Key takeaways
- How to define an acquisition cohort that survives messy real-world data
- How to measure trial-to-paid conversion by source over 30, 60, and 90 days
- How to spot the channels that look great on day one but churn fast
- How to use cohort LTV to set channel budgets and bid caps
- How to feed cohort signals back to the ad platforms with CAPI
Why click and trial metrics lie
Most B2B SaaS dashboards optimize toward cost-per-trial. The problem is that cost-per-trial says nothing about cost-per-customer. Two channels with the same cost-per-trial can have wildly different trial-to-paid rates, and even more dramatically different LTV.
Cohort analysis solves this by grouping trials by when (and where) they came in, and watching what happens to that group over time.
Build the cohort the right way
Group trials by acquisition week or month. For each cohort, track the percentage that converted to paid by day 7, 14, 30, 60, and 90. Pair that with average revenue per converted account, and you have a cohort LTV curve per source.
- Cohort key: ISO acquisition week + first-touch source / campaign
- Conversion windows: 7, 14, 30, 60, 90 days post trial start
- Numerator: paid customers (Stripe "New Customer" trigger)
- Denominator: trials (Stripe "New Trial" trigger)
- Pair with cumulative revenue per cohort to compute LTV
Read the cohort like a pro
A healthy cohort looks like a curve that rises steeply from day 0 to day 14, then plateaus. A flat cohort means the trial isn't activating. A cohort that converts quickly but bleeds revenue by day 60 means the channel is bringing in unfit customers.
Pay closest attention to the slope, not the peak. The slope tells you product-channel fit; the peak tells you raw conversion rate.
Turn cohorts into action
Once a channel has 60-day cohort LTV that beats your CAC payback target, give it more budget. Once a channel underperforms over two consecutive cohorts, cap it. Sync the high-LTV trial events back to Meta, Google, and LinkedIn so the algorithm optimizes for the customers who actually pay over time, not just the ones who click.
Cometly · Guide