Blog · Wallet share and penetration
Follow new-category purchases at existing customers through a comparable reorder window. Separate mature observations, recent first orders, trials and missing purchase history.
A cross-sell report can celebrate the first order in a new category without showing whether the customer purchased it again. The first order might be a trial, an urgent exception or part of a long purchasing cycle. A repeat-demand review separates that initial event from subsequent evidence of ongoing need.
Use cross-sell and upsell measurement for the revenue classification. This guide owns what happens to a customer-category pair after its first qualifying purchase, not another definition of expansion revenue.
Use the customer ID, a stable category, order identity, qualifying purchase date and net amount. Define the history required before calling a category new. A category absent from a short export is not necessarily a new purchase relationship.
Keep returning categories separate from categories never previously bought in the available history. Flag trial purchases rather than deleting them without explanation. The supplier-trial guide describes what acceptance of a trial does and does not establish.
Check product mapping when codes change. The replacement-versus-incremental growth guide helps prevent a renamed or substituted product from being treated as a new need.
Declare an observation cutoff and a reorder window before comparing cohorts. A 60-day window may be useful for one replenishment category and inappropriate for equipment bought annually. Use observed purchasing cadence and business context rather than a universal target.
For a simple fixed-window rate, a pair becomes mature only after the full window has elapsed and purchase history is available through that period. Recent pairs remain in a separate pending group. Record the number and value in both groups.
NIST's explanation of censored observations supplies statistical context: observation can end before an event occurs. This simple maturity gate avoids pretending an unfinished window is a completed outcome; it is not a survival-model estimate.
DEMO-701 contains 30 verified first purchases of a new replenishment category at existing customers. The illustration uses a declared 60-day reorder window and a common reporting cutoff. All purchase records needed for mature pairs are complete.
| Observation state | Customer-category pairs | Repeat purchase within 60 days |
|---|---|---|
| Full window observed | 20 | 12 |
| Less than 60 days observed | 10 | Pending |
| Total first purchases | 30 | Not one comparable denominator |
The mature repeat rate is 12 / 20 = 60%. Dividing the same 12 outcomes by all 30 pairs gives 40%, but wrongly treats the ten newer pairs as though each had a complete chance to repeat.
The mature result is still only 20 pairs. It does not establish a universal category benchmark, and the newer cohort may differ in customer mix. Show counts beside rates and avoid ranking small segments as if their differences were certain.
Use a separate qualifying order or purchasing event, not the number of invoice lines. Multiple lines, partial shipments and corrected invoices can belong to one original purchase. Returns and credits require their declared treatment.
Specify whether a repeat must have positive net value after linked returns. If records cannot resolve that question, report unknown status and recover the evidence. Do not force an unknown into “no repeat” just to complete the table.
The adoption-by-purchase-cohort guide examines longer-term product holdings in education. It can complement this reorder view without replacing its narrower event and maturity rules.
Review mature nonrepeaters for a recorded operational explanation: unsuitable specification, uncompleted trial, long inventory cover or lost access to the buyer. These are questions for the account owner, not conclusions produced by a missing order alone.
Choose an action such as confirming consumption, resolving a trial issue or accepting a one-time purchase. A recent pair may simply need its planned observation date. Preserve the difference between a reminder and a customer-facing pursuit.
Publish first-purchase count, mature count, repeat count, window, cutoff and unknown evidence. Keep resulting revenue separate from the repeat rate. Use an account-growth experiment if the question is whether an intervention caused additional purchases.
Read the cohort beside the customer-growth review example. Bring an authorized order sample to Covirage contact to agree the cohort view, observation window and review scope.
No. If the defined reorder window has not elapsed, report them as immature observations. Excluding them from the mature rate avoids giving newer purchases less time to repeat.
No. Split shipments, rebilling, credits or multiple invoices for one order can create additional records without another customer purchase. Use the agreed order or purchase-event identity.
No. They establish observed purchasing under the stated rules. Causal evaluation needs an appropriate comparison design and consideration of alternative explanations.