Frequently asked questions
What's the right way to slice a cohort?
By the event that started the relationship: acquisition month, contract start date, or feature first-used date. Avoid mixing cohort definitions across reports. The comparisons stop being valid.
What metric should you watch in cohort analysis?
Retention curves (revenue or logo retention at month 6, 12, 24). Then layer in expansion to see net retention. Single-metric cohort analysis misses the texture of which segments grow vs decay.
How is cohort analysis different from regular reporting?
Regular reporting aggregates across time, which hides which customer groups are healthy vs deteriorating. Cohort analysis isolates each group's trajectory, which is what you need to know if early-2024 customers behaved differently than late-2025 customers.
What sample size do you need for cohort analysis to be useful?
At least 30 customers per cohort. Below that, noise dominates signal. Companies under 200 customers usually need 6-month cohorts; above 1,000 customers can use monthly cohorts.
How is cohort analysis useful in outbound?
For identifying which sourcing channels produce the longest-lasting customers. Outbound-sourced customers from Q1 2024 might retain at 95 percent while paid-channel customers from the same period retain at 70 percent. A signal worth thousands in budget reallocation.
.png)
.png)