Retention cohort analysis

When this is useful

You are about to run this process and need one reusable working document.

What the workbook covers

Metric contract

Cohort comparison

Fictional example

Investigation and action

Use with real evidence

Download the full Markdown resource, replace placeholders with your situation and retain sources beside claims. Keep unanswered questions visible rather than filling them with plausible guesses.

Put it to work

1. Read the full worksheet and define the decision it supports.
2. Fill the relevant tables using dated observations and explicit assumptions.
3. Review the output with its owner and record the next decision.

Limits and tradeoffs

The worksheet structures work; it does not supply missing customer evidence or guarantee a business result. Adapt examples and assign a responsible owner.

Worksheet preview

# SaaS retention cohort worksheet

Choose one retention definition and comparable observation periods. Examples are fictional.

## Metric contract

- Decision and owner:
- Unit: account, user, or another defined entity:
- Starting cohort event:
- Entry period and timezone:
- Eligibility/exclusions:
- Meaningful return activity:
- Period activity, exact-day, or another explicitly defined method:
- Observation intervals:
- Cutoff and fully observed periods:
- Original denominator rule:
- Source, definition version, and collection gaps:
- Account merge/deletion handling:

## Cohort comparison

| Cohort | Eligible starting accounts | Period 1 count/rate | Period 2 count/rate | Period 3 count/rate or incomplete |
|---|---|---|---|---|
| Fill in | Fill in | Fill in | Fill in | Fill in |

Period retention = original cohort accounts with qualifying activity during that period / eligible original cohort accounts. Count an account once per period even if it completes several actions. This definition permits return after an inactive period; it is not uninterrupted survival or revenue retention.

## Fictional example

Entry: account enrollment month. Month 1–3: the next three calendar months. Qualifying activity: at least one completed report in a month.

| Cohort | Starting accounts | Month 1 | Month 2 | Month 3 |
|---|---|---|---|---|
| April enrollment | 20 | 14 / 20 = 70% | 10 / 20 = 50% | 8 / 20 = 40% |
| May enrollment | 20 | 16 / 20 = 80% | 12 / 20 = 60% | Incomplete |

The Month 1 and Month 2 differences are two accounts, or ten percentage points. May Month 3 is not zero. Calendar enrollment can produce different within-month ages; use tighter age windows if required by the question.

## Investigation and action

- Comparable periods and maturity checked:
- Event and identity checks:
- Acquisition/product/support differences:
- Returning and inactive account observations:
- Hypothesis and alternative explanation:
- Proposed intervention:
- Customer-value/correctness guardrail:
- Population, window, owner, and next review:


---
Original resource by Jon Bishop. CC BY 4.0. Illustrative examples are not customer results.