeback ratios. These metrics matter. But they’re not the language of budget approval.

CFOs are shifting from asking “How much should we spend to be compliant?” to “What is the expected loss exposure across our payment flows, and how does that compare to the cost of advanced controls?” (PYMNTS). That shift is real. Nearly 70% of financial institutions increased fraud-detection spending year over year, and cost is becoming less of a barrier as firms view fraud technology as core infrastructure (PYMNTS).

But strategic investment still requires a quantified framework. Your CFO needs three things before approving a payment optimization budget:

Net revenue impact in dollars. Not percentages. Dollars.
Payback period. Under 12 months is a straightforward yes by most CFO frameworks; anything under 24 months is within standard approval range (Nucleus Research).
Cost of inaction. What you lose by maintaining the status quo, expressed in annual margin erosion and competitive risk.

The ROI formula below gives you all three.

The ROI Formula: Four Variables with Industry Benchmarks

Here’s the formula. Each variable includes benchmark ranges from industry research so you can plug in your own numbers or start with the midpoint. If you want to skip ahead and run the numbers now, try our Payment Revenue Calculator with your own data.

Net Annual ROI = Acceptance Rate Lift Revenue + Chargeback Cost Reduction + False Decline Revenue Recovery - Total Project Cost

Break it down:

Variable 1: Acceptance Rate Lift Revenue

Formula: Acceptance Rate Lift (pp) x Monthly Transaction Volume x AOV x 12

What it measures: The additional revenue you collect by approving transactions that would otherwise decline.

How to find your baseline: Pull your current authorization rate by processor, card brand, and region. E-commerce card-not-present rates typically range from 85% to 92% (Worldpay; Nuvei). If you run multiple PSPs, average them or (better) use the volume-weighted rate.

What to use if you don’t have exact numbers: Start with a conservative 3-percentage-point lift, which sits at the low end of documented outcomes.

Variable 2: Chargeback Cost Reduction

Formula: Current Monthly Chargebacks x Reduction Rate x Average Cost per Chargeback x 12

What it measures: The savings from reducing chargebacks through better fraud decisioning and prevention tooling.

Input	Benchmark Range	Source
Chargeback reduction rate	70-95%	Chargeflow; industry case studies
Average cost per chargeback	$15-$100 in fees alone; true cost is $4.61 per $1 of fraud when including merchandise loss, labor, and penalties	Chargeflow; Chargebacks911; Mastercard
Current chargeback ratio	Industry average: 0.56-0.60%; digital goods/subscriptions: 1.85%	Chargeflow; ChargebackStop

Why the cost per chargeback matters more than the fee: The $15-$100 fee is just the processor charge. In 2025, every dollar lost to fraud costs U.S. merchants $4.61 total, a 37% increase from 2020 (Chargeflow). That multiplier accounts for the transaction amount, merchandise loss, shipping, administrative labor, and potential card network penalties. A $100 chargeback really costs you $461.

The penalty cliff: Merchants exceeding a 0.9% chargeback ratio enter Visa’s Dispute Monitoring Plan, with penalties of $50 per chargeback plus $25,000 review fees in months five through 12 (Visa VAMP). This isn’t a theoretical risk. Digital goods and subscription merchants average a 1.85% chargeback rate (Chargeflow), well above the threshold.

Variable 3: False Decline Revenue Recovery

Formula: Recovered Transactions per Month x Percentage Who Would Have Been Lost x Average Customer Lifetime Value x 12

What it measures: The lifetime value preserved by approving real buyers your current system would have blocked.

Input	Benchmark Range	Source
False decline cost (global, annual)	Estimates range from $50 billion (PYMNTS, conservative) to $443 billion (ClearSale; Riskified), depending on methodology and scope	ClearSale; Riskified; PYMNTS
Customers who never return after a false decline	40%	ClearSale
Loyal customer order volume reduction after a false decline	65%	Riskified
Average customer lifetime value	Your data (or estimate based on AOV x purchase frequency x retention period)	Internal reporting

This variable is the hardest to quantify precisely, which is exactly why most business cases undercount it. The acceptance rate lift (Variable 1) captures the immediate transaction revenue. Variable 3 captures the downstream impact: when you decline a real buyer, 40% of them never come back (ClearSale). Among loyal customers who do return, their order volume drops by 65% (Riskified).

