River
Y CombinatorBacked by Y Combinator
FREE TEMPLATE

Nonprofit Annual Appeal Letter Template

Four documents and three sheets that derive every printed ask amount from each donor's own giving history rather than one uniform string.

Free download  ·  No account needed

Every appeal guide gives you the same reply slip: three amounts, printed on every piece. Then the appeal gets reported on one blended response rate. Both hide the same thing, which is that the file is not one audience. A donor who has given three years running sits in it next to somebody who has never given, and one number over both of them is not a summary. It is an average of two different answers, and it is the number every board report shows.

So the ask amount here is derived per record, and the base is the donor's highest previous contribution rather than their last gift. That distinction only bites where giving fell, which is the group closest to lapsing and the group a last-gift rule asks least of. Then the slip gets audited backwards: count the gifts that arrived at exactly a printed amount, keep the donors whose own previous best was higher, and sum the differences. That total is what the old slip cost, measured rather than modelled.

In the worked example an 8,400-piece appeal reports 7.57% response and $74,513 net. Inside it, 120 donors gave $7,700 against a previous best of $13,550, so the slip left $5,850 uncollected, and two segments covering 53.5% of the volume are net negative. The gifts that come back still need the acknowledgment their own facts require, and the restricted ones carry conditions grant reporting will draw on. Every outcome figure in the case for support traces to an indicator you can actually collect, or the sentence gets cut rather than estimated.

What the blended response rate could not show

Every segment netted against its own cost, every ask derived from one donor's history, and the gap the uniform slip left behind.

Segment Definitions

Illustrative, for a fictional land trust called Sable Point Conservancy. FY2025 annual appeal, 8,400 records, one ask string of $100 / $50 / $25 on every reply slip. Cost per piece $1.18, which is print, personalisation, slip, business reply envelope, mail house handling and postage. Every column is a sum over the same file and the bottom row is what the board was shown.

SegmentPiecesCostGiftsRateRevenueAvgNetCost per $
Gave in each of the last three years612$72220733.82%$45,950$222+$45,2280.016
Gave in each of the last two years448$5299420.98%$13,025$139+$12,4960.041
First gift last year, nothing before1,105$1,30419617.74%$13,250$68+$11,9460.098
Lapsed one year934$1,102667.07%$6,400$97+$5,2980.172
Lapsed two years806$951323.97%$3,200$100+$2,2490.297
Lapsed three or more years1,690$1,994201.18%$1,550$78-$4441.287
Never given2,805$3,310210.75%$1,050$50-$2,2603.152
Whole file, as reported8,400$9,9126367.57%$84,425$132.74+$74,5130.117

The 612-record segment at the top is 7.3% of the volume and 54.4% of the revenue. The two negative segments are 4,495 pieces, which is 53.5% of the mailing, and they netted minus $2,704 between them. A cost per dollar raised of 0.117 is under one, so the appeal reads as profitable and is, across half its volume, not. The useful move is not suppression, it is division: the long-lapsed segment produced 20 donors for a $444 loss, so reactivation cost $22.21 each, while the never-given segment produced 21 for $2,260, so acquisition cost $107.61 each against a $50 first gift. Same sign, same line in the report, 4.8 times the price, and only one of them is obviously worth paying.

Ask Amount by Giving History

One row per record, with the string a last-gift rule would have printed kept alongside the derived one. The base is the highest previous contribution. Multipliers are the conventional 1x, 1.5x and 2x, rounded to a figure somebody would write, floored at $25.

Giving historyLastBestOff the last giftOff the previous bestGaveGap
1000, 1000, 500$500$1,000500 / 750 / 10001000 / 1500 / 2000$100$900
500, 500, 500$500$500500 / 750 / 1000500 / 750 / 1000$100$400
250, 500, 1000, 500, 250$250$1,000250 / 375 / 5001000 / 1500 / 2000$1,000none
500, 250, 150, 100$100$500100 / 150 / 200500 / 750 / 1000$0none
500, 1000, 500, 250$250$1,000250 / 375 / 5001000 / 1500 / 2000$0none
100, 75, 50, 25$25$10025 / 40 / 50100 / 150 / 200$0none
150, 250, 250, 150, 150$150$250150 / 225 / 300250 / 375 / 500$100$150
100, 150, 250$250$250250 / 375 / 500250 / 375 / 500$100$150
250 (one gift)$250$250250 / 375 / 500250 / 375 / 500$100$150
50, 50, 75, 50, 75$75$7575 / 110 / 15075 / 110 / 150$75none
no historynonenoneno base exists50 / 100 / 250$0none
150, 150, ?, 150$150$150flagged150 / 225 / 300$0none

Row three is the reason to keep both columns. That donor ignored the slip entirely and wrote in $1,000, which is exactly their previous best and four times what a last-gift rule would have led with. Rows one, two, seven, eight and nine did the opposite: they took a number off the slip that was below their own record, and the top amount printed was $100. Across the file the two bases can only disagree on records whose giving fell, because a rising donor's last gift is their best gift by definition. There are 1,009 such records out of 3,714 with more than one gift, and for those the median lead amount is $200 off the previous best against $100 off the last gift.

Response Tracking by Segment

Built before the mail drops rather than after. Each touch carries its own code and the code stays on the gift when it is recorded, so next year's curve can be read per touch instead of collapsing into one blended figure.

