Donor retention rate
The sector's headline health metric, and the one most distorted by the duplicate records nobody has time to merge.
Also called Supporter retention, Repeat giving rate.
What it is
The share of donors who gave in one year and gave again in the next. It is the closest single measure of whether supporters feel like their giving mattered, which is why a fall in it is treated as a crisis and why so much depends on it being computed correctly.
donors who gave in both years ÷ donors who gave in the first year
- donors
- Resolved people or households, not database records
- gave
- A hard credit, counted once, with soft credits attributed to their origin
The part the formula leaves out
Retention measured on records is a measurement of your data entry. A supporter giving every year under three records looks like three supporters who each gave once and left. In the worked example, resolving people before counting moves retention from 47% to 58% and removes 312 names from a win-back appeal they should never have been on.
The second distinction is why someone stopped. A cancelled monthly gift and a declined card look identical in a totals report and require opposite responses. In the example, 214 of 300 lapsed monthly donors stopped within three days of a payment failure, 63 of them in a single processor batch, and none of it had anything to do with the rebrand it was blamed on.
Households or individuals is a genuine fork rather than an error. Two people at one address giving separately are two donors and one household, and the answer changes the number. It moves the example by 1.8 points and merges two of the top eleven donors into one, so both should be shown rather than one being chosen quietly.
How it is usually computed wrongly
01
Counting records instead of people
Every duplicate turns one retained donor into two lapsed ones, and the error compounds every year the file gets messier. A falling retention rate is frequently a file getting worse rather than supporters leaving.
Deduplicate on email, and on name with address and giving pattern, before any counting. Report the number of records that collapsed alongside the rate.
02
Counting a soft credit and a hard credit as two gifts
A matched gift or a donor-advised-fund distribution appears twice by design. Counting both overstated the worked example's year by $84,000, which is the kind of error that survives for years because it makes the chart look right.
Credit the originating donor once and treat the soft credit as attribution rather than income.
03
Treating payment failures as lapses
They are recoverable and the response is a phone call, not a re-acquisition campaign. Rolling them into the lapsed count both understates retention and misdirects the budget.
Separate involuntary from voluntary lapse by looking at what happened in the three days before the gap started.
04
Comparing your rate to a published benchmark
Published rates are computed on different denominators, different definitions of a donor and different treatment of monthly giving. A number computed a different way is not a comparison.
Compare to your own prior years on one definition, stated. If you must use a benchmark, use its definition rather than yours.
What your file needs
- Gift transactions with a date, an amount and a donor reference
- Donor records with enough fields to resolve identity: email, name, address
- At least two full years, and ideally three
Anything missing is reported as unavailable rather than substituted with something weaker computed on worse evidence.
Compute it on your own file
No account needed to start. You only pay when you like what you see.