Skip to content

Trial Conversion

The clearest leading indicator in a subscription business — and the one metric where last quarter's number is expected to move.

The idea

Trials run 7–14 days typically, 30 for enterprise. The conversion rate is the fraction of started trials that become paid subscriptions. Freemium conversion, a related but distinct motion, typically runs 2–5%.

Trial conversion leads MRR by roughly one billing cycle, which makes it the earliest reliable signal that acquisition quality or onboarding has shifted — well before it shows up in revenue.

The retroactive cohort problem

A trial started in January that converts in March: which period does it count in?

Tidemill uses the retroactive cohort model, matching ChartMogul. The trial is fixed to the period of its trial_started event and its eventual outcome rolls back to that cohort no matter when it lands. A March conversion updates January's rate.

The alternative — counting conversions in the period they occur — keeps history immutable but produces a number that is not a conversion rate of anything coherent, mixing numerator and denominator from different populations.

The trade-off is real and worth stating plainly: recent periods are provisional. Trials that are neither converted nor expired are still pending, so the most recent months will move as they resolve. A dashboard reader who does not know this will misread a dip.

How Tidemill computes it

Conversions over trials started, both scoped to the cohort period by started_at, with conversion counted at any later time.

Authoritative formulas: definitions.md — Trial Conversion Rate. Implementation: tidemill/metrics/trials/.

Known documentation drift

definitions.md files Trial Conversion under "Planned (P1) — designed but not yet implemented", but tidemill/metrics/trials/ ships and is registered. The formulas are current; the status label is stale. Recorded in the wiki log on 2026-07-23.

Contested ground

  • Do trials count as active subscriptions? They affect customer counts, ARPU, and therefore LTV if included. Tidemill tracks trials as their own funnel.
  • Expired vs. cancelled trials. A trial abandoned on day two and one that ran its full length are both non-conversions, but they mean different things about the product.
  • Provisional recent periods. Shared with ChartMogul, and a frequent source of "why did last month's number change?" confusion.
  • MRR — what conversion feeds
  • Churn — the other end of the lifecycle
  • Baremetrics — Trial Insights as a product feature
  • Cohort Retention — same cohort logic, different event

Sources