What documented case files reveal about nonprofit content output

What to take away

The table below lists the six fields to score a case on.

What a documented case file contains

A case file is the working paper behind a published result. It has a brief that names the audience and the ask, two or three draft versions, the sign-off record with names and dates, and the response data from the send.

Public files show what that looks like. Wikimedia posts each English Wikipedia fundraising banner on Meta-Wiki with the exact wording, the country it ran in and the amount it raised. charity: water publishes an annual report with a project cost table.

charity: water also states that 100 percent of public donations fund water work, with private donors covering overhead. ProPublica attaches a method note to its data stories; its 2016 Machine Bias analysis named Broward County, Florida, and covered more than 7,000 defendants.

The Trevor Project's youth mental health survey publishes its sample size and reports more than 28,000 respondents. Doctors Without Borders names the countries and figures behind its annual Top Ten crises list.

The output those files describe is narrower than a case study suggests. Wikimedia rewrites banner copy for each test. charity: water anchors a year of content to one September push. ProPublica ships a method note with every data story. The cadence is annual or quarterly, not weekly.

Most published case studies show the last of those four items and skip the first three. That is why they are hard to reuse. You can see that a food bank raised 40 percent more in its spring appeal, but not what the letter said, who approved it, or how many people it went to.

Ask for the artifact. A screenshot of the donation page, the subject line as sent, the segment definition, the approval thread. If a vendor cannot produce one, you are reading a summary of a summary.

The six fields to score a case on

Score every case on the same fields so two cases can be compared. The table above is the audit table a marketing lead keeps in the shared drive.

A case that fails two fields is a lead. A case that fails four is marketing.

Run the table above over the Wikimedia file. The brief is the banner copy, the artifact is live on Meta-Wiki, the method is the per-country test, the result is the amount raised, and the publisher is the Foundation itself.

Five documented files worth opening

These are public and checkable. Read them for structure, not for numbers you can copy.

M+R Benchmarks (mradvocacy.org, published annually with NTEN) reports email list growth, response rates and giving totals across hundreds of nonprofits, split by size and sector. The 2023 report states: 'The median email response rate was 0.10 percent,' based on emails delivered.

It is a benchmark, not a case, but it gives you the denominator most case studies omit. Use it to sanity-check any single-organization result before you repeat it.

NTEN's case study library (nten.org) holds member-submitted write-ups on CRM migrations, email program rebuilds and digital campaign launches. The index page is titled "Case Study Library." Each entry opens with a line that names the organization and the vendor, so the commercial tie is visible.

The Nonprofit Technology Conference session archive (nten.org/events) records practitioners walking through a campaign with the actual screens. The archive page is titled "NTC Session Archive." It sits closer to a case file than a written study, because you watch the person open the tool.

Wikimedia fundraising reports (meta.wikimedia.org) post banner copy, test results and per-country amounts for the English Wikipedia campaigns. One banner line reads: "Please don't scroll past this. 98% of our readers don't give; they simply look the other way." Those banners run to logged-out readers, usually in December. Each campaign has a named owner and a dated launch.

ProPublica method notes (propublica.org) attach to every data story and name the dataset, the cleaning steps and the limits. The 2016 Machine Bias note states: "We obtained data from the Broward County Sheriff's Office on more than 7,000 defendants."

For Canadian receipting, disbursement quota and consent questions raised by any of these, name the authority and stop: the CRA sets receipting rules. A compliant receipt must show the charity registration number, the donation date and amount, and the statement that it is an official receipt for income tax purposes.

A qualified charity lawyer reads the rules against your situation.

The four-row evidence audit

The file you build when someone proposes copying a published result has four rows.

The gap row is the one teams skip. Write down what the case leaves out before you write the proposal. If the gap is the audience size, you cannot estimate lift. If the gap is the comparison group, you cannot claim the change caused the result.

Run the local test on one segment before you change the whole program. A 2,000-name segment tells you whether the idea survives contact with your list.

Point the same four rows at the Wikimedia file and the gap row fills fast. The banners test wording, not your list, so the audience size and the comparison group are missing from the public record.

Where the standards come in

Two external checks belong in the file, and neither one approves anything.

The FTC advertising substantiation policy holds that objective advertising claims need a reasonable basis before they run. If your appeal copy promises a specific outcome, that promise needs evidence behind it. The policy is a U.S. standard; check your own regulator.

The W3C guidance on information and relationships covers whether visual structure is also available to assistive technology. A donation form built as a table of divs fails it. Test the live page, not the mockup.

Two more belong in the purchase record when you buy tools or research. The CISA software acquisition fact sheet lists supply-chain and deployment questions to ask a vendor. The NIST guidance on experiment design starts design choice with the objective and the practical constraint, which is the right frame for a small list test.

Common questions

What separates a case file from a case study?

A case file holds the brief, the drafts, the sign-off record and the raw response data. A case study is the published summary, usually written by the vendor and usually showing only the result.

How many cases should a team review before acting?

Three, scored on the same six fields. One case is an anecdote, two is a pattern, three lets you see which field keeps failing. If the same field fails in all three, the problem is the field, not the cases.

What if the case has no comparison group?

Then you have a description, not a result. Record it as a lead and run your own test on one segment. For donor acquisition benchmarks to compare against, see our donor acquisition case studies.

Where do I find more of these files?

Our nonprofit content marketing questions page covers sourcing, and the content marketing improvement guide walks through the sign-off record. For the wider program, start with nonprofit content marketing.