Merchant Activation and Retention Metrics: Define the Lifecycle Before the Dashboard

Define activation as the first repeatable customer value event, not an account-created event, then measure retained merchants as cohorts with payment-quality guardrails. For a payments platform, a merchant may be verified but never integrate, integrate but never process, or process once through a test credential. Those are different states with different owners.

Retention tools make the definition choice visible: Amplitude distinguishes “return on” from “return on or after,” and each uses a different denominator and time rule. Amplitude retention documentation The lesson applies regardless of tool: publish the event and denominator before comparing cohorts.

Create a merchant lifecycle dictionary

Choose the activation event by product. A gateway may use the first settled production transaction; a payout API may use the first completed payout with a reconciled result. Do not use “API key created” if the product promise is money movement.

Worked illustrative example

Illustrative example, not a reported implementation. A platform finds many merchants create credentials but few complete a first live payment. Build a funnel for eligible merchants: application started → verification completed → production key issued → first valid request → first settled transaction → second settled transaction within 30 days.

The team sees a drop between key issue and first request. Before changing onboarding, inspect integration errors, documentation searches, partner availability and merchant segment. A guided test mode may help, but it could also create false confidence if its responses differ from production. The decision is based on the failure evidence, not the largest chart drop alone.

Track time-to-first-settled transaction and repeat-settled rate by entry cohort. Pair them with guardrails: verification rework, failed-payment rate, support contacts, fraud/risk review rate and exception age. A faster activation metric is not progress if unsafe merchants simply reach production sooner.

A cohort calculation with explicit denominators

Invented numbers for explanation. Take 100 eligible merchants verified in the same entry week, each observed for a full 30 days from verification. Sixty reach their first settled production transaction within that window. The 30-day activation rate is 60/100, or 60%. Forty-five of those 60 complete a second settled transaction within the same window: 75% of activated merchants, or 45% of the original eligible cohort. Name both measures; they answer different questions.

This second-transaction measure is not day-30 retention. If retention means returning in days 31–60 after activation, wait until each included merchant has that full window. Show newer cohorts as incomplete, not as churned. Report time to first settled transaction among activated merchants alongside the activation rate so unactivated merchants do not disappear from the interpretation.

Authorization success uses payment attempts as its denominator. Merchant activation uses eligible merchants. A higher authorization rate can coexist with fewer merchants activating; inspect both before attributing progress to onboarding. Keep the chosen cohort, timezone, non-test filter and late-settlement update rule with the metric definition.

Make retention interpretable

State whether retention means a merchant transacts on day 30, on or after day 30, or reaches a monthly processing threshold. Report the cohort start date, timezone, eligible population and missing-data rule. Segment only after the base view is trusted; otherwise a partner outage or a country mix change can be misread as an onboarding result.

Review a cohort with operations. Payment-state delays can make a transaction appear absent today and settled tomorrow. Preserve event timestamps for request, authorisation, capture, settlement and payout rather than replacing them with one generic “payment completed” timestamp.

Practical next steps

1. Publish the lifecycle dictionary and owners.
2. Audit three recent merchant journeys against the event definitions.
3. Establish one activation and one retention cohort baseline.
4. Investigate the largest evidence-backed drop before proposing a feature.
5. Review the selected intervention with the onboarding conversion guide and the payments toolkit.

Metrics do not create retention. They make the customer’s journey, the product promise and the operating limits visible enough to improve together.