No repeat purchase does not always mean dissatisfaction. Check the natural consumption cycle, product lifetime and customer situation. Compare customers who had the same opportunity to return.
Separate possible causes: the need ended, experience disappointed, the offer was forgotten or repurchasing is inconvenient. AI can group de-identified observations; it cannot establish a customer’s motive without verification.
An appropriate AI use case
Ask AI to propose segments based on observed behaviour and explain the evidence for each hypothesis. A person checks whether contact is permitted and which offer is actually useful.
For businesses across languages and cities, compare equivalent groups. The chosen communication language and purchase terms matter more than model assumptions about a person. Do not infer sensitive traits from purchase history.
Example retention map
Illustration. Purchase situation → expected next need → observed action → permitted channel → useful message → repurchase or opt-out. The test brief records a segment, observation period and comparison group.
Review purchases, unsubscribes, complaints and contact cost. More replies without useful repeat actions are a reason to revise the hypothesis. Message opens alone do not measure retention.
Common questions
Can we upload a customer database into AI?
Use de-identified or synthetic examples initially. Before using personal data, separately establish the lawful basis, access, storage, deletion and permitted provider.
What if there is no natural repeat purchase?
Explore a related need, service or referral rather than repeatedly pushing the same product. Choose a use case based on an actual customer need.
AI helps identify patterns and generate hypotheses. Verify outputs against source evidence, use de-identified data and keep a person accountable for decisions.
Set the period, cohort and repeat-purchase event. Compare like with like.
Cohort, period, repeat-purchase rate and missing data.
Step deliverable: Cohort, period, repeat-purchase rate and missing data.
2. Find return triggers
Review de-identified support themes and order history. Ask AI for patterns and alternative explanations; verify them manually.
Three observations, one hypothesis and an alternative cause.
Step deliverable: Three observations, one hypothesis and an alternative cause.
3. Test outreach
Prepare a relevant offer for one group. Respect communication consent, contact frequency and opt-out.
Who, when, which message, primary metric and guardrail.
Step deliverable: Who, when, which message, primary metric and guardrail.
4. Compare outcome
Compare repeat purchases and complaints with the baseline. Do not attribute impact to AI without a comparable control.
Actual result, limitations and decision: continue, revise or stop.
Step deliverable: Actual result, limitations and decision: continue, revise or stop.
How to judge whether it helped
You should leave with a document that supports a decision and a next action. Completing steps does not guarantee a commercial result. Measure the chosen change separately over an agreed period. If evidence is missing, state that and plan how to collect it.
A baseline, actual result and data limitations are recorded.
A person made the decision after reviewing quality and side effects.
How this differs from S01–S13
This is a user route for a specific problem. The GRI AI research stages S01–S13 and the 13-week programme remain in the library as separate methodological work. This route is not a validated GRI diagnostic scale.
Your route0 / 4 · 0%
Completed 0 · remaining 4
Next action: Define repeat purchaseThe percentage measures completed steps, not proven business growth.
PRACTICE / 1
Define repeat purchase
Set the period, cohort and repeat-purchase event. Compare like with like.
Step deliverable: Cohort, period, repeat-purchase rate and missing data.
No names, contacts, client secrets or sensitive details. Saving happens only when you press the button.
Discuss opens WhatsApp. Your answers are not sent automatically.
Data & privacy
Matching works without saving your query. Practice answers remain in this tab until you enable saving. Do not enter names, contacts, client secrets or sensitive information.
Server saving: answers and steps are linked to a random key in a secure browser cookie for up to 90 days. This is pseudonymous, not anonymous data. It does not sync across devices.
Server saving is not enabled yet: the data controller and hosting configuration are being confirmed. You can work in this tab and download your results.
