Repeat purchase rate is often stated as customers who purchased more than once divided by total customers in a defined period. The formula is simple; the cohort definition is not. First-time customers near the end of the window have less opportunity to repeat, guest and logged-in identities can split one person, returns can change purchase status, and subscriptions can behave differently from discretionary orders. This lesson makes the denominator, observation window, and exclusions explicit before the percentage reaches a dashboard.
VISUAL LESSON
What you will learn
- 01Define the customer cohort and observation window.
- 02Calculate repeat purchase rate from cleaned customer counts.
- 03Use segments and scenarios without inventing benchmarks.

ILLUSTRATIVE WORKED EXAMPLE
Calculate an illustrative 90-day repeat purchase rate
PRACTICAL INTERFACE MAP
Move from raw orders to a decision-ready rate
Define first eligible purchase, business timezone, window length, products, channels, cancellations, returns, subscriptions, tests, and customer identity.
Count every customer once in the denominator and those with more than one eligible purchase once in the numerator, then divide and document.
Compare mature cohorts and useful segments, inspect absolute counts and margins, and choose one retention experiment with guardrails.
STEP-BY-STEP LESSON
Defined cohort → cleaned customer history → repeat fraction → retention action
THE LEAD ATLAS METHOD
Lead Atlas Data can research a fresh done-for-you list of business contacts matched to the growth campaign’s categories, market, and locations, providing an acquisition cohort that is kept distinct from the existing-customer repeat-purchase analysis.See how custom list research works ↗Define the business question and cohort
State whether the team wants to understand early retention, category replenishment, subscription behavior, channel quality, or another decision. Define cohort entry, observation length, calendar or rolling logic, timezone, stores, channels, products, currencies, customer types, and the comparison period.
Give every cohort an equal chance to mature before comparing it. A customer acquired yesterday cannot fairly be compared with one observed for 90 days, and a period-based report can differ from a first-purchase cohort even when both use the same formula.
Create the eligible purchase rule
Define which order states count, when a purchase becomes eligible, and how cancellations, full and partial returns, exchanges, free replacements, tests, fraud, internal orders, subscriptions, gift cards, and marketplace orders behave. Preserve gross and net perspectives when both matter.
Choose the customer identity key and document how guest checkout, email changes, merged accounts, household purchases, business accounts, and privacy-driven deletion are handled. Identity stitching should be supportable and governed, not an attempt to force every order into a person.
Calculate the base rate
Count distinct customers with at least one eligible purchase in the denominator. Count distinct customers with more than one eligible purchase under the same rules in the numerator, then calculate numerator divided by denominator times 100.
In the illustrative example, 280 of 1,000 eligible purchasing customers bought more than once, so the rate is 28%. Do not divide repeat orders by all orders, and do not count a customer three times because they placed three orders.
Segment without breaking comparability
After validating the base, segment by acquisition cohort, first product, customer type, geography, channel, offer, or another decision-relevant dimension. Show customer counts, observation time, order value, gross margin, refund behavior, and confidence or variability next to the percentage.
Avoid declaring a universal good rate or optimizing tiny segments. Diagnose whether the difference reflects product replenishment cycles, discounts, seasonality, subscription mechanics, changing mix, or data coverage before attributing it to a campaign.
Choose and measure a retention action
Select one intervention for an eligible cohort, such as onboarding, education, replenishment timing, service follow-up, product recommendation, or loyalty benefit. Define the control or baseline, message eligibility, suppression, cost, margin guardrail, primary outcome, and follow-up window.
Deliverable: business question, cohort and observation contract, eligible-purchase rules, identity method, cleaned customer table, numerator and denominator query, worked calculation, segment table with counts, margin and refund checks, experiment brief, privacy and suppression controls, and dashboard annotation.
THE TAKEAWAY
Define who enters the cohort and how long they can repeat, resolve identity and returns, calculate customer—not order—counts, segment only after the base metric is sound, and choose a retention action tied to the result.OFFICIAL REFERENCES