Repeat Purchase
Part of Retention for low-frequency purchases
Distinguishing customer engagement from future revenue
Separate post-purchase interactions, genuine buying opportunities, observed sales and conditional revenue forecasts.
Engagement records an interaction now; future revenue requires a later completed purchase. For a long-lasting product, reading a guide, clicking an email or asking a service question may be useful without indicating a near sale. Keep interactions, emerging buying opportunities, observed sales and forecasts separate.
Label what each event shows
Start with the action, not the dashboard's name for it. A recorded email open may reflect an automated image load rather than a person's attention. A care-page visit shows access to information.
An accessory enquiry suggests a possible related need. A completed order records a purchase, subject to later refunds. None of these events alone reveals what would have happened without the contact.
| Recorded event | Defensible reading | Claim to avoid |
|---|---|---|
| Care content viewed | A visit to help content was recorded | The customer is ready to replace the product |
| Service case marked resolved | A case status changed | The customer received a satisfactory remedy |
| Accessory enquiry | A possible related need was expressed | The accessory will be purchased |
| Completed related order | A sale occurred under the stated order rule | Earlier engagement caused it |
A service contact can begin with a fault. Keep the issue, action and customer's response visible instead of treating contact volume as an unqualified positive score.
Connect activity to a buying opportunity
For each activity, write the plausible sequence before reporting results. A setup guide may help owners use the product and has no necessary immediate sales step. A compatibility guide may lead from a relevant question to an accessory enquiry and a completed order. A replacement reminder needs a credible replacement occasion.
Name who could reach each step. An accessory is relevant only to owners of compatible models; a replacement decision may be years away. Keep people without enough follow-up pending.
The sequence is a hypothesis. A customer might have intended to buy an accessory before viewing the guide, while another might solve a problem without buying anything.
Keep forecasts conditional
If a financial plan needs an estimate, start with an eligible group and a defined opportunity. State the proportion assumed to develop a relevant need, the share assumed to buy, expected order contribution and time horizon. Include people expected to place no order. Show how the estimate changes when a material assumption changes.
Call the result projected contribution under assumptions. It is not observed revenue, guaranteed customer value or proof of what the content caused. If an enquiry-to-order rate has no defensible basis, leave that step unquantified rather than borrowing an unrelated benchmark.
A report can label the lines plainly: guide use observed; accessory enquiries observed; completed accessory orders observed; future accessory contribution projected.
Ask the causal question separately
To assess whether an optional message changes purchasing, compare customers assigned to receive it with customers assigned to the usual experience. Use the same eligibility, outcome definition and follow-up. Review completed orders and contribution for everyone assigned, including people who did not click. Comparing clickers with non-clickers is weaker because the groups selected themselves after the message.
Rare purchases may leave too few orders for a reliable revenue conclusion. Report that uncertainty. Content use and the ability to find support may still be assessed, but each result needs its own label.