The capping gap, by segmentCappedGavePrevious bestGapMean
Gave in each of the last three years45$3,150$6,225+$3,075$68
Gave in each of the last two years18$1,100$1,650+$550$31
First gift last year, nothing before35$2,150$3,525+$1,375$39
Lapsed one year17$1,000$1,700+$700$41
Lapsed two years3$200$300+$100$33
Lapsed three or more years2$100$150+$50$25
Total120$7,700$13,550+$5,850$49
The curve the sequence is timed offHalf in byNine tenths in byTouchDay
Gave in each of the last three yearsday 7day 11Warm reminder16
Gave in each of the last two yearsday 6day 13Warm reminder16
First gift last year, nothing beforeday 8day 15Warm reminder16
Lapsed one yearday 8day 21Lapsed reminder27
Lapsed two years and beyondtoo thin to readtoo thin to readLapsed reminder27
Whole fileday 8day 15Peak day 6, last gift day 43
What suppression does to the headlineWhole fileLosing segments out
Pieces8,4003,905
Cost$9,912$4,608
Gifts636595
Revenue$84,425$81,825
Net$74,513$77,217
Response rate7.57%15.24%

The response rate doubles without a word of the letter changing, because the appeal did not improve, the list did. That is the whole case for coding the touches: a year-over-year comparison on a blended rate compares list decisions rather than appeals, and it is the figure that reaches the board. The curve does the same job for timing. Warm donors were nine tenths in by day 11 and lapsed donors by day 22, so the reminders go out eleven days apart rather than on one date somebody picked, and on or after day 15 there were still 61 gifts worth $6,325 to come, which is what a reminder has to catch to pay for itself.

What is in the pack

01

Ask Amount by Giving History

One derived string per record, with the string a last-gift rule would have printed kept in the next column so the base can be argued about.

02

The base is the previous best, not the last gift

The two only disagree where giving fell, which is the group closest to lapsing and the group a last-gift rule asks least of.

03

Segment Definitions, netted

Pieces, cost, gifts, revenue and net per segment, so a mail plan is a decision about numbers rather than about a blended rate.

04

The capping gap, measured

Gifts that came in at exactly a printed amount, from donors whose own previous best was higher. Measured history, not a projected lift.

05

Seven letters, one per behaviour

No dollar figure in any ask sentence, because the slip is personalised and the letter is not. The never-given letter thanks nobody.

06

A sequence timed off the curve

Warm and lapsed reminders on separate days, derived from when each group actually decided, with a conditional fourth touch gated on its own cost.

How it works

  1. 1

    Send the giving history

    One row per gift, plus the real cost per piece, which becomes fundraising expense at year end. If last year's printed ask amounts were never recorded, that absence is the first finding.

  2. 2

    Cut on behaviour, then net

    Recency, frequency and level, every record in one segment, counts reconciled. A donor holding two records lands in two segments and neither one is right. Then pieces, cost, revenue and net for each against its own cost.

  3. 3

    Derive the string per record

    Base at the highest previous contribution, conventional multipliers, and the year of that best gift carried in its own column. A gift missing from that history lowers the base silently.

  4. 4

    Audit last year's slip backwards

    Count the gifts at exactly a printed amount whose donor had already given more, and sum the differences before this year's string is signed off.

Frequently asked questions

Why not multiply the last gift, like every other guide says?

Because for a donor whose giving is flat or rising, the last gift and the best gift are the same number, so the rules agree and nothing is at stake. They can only diverge where giving fell. In the worked example that is 1,009 of 3,714 multi-gift records, and it is the group closest to lapsing.

What is the capping gap, exactly?

Take the gifts that came in at exactly an amount printed on the slip. Keep the donors whose own highest previous contribution was higher than that amount. Sum the differences. Here it is 120 donors who gave $7,700 against a previous best of $13,550, so $5,850, which is 6.9% of revenue.

Is that $5,850 what we would raise with a better slip?

No, and the pack says so in three places. It is measured history: what a specific, countable set of donors gave against what each had already given at least once. Nobody has been asked yet. The per-donor version, $49, is the honest figure to plan against.

Two segments lose money. Do we stop mailing them?

Price them first. The long-lapsed segment lost $444 and produced 20 donors, so $22.21 each. The never-given segment lost $2,260 and produced 21, so $107.61 each against a $50 first gift. Same sign, 4.8 times the price, and the renewal rate in your own file decides the second one.

We enclose address labels. Does that change the letter?

Not for labels. Low-cost articles sent free to people who did not order them are treated as insubstantial, so the payment stays fully deductible. Anything more valuable makes it a quid pro quo contribution, and the disclosure has to accompany the solicitation itself, not just the receipt afterwards.

Can we share the mailing with a partner organisation?

Only if both are separately authorised. An authorised organisation may mail only its own matter at nonprofit prices and cannot lend that authorisation, and a cooperative mailing carried for an unauthorised party pays regular prices. At 8,400 pieces that is a real line in the $9,912.

Our response rate rose last year. Was the appeal better?

Check what changed in the list before crediting the letter. Removing the two negative segments here takes the blended rate from 7.57% to 15.24% with no copy change at all, and that figure is what reaches the board packet. Segment-level tracking is what makes the comparison mean something, and donor retention analysis cuts the same file by gift sequence.

Find out what last year's reply slip cost you

Send the giving history and last year's result. The first thing back is the count of donors who gave a printed amount below their own previous best.

Segment my file