PLG conversion is laggy. Trials started in week 1 don’t become customers until week 2–4, and they don’t pay back the CAC until month 6–12. A cohort report is the only honest way to look at that economics — and the only way to keep your team from cutting paid budgets that haven’t had time to mature.
Building the report
Use the Cometly Report Builder to create a Cohort report. Group rows by trial-start month and source. Add columns for Spend, Trials Started, New Customers, MRR at 30/60/90 days, and LTV ROAS.
Set the attribution model to Source-Specific (the model that gives a channel credit any time it appears in the journey) and the window to Lifetime. This is the combination that produces the most honest channel-level cohort numbers for PLG.
- Rows: trial-start month × source
- Columns: Spend, Trials, New Customers, MRR @30/60/90, LTV ROAS
- Model: Source-Specific
- Window: Lifetime
- Filter to paid sources for clean ROAS comparisons
Reading cohorts
Cohorts mature over time. The trials you started in May won’t have a meaningful 90-day MRR until August. Always read the report top-down (oldest cohort first) and remember that the most recent cohort always looks worst because it hasn’t had time to convert.
Channel-level cohort patterns are stable. If LinkedIn’s 90-day cohort ROAS has been 4.2x for six months, you can budget against that. If Meta’s 90-day cohort ROAS has been bouncing between 0.8x and 3.5x for six months, that’s a creative or audience problem worth investigating.
What to watch for.
- Comparing cohorts that haven’t matured
Last month’s cohort is always lower than two months ago’s. Compare cohorts at the same age, not the same calendar date.
- Using last-touch attribution on cohorts
PLG journeys are multi-touch. Last-touch over-credits direct and branded search and hides the channels that actually drove the trial.
- Reporting cohort ROAS at 30 days
Most PLG cohorts have negative ROAS at 30 days because the LTV hasn’t happened yet. Use 90-day or longer for serious decisions.
Recap.
- Group rows by trial-start month and source
- Columns: Spend, Trials, New Customers, 30/60/90-day MRR, Year-1 LTV ROAS
- Filter to paid sources only when calculating channel-level ROAS
- Cohorts mature as time passes — the most recent cohort always looks worst
- Use this report to justify holding spend through the conversion lag