PLG Leak Triage: Find Where Trials Die
A systematic 10-day hunt through your product funnel to find — and rank — the leaks killing trial-to-paid.
Who this is for
Teams with real signup volume but disappointing trial-to-paid conversion.
What you leave with
Your three biggest leaks, quantified in lost revenue, with a fix sequenced for each.
Define the moments that matter
Days 1–2
You can't find leaks in an undefined pipe. First, agree on what the journey is supposed to look like.
Define activation precisely
The single moment a user first experiences your core value — specific enough to query ('created first agent AND ran it successfully'), not a vibe ('engaged user').
Map the intended path
Signup → setup complete → aha moment → habit (2nd/3rd value moment) → conversion trigger (limit, seat, feature gate) → paid. Six stages, one line each.
Pull the cohort baseline
Last 90 days of signups, stage by stage. The drop-off table IS the leak map — most teams have never actually looked at it end to end.
Phase artifact
The funnel definition + a cohort table showing % surviving each stage.
Hunt the leaks
Days 3–7
Numbers locate leaks; only evidence explains them. Combine the cohort math with session-level forensics.
Rank drop-offs by revenue impact
For each stage transition: (users lost) × (downstream conversion) × (ACV). The biggest percentage drop is often NOT the biggest revenue leak — do the math before picking targets.
Watch 20 real sessions
Ten who converted, ten who stalled at your biggest leak. Session replays or recorded onboarding calls. You're looking for the moment confusion sets in — it's rarely where the team guesses.
Interview five stall-outs
Users who activated but never converted. One question matters most: 'What were you hoping would happen that didn't?' Their words become your fix spec (and your landing-page copy).
Segment before concluding
Split every leak by source and ICP-fit. A 'product leak' that only exists for off-ICP traffic is actually a targeting leak — fixing onboarding won't move it.
Phase artifact
Leak ledger: each leak with size ($/quarter), evidence, and suspected cause.
Fix in sequence
Days 8–10 (then weekly)
Triage means fixing by leverage, not by which leak is most interesting to debate.
One leak, one experiment, one metric
Start with the top revenue leak. Design the smallest intervention that could plausibly halve it — a checklist, a template, a default change — with a success metric and a 2-week decision date.
Instrument the conversion triggers
Limit-hits, seat invites, gated-feature attempts are buying signals. Make sure each fires an event — and route high-fit accounts hitting them to a human while they're warm.
Install the weekly leak review
Fifteen minutes: cohort table, active experiment, decision. Leak triage isn't a project — it's a metabolism. The table you built in phase one becomes the standing scoreboard.
Phase artifact
A running experiment queue ranked by revenue, reviewed weekly against the cohort table.
Want this run for you, with you?
Every playbook here is a phase of the Growth Intelligence OS. We run them inside your team — your Slack, your stack, your numbers.