Retention cohort analysis shows whether people or accounts that started together continue to receive value over time. For a Series A discussion, it can turn a broad growth story into a more useful question: which customers return, how do comparable groups differ and what is the evidence mature enough to show?
Use the worked worksheet below to prepare a clear, inspectable analysis. It does not set a universal Series A retention threshold. The right event, period and population depend on the product and business model.
What is a retention analysis?
A retention analysis follows a defined population after a starting event and measures whether its members meet a specified condition later. The condition might be meaningful product use, continued subscription or recurring revenue from the original customers. Those measures answer different questions and should not share an unlabeled “retention” heading.
Sequoia's retention guidance emphasizes choosing relevant activity and time windows. Start there: a login can be a poor proxy for customer value if the product's useful action is completing a task or transaction.
| Definition | Decide explicitly | Example for this worksheet |
|---|---|---|
| Unit | User, account, location or another entity | Customer account, deduplicated |
| Starting event | What makes an entity enter the cohort | First completed production workflow |
| Return event | What counts as meaningful activity | At least one completed production workflow |
| Period | Calendar buckets or fixed age windows | Successive 30-day age windows after entry |
| Exclusions | Tests, internal use, fraud or invalid records | Internal and test accounts excluded consistently |
| Observation date | The latest reliable data point | Recorded with the export |
The example uses return activity within each age window, excluding the initial completion that created the cohort. It is not an “ever returned after this day” measure. A customer can be inactive in one window and return in another. State your own method so readers do not mistake one retention definition for another.
Can you give me an example of a cohort analysis?
These fictional accounts and results are invented for teaching. They are not Finta analytics, customer outcomes or suggested performance targets. All accounts in a displayed cell have completed the relevant observation window.
| Entry cohort | Starting accounts | First 30-day return window | Second return window | Third return window |
|---|---|---|---|---|
| A | 100 | 70 / 100 = 70% | 65 / 100 = 65% | 63 / 100 = 63% |
| B | 80 | 62 / 80 = 77.5% | 58 / 80 = 72.5% | Not yet observable |
| C | 50 | 40 / 50 = 80% | Not yet observable | Not yet observable |
Each percentage uses that cohort's original eligible account count as the denominator. For cohort B's second window, divide 58 by 80, not by the 62 accounts active in the preceding window. The latter calculation measures a different conditional question.
At the same first-window age, the fictional cohorts show 70%, 77.5% and 80%. That is a descriptive pattern, not proof that a product change caused the improvement. Acquisition channel, customer mix, onboarding and measurement changes could also matter.
Compare equal ages, not the newest cells
It would be misleading to compare cohort A's third-window 63% with cohort C's first-window 80% and claim a 17-point improvement in long-term retention. The groups are at different ages. Compare the same column, with the same definition and enough observation time.
Keep immature cells as unavailable. They are not zero and should not enter an average as zero. If only some members of a cohort have completed a period, either wait for the cohort to mature or define and label a separate eligible subset consistently.
For the first window across all three fictional cohorts, 172 of 230 accounts return: 74.8%. That weighted rate differs from the simple average of the three percentages. Show the calculation and cohort sizes, particularly when the groups are uneven.
What does 90% retention mean?
It means little without a definition. Ninety of 100 original accounts completing a meaningful action in a specified period is 90% activity retention under that definition. Keeping 90% of starting recurring revenue is a revenue-retention measure, not a count of returning accounts.
Always attach the population, event, elapsed time and denominator. Do not treat a generic percentage as evidence that every investor will consider the company ready for Series A.
Build a Series A cohort evidence pack
- A one-paragraph metric definition and explanation of why the activity matters.
- The reproducible source query or export, observation date and responsible owner.
- A cohort table with counts, percentages and immature cells identified.
- A concise account of segment changes, tracking changes and known data gaps.
- The business questions suggested by the pattern and the next test.
Keep product activity, paying-customer retention and revenue retention separate. Reconcile any revenue measures with the finance owner's definitions before sharing them. An activity export alone does not establish an accounting result.
Turn the analysis into a clear investor explanation
Keep the accepted definitions, evidence and investor questions together in Finta. Aurora can help draft a concise explanation from your reviewed analysis while preserving the limits. Your analytics source remains the place to verify the underlying numbers.
Review this retention cohort worksheet for an investor discussion.
Unit, entry event, return event and age-window definition: [paste]
Source data, exclusions and observation date: [paste]
Cohort starting counts, return counts and immature cells: [paste]
Calculate percentages using original eligible cohort denominators.
Compare equal-age windows and show counts beside percentages.
Keep unavailable periods distinct from zero and identify tracking or mix changes.
Draft a factual explanation and next questions, without causal claims or a universal Series A benchmark.
Prepare a review only. Do not change analytics, share private data or send an update.Place this evidence beside the Series A fundraising guide and use the investor update workflow for a reviewed communication.
By Finta Editorial Team. Research reviewed November 12, 2026. All worksheet numbers are fictional.