# Retention and Loyalty Workbook

Use equal observation windows to understand repeat buying, then model whether an incentive can create profitable additional purchases. All sample numbers below are hypothetical.

## Define a qualifying purchase

- Customer identity rule: [how duplicate accounts or guest orders are handled]
- Included first orders: [definition]
- Treatment of tests, cancellations, full refunds, exchanges and renewals: [rules]
- Follow-up window: [number of days and reason]
- Observation cutoff: [date]
- Eligibility rule: [each included customer's full window has elapsed]

## Cohort table

| First-order cohort | Eligible first-time customers | Customers with second order within window | Second-purchase rate | Contribution per eligible customer | Important mix changes |
| --- | --- | --- | --- | --- | --- |
| Example A | 100 | 22 | 22% | [calculate] | [record] |
| Example B | 80 | 12 | 15% | [calculate] | [record] |
| [your cohort] | [A] | [B] | [B/A × 100] | [total contribution/A] | [record] |

If A is zero, the rate is undefined. Count each second purchaser once, even if they place several later orders. Keep incomplete cohorts separate. A seven percentage-point difference between the examples does not establish a cause.

## Customer explanation

- Natural next purchase or need: [replenishment, accessory, gift or none]
- Evidence about typical timing: [order history and customer feedback]
- First-order problems to fix: [delivery, use, quality or availability]
- Helpful message and eligible audience: [details]
- Suppression rules: [recent purchase, return, complaint, marketing eligibility]
- Primary measure and guardrail: [details]

## Loyalty break-even model

Keep the cost treatment consistent. Contribution before program costs should already include ordinary product, payment, fulfillment and other variable costs. Do not subtract redeemed discounts again if they are already reflected in contribution.

| Input | Hypothetical example | Your estimate |
| --- | --- | --- |
| Pilot software and extra staff cost | $500 | [amount] |
| Reward expense not already in contribution | $300 | [amount] |
| Total incremental program cost | $800 | [amount] |
| Contribution per additional order before those costs | $20 | [amount] |
| Additional orders needed to break even | 40 | [cost / contribution] |
| Expected genuinely additional orders | 25 | [estimate and evidence] |
| Modeled result | -$300 | [additional orders × contribution minus cost] |

If contribution per additional order is zero or negative, this model has no positive break-even order count. Reward redemptions and attributed orders are not the same as additional orders caused by the program.

## Pilot decision record

- Eligible population and assignment method: [details]
- Offer and usual-experience comparison: [details]
- Pilot dates, cost limit and stop condition: [details]
- Measure for all assigned customers, not just enrollees: [result]
- Contribution, reward cost and support workload: [results]
- Confounding changes and uncertainty: [details]
- Decision and owner: [expand, revise, stop or continue observing]

Read the [second-purchase review](https://ecomcoach.xyz/learn/second-purchase-cohort-analysis/) and [loyalty economics guide](https://ecomcoach.xyz/learn/ecommerce-loyalty-program-economics/) for interpretation.