Why customers do not return: a practical repeat purchase review

First ask whether another purchase should happen yet

A customer who buys a durable desk may not need another desk soon. A customer buying a consumable product may have a different rhythm. Treating both businesses with the same repeat purchase target creates false alarms. Start with the job the product performs and the event that would create another genuine need.

Write a proposed repurchase window using product use, customer conversations and observed order histories. State the uncertainty. If different products have different cycles, analyze them separately. For a one-off service, a useful later outcome may be a new project or a referral rather than an identical purchase. The workflow below is a proposed diagnostic method, not an industry retention benchmark.

Reconstruct the first experience before sending reminders

Read the journey from payment to successful use. Did the order arrive as promised? Was the product what the customer expected? Could they understand instructions, get help and resolve a problem? A promotional message cannot repair an unresolved delivery issue.

Build a simple experience log:

Contact a small set of customers with permission and an open question: “What happened after you received it, and was anything harder than expected?” Avoid asking only satisfied customers if your goal is to discover failure. Separate complaints that need immediate resolution from broader ideas for improvement.

Define a cohort and a complete observation window

A cohort is a group sharing a defined starting event. For this worksheet, group customers by their first completed purchase month, using a consistent rule for canceled or fully refunded orders. Document that rule. Shopify’s customer reports include a cohort analysis organized around first orders and subsequent activity.

Choose the same elapsed window for each group. If you are evaluating a ninety day reorder window, compare only customers who have had the full ninety days. A customer acquired yesterday has not had the same opportunity as one acquired several months ago. Do not interpret an incomplete window as a confirmed failure to return.

Your own spreadsheet definition may differ from a platform’s report. Label it as such and reconcile exclusions before comparing the two. Keep the report date, first order date, customer identifier, product group, eligible observation window and repeat purchase status. Use only information needed for the analysis.

Separate people, orders and revenue

Write the metric in words before calculating it. For this proposed worksheet, the windowed repeat buyer rate is customers with at least one qualifying later purchase inside the chosen window divided by first purchase customers with a complete window. Define what counts as a qualifying purchase and how refunds are handled.

This is different from the share of this month’s customers who have bought before. It is also different from repeat orders divided by all orders. Those measures can all be useful, but their denominators answer different questions.

Track these alongside the rate:

• Eligible customers: the people with enough elapsed observation time.
• Repeat buyers: eligible people who purchased again under the written rule.
• Repeat orders: qualifying later orders, which can exceed the number of people.
• Net repeat revenue: later order revenue after the adjustments you explicitly include.
• Contribution: revenue after the costs you define, including any campaign discounts.

Do not announce improved retention solely because a few frequent buyers placed more orders. Check whether more people returned and whether those orders were economically useful.

Work through a hypothetical comparison

Suppose two fully observed cohorts each contain twenty first buyers. In an entirely hypothetical example, six customers in the first group and four in the second reorder within the chosen window. The worksheet rates are six divided by twenty and four divided by twenty. These invented figures illustrate arithmetic, not good or bad performance standards.

Now inspect context. Perhaps the second group bought a larger pack, so it lasts longer. Perhaps it arrived late. Perhaps it came through an introductory discount attracting a different need. The numbers identify questions; they do not prove which explanation caused the difference.

Do not split a tiny dataset into many categories and treat every difference as meaningful. Keep the small counts visible and combine them with conversations.

Match the intervention to the reason

If customers struggle with first use, test clearer onboarding or instructions. If delivery was unreliable, repair fulfillment. If the product is unavailable when needed, investigate stock and reorder visibility. If people simply do not need it again yet, a more aggressive message is unlikely to answer the real issue.

A proposed postpurchase sequence is: confirm receipt, offer relevant help, ask about unresolved problems, then offer a timely reorder route when a genuine need is plausible. Use the customer’s communication choices and make stopping messages straightforward. Avoid sending replenishment reminders to someone whose complaint remains open.

For a physical brand, the packaging and identity guide can help review instructions and the experience around the product. If a satisfied one-time buyer has no natural reorder need, a respectful referral process may be more appropriate than pushing another purchase.

Test one change without hiding its cost

Create a test record with the eligible group, reason for the change, message or service improvement, comparison method, observation window and decision date. Where feasible, compare similar customers receiving different treatments during the same period. If you cannot do that, describe a before-and-after comparison as directional and note other changes.

Shopify’s reporting overview explains that cohort views help inspect retention over time. A report is a way to examine behavior, not proof that a campaign caused it.

Measure support demand, opt-outs, refunds and contribution alongside purchases. A discount may increase orders while reducing what remains after costs. A useful instruction message may improve the experience even before another purchase is due. Choose the success measure that matches the intervention rather than changing it after seeing results.

Make the next review small and actionable

Choose one product group, one completed window and one experience issue. Export or record the necessary order history, calculate the written metric and read a few relevant support conversations. Select one repair with an owner and a review date.

If there is too little history, begin with reliable data and customer learning rather than a dramatic retention claim. A brand consultation can help connect the promise, product experience and follow-up. The immediate goal is to understand why another purchase should happen and remove the obstacle that prevents an appropriate customer from making it.