Retention rate in marketing: cohorts and campaign results

Retention rate in marketing measures how many customers acquired through a campaign return at a defined age. Compare the same customer event over the same elapsed time, with the original cohort fixed. A higher combined rate can come from a different campaign mix even when neither campaign improves.

Define the customer event before comparing campaigns

For a US office-supply distributor, a useful question is whether new paying accounts place another qualifying order after their initial buying period. A campaign that produces many first orders but few subsequent orders may need a different budget or onboarding approach.

This is an acquisition-cohort question. The general distinction between renewal, retention and churn rates belongs to their separate definitions: a contract coming due is a different denominator from customers originally acquired.

Write the measurement rule alongside the report:

• Entry: the customer's first qualifying paid invoice, established from the available purchase history.
• Membership: acquisition month and a single approved campaign assignment, frozen at entry.
• Return: at least one qualifying repeat purchase from day 61 through day 90 after entry.
• Completion: transaction records cover the customer's entire return window.

Day zero is the first-purchase date. This example assumes the distributor normally expects repeat orders within this window; another business should choose a window suited to its purchasing cycle. Returns are netted against the original sale before deciding whether a purchase qualifies.

The distinction between cohort membership and a later return condition also appears in Google Analytics' cohort documentation. Here the unit is a business customer in the invoice export, so website users and devices are not substituted for customer accounts. Covirage is not affiliated with Google.

Assemble customer rows, then count them once

Bring a customer export with customer ID, first-purchase date and acquisition campaign, plus a transaction export with customer ID, transaction date and net eligible purchase amount. Keep an acquisition record even when that customer never buys again.

One customer with several orders in the return window contributes one retained customer. Invoice counts would reward frequent ordering and could push a supposed customer percentage above its possible maximum.

The acquisition campaign must have an agreed business rule. If several campaigns touched an account, resolve the primary assignment before counting; do not copy the customer into every campaign. Changing the assignment after seeing later orders changes the question.

For spreadsheet construction, use the cohort retention table guide. The task here is interpreting mature campaign cohorts, rather than rebuilding that table.

A worked comparison with the same customer age

The distributor reports through May 31, 2026. Every January and February account below has completed day 90. Counts come from distinct customer IDs under the rules above; money does not enter this customer-share calculation.

Campaign cohort retention = Customers with a qualifying return in the window / Customers originally acquired in that campaign cohort

January webinar retention is 64 / 80 = 80.0%. January's combined rate is (64 + 72) / (80 + 120) = 68.0%; February's is (96 + 48) / (120 + 80) = 72.0%. Percentages are shown to one decimal place.

The count checks are exact: January's 136 retained plus 64 not retained equals 200 acquired; February's 144 plus 56 also equals 200. Campaign starting counts must match the distinct new-customer list, and retained counts must trace back to those same IDs.

These labels mean no qualifying purchase in the chosen window. They do not establish that the remaining accounts have permanently left the business.

Separate campaign performance from campaign mix

The combined rate rose by 72.0% minus 68.0% = 4.0 percentage points. Yet webinar retention stayed at 80.0%, and paid search stayed at 60.0%. More February customers came through the campaign with the higher observed return rate.

Use January's customer mix as a comparison weight:

At the original mix, February would also show 68.0%: 40.0% times 80.0% plus 60.0% times 60.0%. The combined improvement is entirely a mix change in this example. Standardizing makes the comparison clearer; it does not prove either campaign caused the difference between its customers.

Marketing owns the acquisition assignment and campaign interpretation. Customer operations owns the repeat-purchase definition and follow-up. Finance checks whether the retained customers produce enough contribution to justify acquisition spending; customer acquisition cost supplies that separate cost measure.

Choose the next investigation

Investigate whether webinar accounts differ in size, initial product or purchasing need before reallocating spending. Use retention analysis to compare those segments at the same tenure. If paid search attracts a different customer group, a campaign-level comparison may hide a useful segment within it.

If the decision requires evidence of incremental impact, design a suitable controlled test rather than treating observed differences as lift. NIST's guidance on completely randomized designs describes random assignment of experimental units to factor levels, the principle behind that distinction.

Where marketing retention goes wrong

• Unequal maturity. A cohort with an unfinished return window is unavailable, not zero. Show only completed customer windows when comparing rates.
• A changing entry event. Trial registrations in one campaign and paying customers in another do not share a denominator.
• A moving attribution rule. Reassigning customers to a later campaign makes acquisition performance impossible to compare consistently.
• An ignored mix change. The combined rate can improve while every campaign's own rate is unchanged, as the tables demonstrate.

Compare your own campaign cohorts in Covirage

Bring your customer, campaign and invoice exports; Covirage maps the columns, and its tools compute cohort counts, retention percentages and reconciliation checks. The external AI model explains the results with every figure traced to its rows. Create a free Covirage account and bring a sample of your own data.

See renewal watch for the supporting customer-retention use case.