Conservative approach for your CFO: If you can’t model CLV precisely, present Variable 3 as upside rather than a core number. Note that merchants lose up to 75 times more revenue to false declines than to actual fraud (Aite Group, via Riskified). That ratio reflects the compounding effect of lost lifetime value, not just the declined transaction. Frame the CLV recovery as additional return beyond the hard-dollar figures in Variables 1 and 2.

Variable 4: Total Project Cost

Formula: Implementation Cost + Annual Ongoing Cost

Input	Benchmark Range	Source
Implementation cost	~$50,000 for mid-market implementations (illustrative; actual costs vary by vendor and scope)	Industry estimates
Ongoing cost	Varies: monthly SaaS fee, percentage of recovered revenue, or per-transaction pricing	Vendor-specific
First-year ROI on sub-$50K implementations	10-26x	Signifyd; industry benchmarks

For most mid-market merchants, the implementation cost is small relative to the revenue at stake. That’s why first-year ROI on sub-$50K implementations consistently falls in the 10-26x range (Signifyd; industry benchmarks).

Worked Example: A $120M Mid-Market Merchant

Let’s run the numbers for a merchant that looks like a typical multi-PSP e-commerce operation.

Current State

Metric	Value
Annual transaction volume	$120,000,000
Monthly transactions	150,000
Average order value	$67
Current authorization rate	87%
Monthly approved transactions	130,500
Monthly declined transactions	19,500
Current chargeback ratio	0.70%
Monthly chargebacks	~910
Average all-in chargeback cost	$200 (fee + labor + merchandise loss)

An 87% authorization rate sits in the lower half of the 85-92% e-commerce range (Worldpay; Nuvei). That’s not unusual for a multi-PSP merchant without optimization tooling.

Optimized State (Conservative Scenario: 3pp Lift, 70% Chargeback Reduction)

Implementation and ongoing costs below are illustrative for a mid-market deployment. Your actual costs will vary by vendor, scope, and pricing model.

Variable 1: Acceptance Rate Lift Revenue

A 3-percentage-point lift (87% to 90%) recovers 4,500 additional transactions per month.

4,500 recovered transactions x $67 AOV x 12 months = $3,618,000 per year

Variable 2: Chargeback Cost Reduction

A 70% reduction drops monthly chargebacks from 910 to 273, saving 637 chargebacks per month.

637 saved chargebacks x $200 all-in cost x 12 months = $1,528,800 per year

Variable 3: False Decline Revenue Recovery (Upside)

Of the 4,500 newly approved transactions per month, assume 40% of those buyers would have never returned. At a conservative CLV of $200 (roughly 3x AOV), that’s:

4,500 x 40% x $200 = $360,000 per month in preserved lifetime value

This figure is directional, not exact. Present it as upside in your business case, not as a guaranteed return.

Variable 4: Total Project Cost

$50,000 implementation + $30,000 annual ongoing = $80,000 first year

The Math

Component	Annual Impact
Acceptance rate lift revenue	+$3,618,000
Chargeback cost reduction	+$1,528,800
Hard-dollar gross benefit	+$5,146,800
Total project cost (Year 1)	-$80,000
Net first-year impact	+$5,066,800
Payback period	~6 days
First-year ROI	63x

Even if you cut the acceptance rate lift in half (1.5pp instead of 3pp) and the chargeback reduction to 50%, the net first-year impact is still over $2.3 million on an $80,000 investment.

Realistic Scenario (5pp Lift, 70% Chargeback Reduction)

For context, a 5-percentage-point lift sits in the middle of the 3-12% benchmark range. Klarna achieved 6 percentage points (Optimized Payments). Reach achieved 9.5 percentage points with Checkout.com. Nord Security achieved 10% conversion improvement through AI-driven routing (Juspay; Adyen).

