Start with the business game you are playing
Before naming a single metric, you need to be honest about the goal of your payments product. When I was at Amazon in Prime, we started with the type of customer we were trying to reach that couldn’t with the existing payment offerings we made available for purchasing Prime. Amplitude's North Star Playbook, developed by John Cutler, categorizes digital products into three engagement models: the Attention Game, the Transaction Game, and the Productivity Game. Payments is almost always the Transaction Game. The core question is: how many transactions does your user successfully complete on your platform? How many are recurring with a subscription product like Prime?
That framing immediately eliminates a class of vanity metrics. Session duration, page views, and daily active users are irrelevant at the top of your hierarchy. What moves the business is successful payment volume.
Shreyas Doshi, formerly Stripe's first PM Manager and a Director of PM at Twitter, frames product success across three dimensions: user adoption, customer satisfaction, and business impact. All three must show up in your metric structure, or you are measuring only one dimension of the product.
Map the payments funnel first
You cannot select meaningful metrics without first mapping the transaction flow through your system. The payments funnel has distinct stages, and drop-off at each tells a different story.
A standard e-commerce payments funnel: user lands on checkout, initiates payment, submits to your processor, the processor routes to the card network, the issuing bank approves or declines, and the result returns to the frontend. Each transition is a conversion step that needs measurement.
Checkout initiation rate measures how many users who reach checkout begin the payment flow. Authorization rate captures whether your processor got approval from the issuing bank. Payment success rate measures how many attempted payments resulted in a completed transaction.
As Stripe's documentation notes, if you process $500,000 in monthly orders at an 80% acceptance rate, $100,000 of attempted revenue does not convert to revenue. Why the gap? Where is the failure? Understanding where it is failing can lift your payment success rate to 95%. Getting to 95% recovers $75,000 without spending an additional marketing dollar.
These funnel metrics are your diagnostic layer. They tell you where the system has a leak. However, they are not your North Star.
Selecting the North Star Metric
The North Star Metric (NSM) should capture the core value your product delivers to users and reflect long-term business health. John Cutler puts it directly: "If you can move your North Star directly, it's probably not a good North Star." It should be a composite outcome that requires multiple inputs to move the number.
For a payment gateway or checkout product, two common NSM candidates are:
Total Successful Payment Volume (TPV): The dollar value of transactions successfully completed through your system in each period. This can also be measured in Payment/Transaction Success Rate (PSR/TSR). This ties directly to merchant revenue and signals that your system is performing simultaneously across acquisition, reliability, and user experience.
Authorization Rate: The percentage of submitted payment auth attempts that result in approval from the issuing bank prior to charge attempt. This captures the efficiency of your rails and is a leading indicator of revenue leakage.
Between the two, TPV is the stronger North Star because it integrates volume and quality. A high authorization rate on low volume is not a win. TPV forces your team to care about bringing more transactions in and converting those attempts into successful charges.
For a B2B payments infrastructure product where merchants are your customers, monthly processed volume per active merchant is a sharper signal because it controls for merchant count and measures engagement depth.
Funnel metrics as input metrics
Once you have your NSM, funnel metrics become input metrics, the levers your team can pull to move the North Star. John Cutler describes these as "the key inputs that are both descriptive and actionable" in the North Star Framework.
Here is how the payments funnel maps to input metrics:
Checkout Initiation Rate: Percentage of users who reach checkout and begin the payment flow. Low rates here signal UX friction, trust issues, or poor payment method coverage.
Authorization Rate: Successful bank approvals divided by total authorization attempts. Benchmarks for card-not-present transactions typically run 85-95%, segmented by card type, geography, and MCC code. Multi-Factor Authentication (MFA) also decreases this Auth rate by as much as 35%. A drop here signals issuer-side problems, fraud model over-triggering, or customers finding too much friction and not following through with the purchase.
Payment Success Rate: Broader than authorization rate because it includes retries. Checkout.com notes this should be calculated separately since second and third retry attempts can inflate the number without reflecting real first-pass improvement.
