Operations
Questions nonprofit content teams ask before greenlighting a campaign
The questions nonprofit content teams ask before greenlighting a campaign, and what a defensible answer looks like for each stage of planning.
What to take away
The questions nonprofit content teams ask before greenlighting a campaign arrive in a fixed order, and one weak answer can hold the launch date.
- Greenlight a campaign only when you can name the program evidence behind it and the period that evidence covers.
- Ask who the story is about, what they consented to, and whether the claim survives outside that one case.
- Set the success measure before launchtask completion, qualified support, or correction quality, not impressions.
- Write down the exception that would change your answer, and who owns the correction if the claim is wrong.
- Treat legal, consent and receipting questions as referrals to the CRA or a charity lawyer, not as content decisions.
The questions content teams actually ask
Seven questions stall campaigns in review. Each one has a defensible answer, and it belongs in the brief rather than in a meeting recap.
Campaign greenlight questions
- Which program evidence supports this claim?
- One participant's story or a pattern?
- What did the person agree to in writing?
- What is the ask, and does it match?
- Who signs off, and by when?
- What would make us pull it?
- How will we know it worked?
Questions content teams ask
- Which program evidence supports this claim, and how old is it?An eighteen-month-old output report describes a program that may no longer run that way.
- Is this one participant's story or a pattern?A single interview is a story. A pattern needs a count, a period, and a definition of who was counted.
- What did the person agree to, in writing?Consent to be photographed is not consent to be quoted, and neither is consent to appear in a direct mail appeal.
- What is the ask, and does it match the story?A story about housing stability that ends in a general fund gift asks the donor to make a leap the story did not make.
- Who signs off, and by when?Campaigns stall at sign-off more often than at drafting. Name the approver and the date on the brief.
- What would make us pull it?Decide the correction trigger before launch, not after a donor emails to complain.
- How will we know it worked?Pick one primary measure and one guardrail measure. More than that and nobody reads the report.
Turning one story into a broad impact claim
The failure that costs the most is invisible until someone checks. A team interviews one graduate of a training program, writes that she now earns a living wage, and the headline becomes "we help people out of poverty."
Sizing the claim to the evidence
Defensible claim
- Scope
- One of 40
- Period
- 2025 cohort
- Count
- Count available
- Comparison
- None needed
- Wording
- One of 40 participants
Unsupported claim
- Scope
- All participants
- Period
- No period
- Count
- No count
- Comparison
- No comparison group
- Wording
- We help people out of poverty
The story is true. The claim is not supported by it. A reader who checks finds no count, no comparison group and no period, and the credibility loss spreads to every other number on the site.
The fix is a sentence, not a study. Write the claim at the size the evidence supports: "one of 40 participants in the 2025 cohort" is defensible. "Participants in our program earn a living wage" needs a follow-up count you can point to.
What to measure instead of impressions
What to measure instead
| Measure | What it tells you | Common trap |
|---|---|---|
| Task completion | Whether the reader did the thing the page asked | Counting a click as completion |
| Qualified support | Whether the gift or signup matched the campaign intent | Counting all gifts, including general fund |
| Consent match | Whether the release names the channel and the words the campaign will use | Treating a photo release as consent to quote |
| Correction quality | How fast and how visibly an error was fixed | Logging the fix without a date or owner |
| Cost per qualified action | Actual spend per qualified action, including staff time and design | Excluding staff time and design |
Cost per qualified action is the figure that survives a budget meeting. In most US markets it runs from a few dollars per online action to $25 or more per mail response, driven by list size and postage.
What to measure instead of impressions
Measure
- Task completion
- Reader did the thing
- Qualified support
- Gift matched intent
- Correction quality
- How fast errors fixed
- Cost per qualified action
- What campaign cost
What it tells you
- Task completion
- Click counted as completion
- Qualified support
- All gifts counted
- Correction quality
- Fix logged without date
- Cost per qualified action
- Staff time excluded
Common trap
- Task completion
- Qualified support
- Correction quality
- Cost per qualified action
Impressions belong in a footnote. They move for reasons unrelated to your content, including a platform's own algorithm changes.
Where the source limits bite
The NIST AI RMF Playbook sets out voluntary actions under govern, map, measure and manage. Cite it when a tool changes how content gets drafted or reviewed. It is not a certification and it does not rank vendors, so do not present it as either.
The CISA software acquisition fact sheet covers supply-chain exposure and vulnerability management for buyers. Add those checks to a procurement record. They do not approve a content decision.
When a vendor promises reach or engagement, the FTC guidance on soliciting and paying for online reviews is the reference for fake feedback, conditioned incentives and undisclosed relationships. Read it before you name a provider in a case study.
If you plan to compare two versions of a page or an appeal, the NIST guidance on experimental design selection starts with the objective and the practical constraints. That is the difference between reporting what happened and estimating what caused it.
A worked example
A food bank wants to greenlight a spring appeal built on one client's story.
The program evidence is a 2025 distribution count, current. The story is one household, and the client signed a photo release that does not mention direct mail. The ask is unrestricted. The approver is the communications director, and the brief has no date.
Two changes make it defensible. The team adds the distribution count beside the story so the claim is sized correctly, and the client signs a release that names the appeal channel.
The measure is qualified support: gifts coded to the spring appeal, compared against the same appeal last year. The exception that would stop it: if the household's circumstances changed and the story no longer reflects the program, the piece comes down, and the correction owner is named on the brief.
Common questions
Who should answer these questions?
The content lead writes the brief, but the program team supplies the evidence and the approver signs it. A campaign that only the communications team has seen is a campaign nobody can defend.
Most of the questions nonprofit staff keep raising about email campaigns come back to this split between who drafts and who owns the evidence.
What if the program data does not exist yet?
Say so in the brief and set a date to collect it, or run the campaign without the numeric claim. A story told as one story is honest. A story told as a pattern without a count is not.
This is where staging real gains: how nonprofit content output actually improves becomes a scheduling problem rather than a writing one.
How do consent and receipting rules affect this?
They vary by province and by state, and the authority that sets them is the CRA, the IRS for cross-border gifts, or your provincial privacy regulator. Confirm the specifics with a qualified charity lawyer before a campaign goes out.
Teams that treat these as notes on nonprofit content marketing from the field tend to catch them earlier, before a release is signed rather than after the appeal mails.