Component	Annual Impact
Acceptance rate lift revenue (7,500 x $67 x 12)	+$6,030,000
Chargeback cost reduction	+$1,528,800
Hard-dollar gross benefit	+$7,558,800
Total project cost (Year 1)	-$80,000
Net first-year impact	+$7,478,800

The point isn’t to pick one scenario. The point is that even the most conservative projection delivers returns that exceed any reasonable payback threshold.

Want to see what these numbers look like for your business? Run your own scenario in our Payment Revenue Calculator. Plug in your transaction volume, authorization rate, and chargeback ratio to get a personalized revenue recovery estimate in seconds.

The Cost of Doing Nothing: Margin Erosion You Can Quantify

Your CFO will ask about risk in both directions: “What if the optimization doesn’t deliver?” and “What if we don’t invest?” The second question has a concrete answer.

False declines compound. Industry estimates of the global cost of false declines range from $50 billion to $443 billion per year, depending on methodology (ClearSale; Riskified; PYMNTS). Even at the conservative end, that dwarfs the $48 billion lost to actual fraud. Every false decline isn’t just a lost transaction. It’s a customer relationship at risk: 40% never return, and the ones who do reduce their order volume by 65% (ClearSale; Riskified).

Chargeback penalties escalate. Total chargeback losses will reach $117 billion globally in 2026 (Chargebacks911). Merchants who exceed Visa’s 0.9% chargeback ratio face escalating penalties that can reach $50 per chargeback plus $25,000 in review fees (Visa VAMP). In severe cases, exceeding thresholds can cost you your merchant account entirely, which means you can’t process payments at all.

The fraud cost multiplier grows. Every dollar lost to fraud now costs $4.61 in total impact (Chargeflow). That’s up 37% from 2020, and the trajectory is not reversing. Merchant losses from online payment fraud will exceed $91 billion in 2028 alone (Juniper Research).

Competitors are investing. Nearly 70% of financial institutions increased fraud-detection spending year over year (PYMNTS). Competitors who optimize their payment flows capture the customers you decline. Once those customers form new purchasing habits, they don’t come back.

Frame the cost of inaction as an annual number in your business case. For the $120M merchant in our worked example, maintaining the current 87% authorization rate and 0.70% chargeback ratio costs roughly $5.1 million per year in recoverable revenue and avoidable chargeback costs. That’s not a projection. It’s the gap between where you are and where documented benchmarks say you could be.

From Spreadsheet to Approval: Structuring the Deck

You have the formula. You have the math. Here’s how to package it for the budget conversation.

The Six Slides Your CFO Needs

Slide 1: Executive summary. Lead with the payback period and net first-year revenue impact. One sentence on what you’re proposing. One sentence on total cost. One sentence on what happens if you don’t invest.

Slide 2: Current-state audit. Your authorization rate by processor, card brand, and region. Your chargeback ratio and total chargeback costs (not just fees, but the full $4.61 multiplier). Your estimated false decline volume, based on the gap between your current auth rate and the 92-95% benchmark (Micros Integrated Payments).

Slide 3: The ROI formula with three scenarios. Run conservative, realistic, and optimistic projections using the formula above. Anchor on the conservative scenario. Let the realistic and optimistic numbers show the upside range. This approach demonstrates rigor, not optimism.

Slide 4: Proof points and case studies. Cite real examples: Klarna’s 6% acceptance rate increase across key markets (Optimized Payments). Reach’s 9.5% authorization rate increase through intelligent acceptance (Checkout.com; Juspay). Checkout.com’s $741 million in revenue recovered across its merchant base. A mid-sized insurance company saving $700,000 annually through payment process optimization (Optimized Payments). Worldpay recovering $200 million in revenue for merchants in 2024 (Worldpay).

