Lapsed Customer

A grocery customer may return weekly, while someone buying a sofa may wait years, so one 90-day cutoff would misclassify the furniture buyer. Define lapse by product, segment and expected behaviour.

Shopify explains recency, frequency and monetary segmentation and Klaviyo describes winback flows for customers who have not purchased in a chosen window, but neither provides one universal lapse threshold. A fictional coffee subscription renews monthly, so a customer whose plan ends and remains inactive for two cycles may be flagged as lapsed, while a yearly gift buyer would need a different window.

Use reliable purchase and account records, because returns, cancellations, subscriptions and guest checkouts can complicate the last-purchase date. A fictional retailer sees no recent orders under one email when the same person used a new email at checkout, so it reconciles records before sending a "we miss you" message.

A lapse definition can use days since last purchase, missed expected orders or a predictive model, and a simple fixed threshold is transparent but may ignore variation, so document the rule and its limitations. A fictional pet-food shop estimates customers reorder every six weeks and labels a missed second cycle for review, which helps prioritise outreach but is not a diagnosis.

Separate inactive, lapsed and churned states if useful, because 'churned' can imply a more durable loss and definitions vary, and a fictional software firm whose annual contract is still active despite little recent use has an account at risk, not already churned. Look at customers' previous frequency and value, since someone who bought once during a sale may not have the same return expectation as a loyal buyer, and a fictional bookstore finds that regular monthly buyers and one-time holiday shoppers who both have six months without orders are not equally lapsed.

Investigate reasons for inactivity such as price, competitor, poor service, changed need or ordinary timing, because purchase data alone cannot reveal motive and surveys and support history may help. A fictional customer who stopped buying after a damaged shipment will not be helped by a generic discount, so the store first makes the service issue right.

Winback efforts may use reminders, useful content, service fixes or offers, but consent and channel rules govern contact, so do not send marketing to someone who opted out just because a segment flags them; a fictional gym leaves former members who declined email marketing out of campaigns and analyses aggregate reasons for leaving instead. Measure reactivation against a suitable baseline, since some lapsed customers would return without an incentive, and a holdout can estimate incremental impact when feasible and lawful.

A fictional shop sends a coupon to a randomly selected half of an eligible lapsed group and compares repeat purchases with a holdout, which shows whether the offer added sales beyond natural returns. Discounts can erode margin or train customers to wait, so calculate net contribution after incentives, fulfilment and return costs, because a reactivated order is not automatically profitable; a fictional retailer offering 20% off to buyers who normally return unaided sees revenue rise in the report but margin fall, and narrows the offer after testing.

Revisit the lapse rule when seasonality or business model changes, as with a fictional garden centre that compares customers with the same seasonal pattern to avoid false alarms in winter. Lapsed status is an operational hypothesis, so use it to identify people worth understanding, then respect their choices and measure any response honestly.