Field note · 22 July 2026
Your Trial Conversion Rate Is Only as Honest as Its Denominator
Why eligibility, delayed outcomes and mixed offers can make a familiar trial-to-paid percentage answer the wrong question.
A trial conversion rate appears simple: paid starts divided by trial starts. The disagreement begins when a team asks what belongs in either count.
Start with eligibility, not the chart
A store may prevent a customer from receiving an introductory offer twice, while product analytics can still record that person touching the trial paywall. If the denominator includes ineligible visitors, it measures paywall traffic—not trial performance.
Build the cohort from confirmed trial entitlements. Keep direct-to-paid purchases separate. Then attach the offer identifier, start timestamp, platform and market as they were known at that moment.
Let every cohort mature
A seven-day trial that began yesterday cannot be compared with one that began a month ago. The recent cohort contains unresolved members by construction. Either wait for the full conversion window plus billing tolerance or report mature and pending outcomes separately.
Do not blend unlike offers
Monthly and annual plans can have different trial lengths, prices and intent. A shift in their mix changes the blended rate even when each offer’s own performance is flat. Show offer-level rates first, then calculate a mix-adjusted comparison if leadership needs one headline.
The trustworthy version is rarely the fastest percentage to produce. It is the one whose population, observation window and billing states can be explained in a sentence.