Slide 5: Implementation timeline and resource requirements. Be specific about what the project needs: timeline, internal resources, integration scope. Sub-$50K implementations with 10-26x first-year ROI speak for themselves (Signifyd; industry benchmarks). Target a 12-month payback as the benchmark for a straightforward approval; anything under 24 months is within standard CFO approval range (Nucleus Research).

Slide 6: Cost of inaction. Quantify the annual cost of maintaining the status quo using the formula in reverse. Show the chargeback penalty risk if your ratio crosses 0.9%. Note the competitive context: competitors who optimize will capture the buyers you decline.

Three Tips for the Conversation

Speak in dollars, not percentages. “$3.6 million in recovered revenue” lands differently than “a 3-percentage-point auth rate lift.” Both describe the same outcome. Only one gets a budget approved.

Present the conservative scenario as your ask. If the conservative projection justifies the investment (and at these benchmarks, it almost always does), the realistic and optimistic scenarios become upside. Your CFO will appreciate the intellectual honesty.

Address the fraud question directly. “If we approve more transactions, do we increase fraud exposure?” The answer: properly implemented optimization actually reduces fraud. Network tokenization delivers a 2.1-6% authorization rate lift while simultaneously reducing fraud by 30% (Visa; Mastercard). The goal is approving more real buyers, not lowering the bar.

Start with Visibility, Then Build the Case

The ROI formula works when you have clean data to feed into it. That means knowing your authorization rate by processor, your true chargeback cost (not just the fee), and your false decline volume. Most merchants running multiple PSPs don’t have unified visibility across these metrics.

If your payments data sits in three different dashboards with three different reporting formats, the first step isn’t buying optimization tooling. It’s getting a unified view of where your revenue is actually leaking. Once you can see the full picture across processors, building the business case is arithmetic.

Corgi Intelligence surfaces these metrics across your payments stack, and Corgi Model applies merchant-specific machine learning to approve more real buyers while reducing chargebacks. Both work on your existing platform with no development work and deliver results in days.

Your CFO doesn’t want a pitch about payment technology. She wants a spreadsheet that shows payback period, net revenue impact, and the cost of standing still. Now you have the formula to build one.

Corgi Labs Payment Revenue Calculator

Sources

Your Fraud System Is Your Most Expensive Revenue Leak.

2026-02-10T18:19:00.000Z

Most ecommerce finance teams track fraud losses closely. Chargebacks get dashboards. Fraud rates get quarterly reviews. But false declines (legitimate orders your fraud system blocks by mistake) rarely get the same attention, even though they cost you far more.

In 2023, US merchants lost $81 billion to false declines. For comparison, total ecommerce fraud losses were roughly $48 billion globally. For many ecommerce sellers, fraud prevention tools are blocking more revenue than fraudsters are stealing.

That’s worth saying again: for many ecommerce businesses, the system designed to protect their revenue may be their biggest source of lost revenue.

The real cost of a $100 false decline

When your system blocks a $100 order from a real buyer, you don’t just lose $100. You lose in four directions at once.

The immediate sale. That $100 is gone. Research from the Merchant Risk Council shows merchants decline about 6% of orders for suspected fraud. Of those, roughly two-thirds are legitimate buyers who got caught in the filter.

The customer’s lifetime value. This is the big one. Industry surveys consistently show that 40% to 42% of customers never return after a false decline. They don’t call support. They don’t retry. They just leave. If that customer would have spent $100 a month over the next several years, you’ve lost thousands of dollars over a single blocked order.

One study of loyal customers (three or more previous purchases) found that a false decline cut their future order volume by 65%. Even the ones who came back spent 16% less per order. The relationship doesn’t recover.

Your acquisition cost. You spent real money to get that customer to checkout. The average ecommerce CAC is $70, which ranges from $50 to $130 depending on your vertical (industry benchmarks). When a new customer gets declined and walks away, 100% of that acquisition investment is wasted. You paid to acquire someone, then your own system turned them away.

Your brand. 32% of falsely declined customers post about it online. Among Gen Z, that number rises to 35%. One frustrated social media post about being “treated like a fraudster” doesn’t just lose you that customer. It discourages the next 10 who see it.

