Moving fundraising campaigns forward: small, staged wins that compound

What to take away

!image

A staged win is one change to one part of one campaign, measured against the version it replaced. It can be a gift array in Donorbox, a reactivation email, or a shorter form. The size of the lift matters less than the evidence it leaves behind for the next decision.

Start with the gift array, because it compounds fastest

The gift array is the set of suggested amounts on your donation form. It is the cheapest thing to test and the easiest to read. Donorbox, Givebutter, Classy, Bloomerang and EveryAction all let you edit it from the form builder.

Take the middle option, the one most donors pick, and raise it by one step. If your array runs $25, $50, $100, $250, the middle is $50. Run $25, $75, $150, $300 on half your traffic for two weeks and compare average gift.

Net funds are what you keep after processing. Stripe publishes 2.9% plus 30 cents per US card charge, so a $50 gift nets about $48.25. Its ACH rate is 0.8% capped at $5, which nets about $49.60 on the same gift. Platform fees sit on top.

A higher average gift lifts every channel that sends traffic to that form. The same email list and the same ad spend produce more net funds per donor. Test the array before you test the email, because the email only decides who arrives.

The next change on the same form is a monthly option at the top of the array. Givebutter charges no platform fee and asks donors for tips. Donorbox publishes a platform fee under 2% on its Standard plan, before processing.

Monthly donors stay on file longer than one-time givers. The M+R Benchmarks study, produced each year with NTEN, tracks monthly and one-time giving separately, so you can see whether the change held.

Reactivate lapsed donors in small steps

A donor who gave once and stopped is not a stranger. They already said yes once, and the file holds the date they said it. Bloomerang, Little Green Light and Neon CRM all let you filter on last gift date.

Send them a single ask for a specific, small gift tied to one funded need. Not the annual appeal. Not a newsletter with a donate button at the bottom. One ask, one amount, one deadline.

Track reactivation rate by segment: lapsed under twelve months, twelve to twenty-four months, and beyond. The first group responds at several times the rate of the third. If your list has not been segmented this way, that split is the first staged win available to you.

Time the send to a real deadline. GivingTuesday falls on December 1 in 2026, the Tuesday after Thanksgiving. December carries the heaviest share of US giving, and the final days of the year take a large slice of it.

Plug in your own numbers: 2,000 lapsed donors at a 3% response and a $40 average gift returns $2,400. The same message to an unsegmented list of 20,000 buries that result in the noise.

Test the donation page, one element at a time

NextAfter, a nonprofit research lab, publishes free donation-page experiments. VWO and Optimizely run split tests and quote pricing on request. Google Optimize closed on September 30, 2023, so it is not an option.

A form that asks for a phone number will convert worse than one that does not, and the drop is usually visible within a few hundred visits. Run the shorter form against the longer one and keep the version that raises completed gifts per visit.

Free tools cover most of what a small shop needs. Google Analytics 4 counts completed gifts, and Microsoft Clarity records sessions and heatmaps at no cost. Stripe can add Apple Pay and Google Pay to a hosted page, which removes the card fields and the typing.

Do not stack three changes into one test. If the result moves, you will not know which change moved it, and you will have learned nothing you can reuse.

Keep the ask honest, and keep the proof

The FTC advertising substantiation policy states that objective advertising claims need a reasonable basis before you publish them. That standard covers the promise on your donation page as well as a product ad.

If the appeal says a gift feeds a family for a week, the program team must show the arithmetic. If the photo shows a named beneficiary, you need written consent for that use, and a plan for what happens if they withdraw it.

Written consent should name the use, the channel and the time limit, and it should say who to call to withdraw it.

The CISA software acquisition fact sheet is written for buyers of software, not for fundraisers. Its one transferable point is that you check a vendor's practice, exposure, and vulnerability handling before you sign. Apply that to your CRM and email platform, not to your appeal copy.

Bloomerang, Classy, EveryAction and Mailchimp all publish privacy and security pages, and the answer to a data question should arrive before the renewal, not after.

Run the test like an experiment

The NIST experimental design guidance starts design choice with the objective and the practical constraints. In fundraising terms: decide what you are trying to learn, then decide what you can actually run.

Split your audience into a test group and a control group. Give them the same campaign, the same timing, the same list hygiene. Change one thing. Hold the rest still.

Plan the sample before you send. A few thousand recipients per cell is a working floor for an email test; below that, only a large difference shows up at all.

Keep records of what you sent, to whom, on what date, and what came back. The CISA logging guidance explains why organizations retain and review event records. Your version is a simple log of sends, changes, and results, kept where the next person can find it.

Name one owner with the authority to stop the test, and write the stop rule next to the start date.

A worked example: the spring appeal

A mid-size charity ran its spring appeal to 40,000 email addresses. The first send used the standard array and raised an average gift of $62.

For the second send, the team split the list. Half received the standard array. Half received an array with the middle option raised by one step, plus a reactivation line aimed at donors who had lapsed in the past year.

The test group's average gift came in higher, and the reactivation line pulled back donors who had not given in eighteen months. The team kept the higher array for the fall campaign and made the reactivation line its own test next time.

That is the shape of a staged win: one change, one measurement, one decision that carries forward.

Common questions

How long should a fundraising test run?

Long enough to collect a meaningful number of gifts rather than clicks. Two weeks is a common floor for email, but a small list may need a full month. Stop when the result is clear, not when the calendar says so.

What should I measure besides net funds?

Donor quality: repeat giving rate, average gift over twelve months, and whether new donors from this campaign give again. A campaign that raises money once and loses the donor is not a win.

What if the test shows no difference?

That is a result. Keep the simpler version, record what you tested, and move to the next element. A null result saves you from repeating a change that does nothing.

What does a starter stack cost?

Mailchimp is free up to 500 contacts, with paid plans from around $13 a month. Givebutter charges no platform fee and asks donors for tips. Donorbox publishes a platform fee under 2%. Stripe card processing runs 2.9% plus 30 cents.

CRM costs sit higher. Bloomerang publishes entry plans in the low hundreds of dollars a month, and Classy quotes on request.

Who should own the test?

One named person, with the authority to stop it.

If the owner is a committee, the test will run past its stop rule and the result will be unreadable. For the wider set of campaign questions this work raises, see common nonprofit fundraising campaign questions, and for the campaign planning that sits upstream of any test, see nonprofit fundraising campaigns.