Log Ninja — conversion intelligence
Behavioural analysis for a driver logbook subscription product. I defined the cohorts, ran the analysis and turned the findings into a measurable activation experiment.
The problem
Trial-to-paid conversion looked flat, but the raw number was misleading. Legacy free-product users, trials that had not yet matured, and users who cancelled after paying were all being counted together.
The real question was not who converted, it was who activated during their trial and who did not.
What I did
I separated post-paid-launch users from the legacy free-product cohort, excluded trials that had not yet matured, and defined conversion as any user who had ever made a successful subscription payment. I only measured product behaviour that occurred before the first payment, so activation could be assessed without post-payment noise.
The analysis compared users with zero-or-one meaningful-use days against those with two or more. The gap was large enough to reframe the problem — the main issue was not expired trials, it was failure to activate.
The output
I estimated the recurring revenue upside of closing the activation gap, cleanly separating potential from guaranteed. Then I proposed a controlled second-trial for users who completed a first trial without creating a log, with defined stages to measure: restart, first log, second meaningful-use day, subscription.