Add it up

For a $100 false decline, the math looks roughly like this:

Component	Cost
Lost sale	$100
Lost lifetime value (40% defection × $2,000 avg LTV*)	$800
Wasted acquisition cost (40% defection × $70 CAC)	$28
Support and brand damage	$50–$100
Total cost per $100 decline	$978–$1,028

*Based on repeat-purchase ecommerce benchmarks ($100 AOV × 20 orders). Luxury and subscription verticals typically exceed this figure.

The claim that you lose $750+ in lifetime value and $250+ in wasted CAC for every $100 blocked is directionally right. In some verticals (luxury, subscription, B2B), it’s conservative.

Other research puts it more starkly: for every $1 lost to actual fraud, merchants forfeit $30 by declining real buyers.

To put it in cumulative terms: if a business has a 6% decline rate and two-thirds of those are false declines, total revenue can be boosted 4% simply by improving payment decision logic. For a retailer with $100M in annual sales, that’s $4M in revenue being forfeited needlessly.

Why this stays invisible

False declines don’t show up as a line item. There’s no chargeback notification, no dispute filing, no alert from your payment processor. The customer simply doesn’t come back, and your analytics never register the lost customer at all — or misattribute the drop to routine churn.

Most merchants don’t track their false decline rate at all. The ones who do often discover the number is higher than expected. PYMNTS Intelligence found that 64% of failed payments are difficult to recover, and only 22% of customers will definitely retry after being declined.

The customers you’re losing aren’t complaining to you. They’re complaining to Twitter. Or they’re just buying from your competitor.

What to do about it

Measure your insult rate — or at least estimate it. Track retry success rates and customer complaints on blocked orders. If you have the appetite, run holdout tests on a sample of auto-declined transactions. The number will be imprecise, but even a rough estimate is better than the zero most merchants are working with.

Quantify the full cost. Don’t model a false decline as a $100 loss. Model it as a $1,000 loss. Use your own LTV and CAC numbers to build the multiplier for your business. When finance teams see the true cost, false decline reduction moves from a fraud team problem to a CEO’s or CRO’s revenue priority.

Rethink your fraud approach. Generic fraud rules block good customers because they don’t know your buyers. Custom machine learning trained on your transaction data can separate real buyers from real fraud with much higher accuracy. The goal isn’t less fraud prevention. It’s smarter fraud prevention that blocks fraud, not buyers.

Your design team has optimized every pixel for a good experience and high shopping cart conversion. So why aren’t you thinking about how every 20th legitimate customer is being told to go away — right when they are trying to pay you?

The $81 billion lost to false declines in 2023 is a cost. It’s also an opportunity. Every legitimate order you approve that your current system would have blocked flows straight to your top line.

Your payment data is sitting on gold. The question is whether you’re digging it up, or burying it deeper.

Sources

PYMNTS Intelligence & Nuvei, "Fraud Management, False Declines and Improved Profitability" (November 2023)

Merchant Risk Council, "2024 Global eCommerce Payments & Fraud Report"

Forter, "2023 Consumer Trust Premium Report"

ClearSale, "State of Consumer Attitudes on Ecommerce, Fraud, & CX 2023-2024"

Signifyd, "False Declines Explained: How to Prevent Fraud False Alarms"

Riskified, "How Consumers Respond to False Declines"

Rivo, "Average Customer Acquisition Cost for eCommerce"

Deliberate Directions, "Customer Acquisition Cost Ecommerce: 2026 Benchmarks"

Harvard Business Review / Bain & Company, "The Value of Keeping the Right Customers"

Your Authorization Rate Is a Vanity Metric: What That 91% Is Actually Hiding

2026-02-09T20:00:00.000Z

For payment operations leaders, the authorization rate has become the metric you check but never question. It sits on a dashboard, it trends in a reasonable range, and it gives you a false sense of control. The problem isn’t that the number is wrong. The problem is that it’s incomplete, and the gap between what it shows and what’s actually happening is where your revenue disappears.

