Strategy
Part of Donor retention: the parts worth your attention
Fixing donor retention mistakes: what teams repeat every quarter
Recurring donor retention mistakes usually trace back to source data, gift definitions or approval gaps in the CRM, not to the ask or campaign itself.
What to take away
- Most donor retention mistakes repeat each quarter because the defect sits in the source record, the gift definition or an approval step, not on the slide.
- The same five defects come backsplit donor records, frequency flags set on gift count, rolling-window retention, receipts and thank-you notes sent from two systems, and reactivation mail sent to active donors.
- Fix the record, then reissue every asset built from it, including receipts and thank-you letters.
- A correction is not closed until the same defect has been shown to be unable to recur under normal work.
- Keep the old value. A donor who gave twice in one month is not a monthly donor, and the audit trail has to prove it.
The five mistakes that come back every quarter
1. Duplicate donor records. Two entries for the same person split one giving history in two, and both halves look lapsed. Matching gift records from Benevity or Double the Donation and event registrations through Classy or Eventbrite are the usual sources. A merge rule keyed on email plus postal code, run before any retention count is published, closes most of them.
Five recurring donor retention mistakes
- Duplicate donor records split one giving history
- Wrong gift frequency flag tags one-off gifts recurring
- Rolling twelve months hides the quarter of churn
- Thank-you and receipt sent from different systems
- Reactivation sent to people who never lapsed
2. Wrong gift frequency flag. A donor who gave monthly for three months gets tagged recurring, then churns in the fourth. The flag should follow the payment schedule, not the count of gifts in the period. Stripe and PayPal send each charge as its own transaction, so the CRM infers the schedule from dates. Recheck the flag whenever a retention figure moves.
3. Retention measured on a rolling twelve months. A rolling window hides the quarter a donor actually stopped. Use fixed cohorts: everyone whose first gift landed in a given month, tracked forward. The donor retention figures then compare like with like. Build the cohort from the gift table rather than a dashboard tile, because dashboard defaults rarely show the month a donor stopped.
4. Thank-you and receipt sent from different systems. The email says one amount, the receipt another, and the donor calls to ask which is right. One system owns the number. The other reads from it. IRS Publication 1771 requires a written acknowledgment for gifts of $250 or more. Quid pro quo contributions above $75 need a disclosure statement, so a mismatch can be a compliance problem.
5. Lapsed-donor reactivation sent to people who never lapsed. This is the most expensive one. It reaches active donors with a "we miss you" message, and some of them believe it. Suppress anyone with a gift inside your lapse window, usually 12 or 24 months, and check the count before the send goes out.
Trace the defect, not the symptom
A retention number that drops is an output. The defect is upstream: a source field, a definition, or an approval that let a wrong value through.
Work backward first. Which record changed, when, and by what import or manual edit? Then work forward. Every report, receipt, appeal segment and board slide built on that record carries the error.
Preserve the prior version before you correct anything. The NIST experimental design selection guidance starts design choice with the objective and the practical constraint, which is the right frame for deciding whether you are describing a problem or testing a fix.
| Field | What to record | Acceptance test |
|---|---|---|
| Duplicate merge | Source of the second record | One history, one contact |
| Frequency flag | Payment schedule and gift dates | Flag matches the schedule |
| Cohort window | First gift month and end date | Same window across reports |
| Receipt match | System that owns the amount | Email and receipt agree |
| Reactivation segment | Date of each donor's last gift | No active donor in the send |
Before you scale the fix
- Use a fixed cohort, not a rolling window.
- Track renewal and reason-coded attrition together, with cost and quality.
- Walk one normal donor path and one failure path.
- Keep claims, inputs, versions and decisions.
- Write down what you still do not know.
- Set a stop rule and a review date.
Entry pricing for small-shop systems such as Little Green Light and DonorPerfect typically falls between $40 and $150 a month for one user, as of 2026. Blackbaud Raiser's Edge NXT and Salesforce Nonprofit Cloud are quoted per organization, so ask what an export costs before you sign.
The GOV.UK technology selection guidance asks about data control, security review and ownership cost, which is a fair checklist before you move donor data between systems.
The GOV.UK open standards guidance links open standards to interoperability and reduced supplier dependence. Treat it as a portability prompt: can you export the giving history and take it elsewhere?
A worked correction
A mid-size charity found its monthly giving count 14% high. The cause was a gift-frequency flag set on gift count rather than payment schedule. Three donors had given three one-off gifts in a quarter.
The team corrected the flag, reissued eleven receipts, rebuilt the monthly donor segment, and reran the reactivation send to the corrected list. They then added a check that compares the flag to the payment schedule before any retention report is published.
Common questions
Why does the same retention mistake return each quarter?
Because the fix was applied to the output, not the input. Rebuilding a report takes minutes; changing the import rule or the approval step is what stops the error coming back.
How do we know a correction is finished?
When the corrected path has been run once under normal conditions and the defect did not reappear. Until then the issue is open, whatever the dashboard says.
What should we do with the old, wrong figure?
Keep it, dated, next to the corrected one. Boards ask why a number moved, and the answer has to be in the file. The FTC advertising substantiation policy is the standard to check any public claim against before it goes out.
