If you do one analytical exercise every quarter, it should be a cohort retention analysis. Not a customer count analysis. Not an ARR waterfall. A cohort-level retention analysis that shows what percentage of revenue from each quarterly acquisition cohort is still active at month 6, 12, 18, and 24.
This analysis answers the most important strategic question in your business: are the customers you're acquiring today better or worse quality than the ones you acquired 18 months ago?
If your 2024 cohorts have better 12-month retention than your 2022 cohorts, your product, ICP, and onboarding are improving. Your business is getting healthier over time.
If your 2025 cohorts have worse 12-month retention than your 2023 cohorts, something changed for the worse. Maybe you expanded ICP too broadly. Maybe the product moved away from core value. Maybe the market changed. Something is wrong.
The cohort analysis reveals this before it shows up in aggregate metrics. Aggregate NRR is a blended number that mixes the performance of old, well-embedded accounts with new, not-yet-proven accounts. Cohort analysis strips that blending away.
Running a meaningful cohort analysis requires:
Monthly or quarterly cohort grouping by acquisition date (or contract start date for annual subscribers).
Tracking both logo retention and revenue retention at fixed intervals (6, 12, 18, 24 months).
Including expansions and contractions in the revenue retention, not just cancellations.
Annotating the analysis with what changed in your product, GTM, or market at each cohort boundary. The retention inflection points correspond to real business events. Find them.
Run the cohort analysis quarterly. If you don't like what you see, fix it. If you like what you see, double down on what's driving it.