According to Worldpay, for a business processing $1 billion in annual transactions, a single percentage point improvement in authorization rate equals $10 million in recovered revenue. That lift comes without acquiring new customers, increasing marketing spend, or changing pricing. Yet most merchants lack the visibility to diagnose where those lost percentage points go.

Here’s what your authorization rate is hiding, how to quantify what you’re losing, and what the highest-performing merchants do to close the gap.

The Composite Metric Problem: Why Top-Line Auth Rates Mislead

Industry authorization rates for e-commerce range from 85% to 95%, according to data from Worldpay and GR4VY. These benchmarks primarily reflect North American and European markets; authorization rates in LATAM and APAC regions often trend lower due to different issuer practices and fraud patterns. That said, even within this range, a “good” rate for one merchant may represent millions in lost revenue for another at similar scale. The top-line number tells you your approval percentage, but it doesn’t tell you why transactions fail, which failures you can fix, or how much money sits on the table.

Your authorization rate is a composite of multiple failure modes blended into one figure. It includes hard declines (stolen cards, closed accounts) that you can’t recover. It includes soft declines (insufficient funds, processor timeouts, missing authentication data) that you often can recover. It includes transactions flagged by issuer fraud models for patterns that aren’t actually fraudulent. And it includes a massive bucket of “do not honor” responses that carry no diagnostic value at all.

When you look at a single number on a dashboard, you’re averaging all of these together. A stable 91% might mean your soft decline recovery is excellent and your hard decline rate is creeping up. Or it might mean your fraud flags are increasing but your retry logic is compensating. You can’t tell, because the metric doesn’t decompose itself.

What Your Dashboard Isn’t Showing You: The Anatomy of a Decline

Here’s the first thing most dashboards obscure: 80% to 90% of all payment declines are soft declines, meaning they’re potentially recoverable. Soft declines can often be resolved through retry logic, updated card credentials, or richer transaction data sent to the issuer. Yet most dashboards don’t clearly separate soft declines from hard declines (Spreedly; GR4VY).

The second thing your dashboard hides is even more frustrating. According to a 2016 Visa Global Declines Analysis, over 76% of Visa’s global declined transaction volume fell into just two categories: “insufficient funds” and “do not honor.” Industry sources confirm this pattern persists today (Churnkey). The “do not honor” code (response code 05) is a catch-all that can represent anywhere from 10% to 60% of all refused payments depending on geography (Churnkey). As American Banker reported, issuing banks put most declines into one large bucket of “do not honor,” giving merchants almost no information about why a transaction actually failed.

Think about what that means in practice. Your dashboard shows a decline. The decline code says “do not honor.” You have no idea whether the issue was a fraud rule, a velocity check, an address mismatch, or something else entirely. You can’t fix what you can’t diagnose.

If you run payments across multiple processors, the problem compounds. Each PSP uses different data formats, reporting structures, decline code taxonomies, and settlement timelines. A merchant using Stripe, Adyen, and Braintree may see three different decline rates for the same region with no way to normalize or compare them (Payrails). Every PSP dashboard shows different metrics, with fields that don’t match and varying timeframes, creating fragmented reporting that hides revenue opportunities.

For multi-PSP merchants, your authorization rate isn’t just a vanity metric. It’s three or four different vanity metrics that don’t talk to each other.

The Hidden Revenue Leak: Quantifying What You’re Losing

The revenue buried beneath your authorization rate is larger than most payment teams realize.

False declines (legitimate transactions incorrectly rejected) cost merchants an estimated $308 billion globally in 2023, according to industry estimates compiled by Riskified and others. That figure exceeds actual fraud losses. A more conservative 2026 estimate from PYMNTS places the number at $50 billion. Either way, merchants lose more money to false declines than to fraud itself, a point Chargebacks911 has emphasized (CrowdFund Insider).

