Product-Led Growth (PLG) & User ActivationPlaybook3 min readUpdated September 2026

A Runbook for Growing LTV Cohort by Cohort

Blended lifetime value hides more than it shows. A number that averages every customer you've ever had together can sit flat for a year while a cohort that signed up after a pricing change is quietly worth twice as much as the one before it, and nobody notices because the average smooths the difference out.

Cohort analysis fixes this by grouping customers by when they signed up, then tracking each group's revenue and retention separately. Once you can see which cohorts are actually growing in value and which are shrinking, expansion campaigns can target the specific behavior that's different between them.

Vendors Covered in this Article

Disclosure: We may earn a commission if you buy through some links on this page. It doesn't change what we recommend.

Step 1: define the cohort boundary that matters

Monthly signup cohorts are the default, but they're not always the right boundary. If you shipped a major onboarding change or a pricing update at a specific point, split cohorts around that date instead, since that's the boundary that actually explains a difference in behavior.

Keep the boundary consistent once you've picked it. Switching between monthly, quarterly, and event-based cohorts across different analyses makes it hard to build institutional memory about which cohorts are actually strong, since nobody can compare this quarter's view to last quarter's.

Step 2: track LTV and retention per cohort, not just revenue

Pull retention curves for each cohort side by side, not just current revenue. A cohort with lower starting revenue but a flatter retention curve will often overtake a bigger, faster-churning cohort within a year, and that's the group worth studying for what it's doing differently.

Chart cumulative revenue per cohort over time rather than a single current snapshot, since a snapshot hides the shape of the curve. Two cohorts can have identical revenue today and completely different trajectories, one still climbing and one already flattening out.

Step 3: find what the strongest cohort actually did differently

Compare the top-performing cohort against the others on early product usage, not just demographics. Look for a specific action taken in the first month, a particular feature adopted, an integration connected, a second user invited, that shows up more often in the cohort that's expanding faster.

Resist the temptation to explain the difference with something you can't act on, like the cohort just happened to include better-fit customers. Even if that's part of the story, keep digging for a behavior difference, since that's the part you can actually turn into a campaign.

Step 4: build an expansion play around that action

Once you've found the behavior that predicts expansion, build a targeted campaign that pushes underperforming cohorts toward it. This might be an in-app prompt, a direct email sequence, or an outreach from customer success, depending on how much friction is involved in taking the action.

Match the channel to how much explanation the action needs. A one-click setting is a good fit for an in-app prompt; something that requires understanding a new workflow usually needs a fuller email or a human conversation to actually land.

Step 5: run the play against one cohort before rolling it out

Test the expansion play against a single cohort that hasn't taken the target action yet, and compare its LTV trajectory against a similar cohort that didn't get the play. This is the only way to know whether the campaign actually moved the number or whether that cohort was just going to expand anyway.

Give the test enough time to show up in the retention curve, not just an immediate response rate. A play can get a good click-through rate and still fail to move LTV if the behavior it drove doesn't actually stick past the first use.

Where cohort analysis goes wrong

The most common mistake is redefining cohort boundaries every time you look at the data, which makes it impossible to compare one analysis to the next. Pick a boundary, usually monthly, and stick with it for at least a year before changing it, even if a different boundary looks more interesting in the moment.

A shared board in ClickUp keeps the cohort definitions, the hypotheses you're testing, and the results of each play in one place, so the next person to run this analysis isn't starting from scratch. Once you've identified a play worth automating, ConvertKit can trigger the sequence based on cohort and behavior rather than requiring a manual send each time.

The runbook in short:

  1. Pick one cohort boundary, monthly signup by default or a date tied to a major onboarding or pricing change, and keep it consistent across analyses.
  2. Chart retention and cumulative revenue per cohort side by side, so a flatter retention curve with lower starting revenue does not get overlooked.
  3. Compare the strongest cohort's first-month behavior, such as a feature adopted, an integration connected or a second user invited, against the other cohorts.
  4. Build an expansion play that pushes lagging cohorts toward that action, choosing an in-app prompt, email sequence or customer success outreach based on how much explanation the action needs.
  5. Test the play on one cohort that has not taken the action, compare it with a similar untouched cohort, and wait long enough for the retention curve to show a difference.
Executive Capability Standard

What Good Looks Like

Good cohort LTV work means every signup cohort is tracked separately on retention and revenue, and expansion campaigns are built around a specific early behavior shown to predict higher LTV, not a blanket offer sent to everyone.

Building The Capability (5-Stage Skill Ladder)

1. Learn:Pull retention curves for your last six signup cohorts and see how much they actually differ from each other.
2. Do Manually:Manually compare your strongest and weakest recent cohorts on first-month product usage to find a candidate behavior worth testing.
3. Delegate:Give a lifecycle or growth marketer ownership of running the cohort analysis and expansion plays on a recurring basis.
4. Automate:Trigger expansion plays based on cohort and behavior automatically through a tool like ConvertKit instead of manual sends.
5. Buy:Bring in a data or growth analyst if your cohort curves are noisy enough that you can't tell a real signal from normal variation.

How to Get Started

Disclosure: We may earn a commission if you buy through some links on this page. It doesn't change what we recommend.

Frequently Asked Questions

How many customers do we need before cohort analysis is useful?

You can start looking at cohort trends with as few as a couple dozen customers per cohort, though the patterns get more reliable as that number grows. The bigger constraint is usually time, since you need at least a few months of a cohort's behavior before its retention curve tells you much.

Should we cohort by signup date or by plan type?

Start with signup date, since it captures changes to your product and onboarding over time that plan type won't show. Once you have a signup-date view working, layering plan type on top can help explain why one cohort outperforms another, but it's rarely useful as the primary boundary.

What if two cohorts look the same but one expands and one doesn't?

That's usually the most useful comparison you can run, since it rules out obvious explanations like seasonality or pricing changes. Dig into first-month product usage specifically, since a difference that doesn't show up in signup demographics is often hiding in what customers actually did after they signed up.

About the numbers

This guide doesn't quote a sourced benchmark. Figures in it are estimates or general guidance, so check them against your own numbers.

Related Guides