Checkout Abandonment Rate: The percentage of users who initiate a checkout flow but do not finish the transaction. Research from Optimized Payments suggests a 1% reduction in cart abandonment can generate a 10% revenue lift, depending on average order value.
Average Transaction Value (ATV): Useful for tracking whether the purchase mix is shifting, which affects fraud exposure and interchange costs.
These input metrics form the diagnostic and improvement layer beneath your NSM. They are where you run experiments and ship features.
Guardrail and health metrics
This is where most early and mid-career PMs underinvest when developing a metrics plan. Guardrail metrics ensure that optimizing aggressively for your North Star does not silently degrade another part of the system. These metrics are trip wires that alert you when a new feature is introduced and causes downstream impacts to payments.
Mixpanel's experimentation framework describes guardrails as metrics that prevent teams from "accidentally sending their product off-course while they conduct A/B testing." Airbnb operationalized this with their Experiment Guardrails Framework, monitoring for cross-team metric harm on every experiment. A launch that improves TPV while degrading another critical signal is not a clean win.
For a payments product, these are the guardrails worth maintaining:
Chargeback Rate: The percentage of completed transactions that result in a customer dispute. Card networks enforce hard thresholds, typically 1% for standard merchants and 1.5% for high-risk categories. Exceeding these triggers fines and can cost you card acceptance privileges. If you are running experiments to improve conversion by reducing friction, chargeback rate is the guardrail that tells you whether you removed the wrong friction.
Fraud Rate: The percentage of transactions flagged or confirmed as fraud. Authorization rate and fraud rate sit in direct tension. Loosening your fraud model to approve more transactions improves authorization rate but exposes you to fraud losses. This tension must be explicit in your metric framework.
Payment Latency (p95/p99): The time from payment submission to response. High latency damages conversion and user trust even on a successful transaction. This is a system health metric that sits outside your funnel but has direct impact on abandonment rate.
False Decline Rate: The percentage of legitimate transactions incorrectly declined by your fraud model. This is the counter metric that guardrails fraud rate. Reducing fraud rate too aggressively hurts real customers. Tracking false declines closes the loop.
Processor Uptime: Availability of your payment infrastructure. A gateway delivering 99.9% uptime protects conversion during high-traffic periods. Anything below that becomes a meaningful revenue event at scale.
How to frame the hierarchy
The structure assembled:
• North Star: Total Successful Payment Volume (or Monthly Processed Volume per Active Merchant for B2B)
• Input/Funnel Metrics: Checkout initiation rate, authorization rate, payment success rate, checkout abandonment rate, ATV
• Guardrail/Health Metrics: Chargeback rate, fraud rate, false decline rate, payment latency (p95), processor uptime
Shreyas Doshi categorizes product metrics into six types: health, usage, adoption, satisfaction, ecosystem, and outcome. Guardrails map to health and ecosystem. Funnel inputs map to usage and adoption. The NSM is the outcome metric. Run guardrails in your weekly ops review, input metrics in sprint reviews, and the NSM in your quarterly business review.
The pressure test
Ask whether each guardrail has a corresponding experiment type that could move it in the wrong direction. If it does not, it is probably not a real guardrail. For payments, loosening fraud rules threatens both chargeback rate and false decline rate. Simplifying checkout fields threatens authorization rate if users enter card data with less care. The guardrail and the experiment risk should be obvious to everyone on your team.
Also ask whether your NSM can be gamed without delivering real value. TPV driven by one merchant category or geography is a fragile signal. Segment your North Star. If one segment is masking problems in others, a single-number dashboard will mislead you until it is too late for a fix.
Payments is not hard to measure with metrics. The structure above gives you the foundation to do it right the first time.
---
Related reading: For a comprehensive overview of payments PM interview preparation, including authorization rates, fraud tradeoffs, and compliance questions, see Payments product manager interview questions and answers: the complete guide.