The customer impact is just as stark. In 2024, 56% of U.S. consumers reported experiencing a false payment decline in the prior three months (PYMNTS). Among loyal customers who experience a false decline, subsequent order volume drops by 65% (Riskified). Riskified’s research shows that 27% of loyal customers never return to the merchant after a false decline. Broader studies suggest 32% to 33% of all consumers abandon a merchant entirely after the experience (Riskified; Signifyd).

For subscription and SaaS businesses, the math gets worse. The average SaaS business loses approximately 9% of its recurring revenue to failed payments annually, effectively negating a full month of growth each year (Stripe analysis, via Baremetrics). Involuntary churn, caused by failed payments rather than customer decisions, accounts for 20% to 40% of total churn in subscription businesses (ProfitWell research, via Userpilot).

Consider what that means if you’re running a $40 million ARR SaaS company. Nine percent of recurring revenue is $3.6 million per year lost to payment failures. Only about 70% of failed payments are ever recovered on average (Recurly Research). The remaining 30% becomes permanent revenue loss that your dashboard attributes to “churn” without distinguishing whether the customer chose to leave or their payment simply failed.

Your authorization rate doesn’t tell you any of this. It shows you a percentage. It doesn’t show you the customers who left because their renewal was declined, the revenue that could have been retried, or the fraud rules that are blocking your own subscribers.

What High-Performing Merchants Do Differently

The gap between average and top-performing merchants isn’t luck. It’s instrumentation and process. Here are the specific techniques that move authorization rates by meaningful amounts, backed by data.

Network tokenization. Replacing stored card numbers (PANs) with network-level tokens delivers a measurable 2 to 6 percentage point lift in authorization rates. Visa reported a 4.6% global authorization rate lift for tokenized transactions versus PAN-based transactions, along with a 30% reduction in fraud (Visa Acceptance Solutions). Mastercard reports a 3 to 6 percentage point improvement (Mastercard). By 2024, 47% of merchants had adopted tokenization, up from 44% in 2023. By 2025, six in 10 merchants using tokenization cited authorization rate improvement as a primary benefit (MRC Global Reports).

Intelligent retry logic. Smart retry engines, combined with card account updater services, recover 60% to 70% of failed payments (Slickerhq; Cleverbridge). The timing, sequencing, and data enrichment of retries matters enormously. A retry sent at the wrong time or without updated card information fails just like the original attempt. A retry sent with fresh credentials, when the cardholder’s account is more likely to have funds, succeeds at significantly higher rates.

Stripe’s Adaptive Acceptance recovered $6 billion in falsely declined transactions in 2024, reflecting a 60% year-over-year increase in retry success rate (Stripe). Their average authorization rate lift across merchants is roughly 2.2%.

Richer transaction data for issuers. Issuers decline transactions when they lack confidence that the transaction is legitimate. Sending additional data fields (device fingerprint, customer tenure, transaction history) gives issuers more context to approve. One athletic apparel brand working with Riskified lifted authorization rates from 82% to 95% by enriching transaction data and optimizing the fraud decisioning layer before transactions reached issuers (Riskified).

Unified cross-processor analytics. Merchants who normalize their decline data across processors can spot patterns invisible in siloed dashboards. Is one processor declining more transactions in a specific BIN range? Are issuer fraud rules triggering differently depending on which acquirer routes the transaction? Without unified analytics, these questions go unanswered.

The proof points are compelling. Zapier achieved a 4% authorization rate uplift by combining Adaptive Acceptance, network tokens, and card account updater, translating to over $3 million in additional revenue (Stripe Newsroom). GAIA, a streaming company, moved from 80% to 89%+ authorization rates after gaining visibility into why transactions were failing and applying targeted optimization (Stripe). Worldpay reports that their optimization tools deliver a 1.5% revenue uplift within 90 days for participating merchants.

From Vanity Metric to Revenue Recovery: Five Steps to Start

Closing the authorization rate blind spot doesn’t require a full platform migration. It starts with visibility.

1. Disaggregate your authorization rate. Break your top-line number down by decline type (soft vs. hard), decline code, issuer, BIN range, card brand, and processor. This single step often reveals that 80% or more of your declines are soft and potentially recoverable.

2. Build a soft decline retry strategy. Not all retries are equal. Map your most common soft decline codes to specific retry actions: timing adjustments, credential updates, data enrichment, or alternative routing. Target a 60% to 70% recovery rate on soft declines as your benchmark.

3. Adopt network tokens. If you haven’t moved to network tokenization, you’re leaving a 2 to 6 percentage point authorization rate lift on the table. The merchant adoption curve is accelerating, and the data on authorization rate improvement is consistent across card networks.

4. Unify your payments data across processors. If you run multiple PSPs, you need a single view that normalizes decline codes, standardizes reporting periods, and lets you compare performance across processors for the same transaction types. Fragmented dashboards make optimization guesswork.

5. Measure what matters. Track false decline rate, soft decline recovery rate, involuntary churn rate, and revenue recovered per retry cycle. These metrics tell you whether your authorization rate is improving because you’re actually approving more real buyers, or just because your transaction mix shifted.

Each of these steps moves you from treating your authorization rate as a number to check toward treating it as a system to optimize. The merchants recovering millions in previously lost revenue aren’t doing anything mysterious. They’re looking at data that was always there, just buried beneath a single percentage on a dashboard.

For payment teams ready to dig into their decline data across processors and pinpoint exactly where revenue is leaking, Corgi Intelligence and Corgi Model provide this level of visibility and optimization, with results in days and no development work required.

The 91% on your dashboard isn’t wrong. It’s just not telling you the whole story.

Sources

American Banker, “05: Do Not Honor Card Refusals Are Confusing to Merchants”
Baremetrics, “5 Ways to Prevent Involuntary Churn in SaaS” (citing Stripe analysis on SaaS failed payment revenue loss)
CrowdFund Insider, “False Declines Costing Merchants More Than Fraud, Report Claims” (Chargebacks911)
Churnkey, “Do Not Honor Decline: Meaning, Stats, and How To Fix” (citing Visa Global Declines Analysis, 2016)
Cleverbridge, “Recover Failed Payments and Prevent Involuntary Churn with AI-powered Retry Logic”
GR4VY, “Approval Rates in Payments: Meaning and Deep Dive for 2025”
GR4VY, “What Is the Difference Between Hard and Soft Decline in Payments?”
MRC, “2025 Global eCommerce Payments and Fraud Report”
MRC, “2024 Global eCommerce Payments and Fraud Report”
Optimized Payments, “Network Tokenization: A Strategic Advantage in Modern Payments” (Mastercard data)
Payrails, “From Data Fragmentation to Strategic Control”
PYMNTS, “56% of US Consumers Experienced a False Payment Decline in Last 90 Days” (2024)
PYMNTS, “47% of Merchants Say False Declines Cost Them Sales” (2026)
Riskified, “The True Cost of Declined Orders”
Riskified, “Unlock Revenue by Optimizing Payment Authorization Rates” (athletic apparel brand case study, 82% to 95% lift)
Signifyd, “5 Strategies to Increase Bank Authorization Rates for Merchants”
Slickerhq, “Cut Involuntary Churn by 70% in 2025”
Spreedly, “How to Improve Soft and Hard Decline Rates”
Stripe, “AI Enhancements to Adaptive Acceptance” ($6B recovery in 2024)
Stripe, GAIA Customer Case Study (80% to 89%+ authorization rate)
Stripe Newsroom, “Zapier sees 4% uplift in auth rates with Stripe” ($3M+ additional revenue)
Userpilot, “Involuntary Churn vs Voluntary Churn in SaaS” (citing ProfitWell research, 20-40% of total churn)
Visa Acceptance Solutions, “Tokens Are Key to Future Proofing Payments” (4.6% auth rate lift, 30% fraud reduction)
Worldpay, “The C-Suite’s Guide to Payment Authorization Rates” ($10M per percentage point at $1B volume)
Worldpay, “Smarter Payments, More Revenue” (1.5% revenue uplift within 90 days)