The Case for Support: Drafting a Campaign Statement Without Losing Your Voice

The Internal Case, the External Case, and What Comes From Them

Campaign counsel distinguishes between two versions, and organizations that skip the first pay for it later. The internal case is written for the board, the leadership team, and the staff who will defend this campaign in public. It is long, unpolished, and comprehensive, stating the argument in full including the awkward parts: why the current facility is inadequate in specific terms, what happens if the campaign underperforms, why this program and not the one the board has debated for a year. As the Moran Company notes in its overview of the role of the case statement in fundraising, the internal case forces leadership to put the fundamental reasons for the appeal in writing, and the process matters at least as much as the artifact.

That process is where disagreements surface. A board that has never written an internal case frequently discovers, during the writing, that three members hold materially different theories of what the campaign is for. Surfacing that in a drafting session is uncomfortable and cheap. Surfacing it eighteen months later, when two board members have described the campaign differently to the same prospect, is expensive. This is the step organizations skip most often, precisely because it produces nothing they can print.

The external case is shaped for a reader who does not work for you: shorter, sequenced for interest rather than completeness, written in the language of the person being asked. Everything in it should trace back to something the internal document established. From it a family of derivatives follows, worth naming because organizations tend to commission them separately and end up with five documents that disagree. The leadership brief is two or three pages a board member can absorb and speak from. The one-pager is the leave-behind, often the only thing a prospect actually reads, so it must carry the problem, the solution, the number, and the ask at a glance. The brochure is the designed artifact for lead prospects, and the website page has to work for someone arriving from a search result with no context. The verbal ask is the hardest derivative, because the person delivering it will improvise, and improvisation stays on message only when the argument is genuinely internalized. That piece connects to how you build major gift proposals, since a tailored proposal is the case narrowed to one person's interests.

Drafting the derivatives is a low-risk use of AI, because the source material is already yours. Give a model the approved external case and ask for the one-pager, the web version at two lengths, the board talking points, and a ninety second script. Voice drift is far less likely when a model compresses your words than when it generates new ones.

Why Most Cases Fail

Weak cases fail in a small number of predictable ways, and every one is a failure AI will reproduce and amplify unless you watch for it. Knowing the patterns turns a model from a copy generator into a diagnostic tool, because you can ask it to hunt for these specific problems in a draft, including a draft it wrote.

The first is being organization-centered rather than donor-centered. Count the sentences whose subject is your organization; in a weak case it runs well above half. The reader is repeatedly told what the organization has achieved, believes, and requires, positioned as an audience for institutional self-description rather than a participant. The correction is not to remove the organization but to restructure, so the community's situation and the donor's role occupy the frame and the organization appears as the mechanism connecting them.

The second is abstraction where specificity belongs. Weak cases describe categories: underserved populations, systemic barriers, wraparound services, transformative outcomes. Strong cases describe instances: a wait time, a named neighborhood, a particular thing that happens on a Tuesday afternoon in your building. Abstraction feels safe because it cannot be contradicted, which is exactly why it moves no one. The third is hedged institutional language, common where a case passes through four approvers. Programs seek to address. Initiatives aim to foster. Each hedge removes a little meaning and collectively they remove all of it, and hedging usually means nobody wants to own the claim.

The fourth and fifth travel together. If you cannot name what makes this year different from next year, you do not have urgency, and asserting it anyway costs credibility on every other claim. Likewise a goal of twelve million dollars is an abstraction to almost every reader, while the same twelve million split into a building, an endowment, and three years of bridge funding, each described by what it produces, is legible. Donors are increasingly literate about outcome claims, and a number offered without the context to evaluate it reads as evasion rather than evidence.

A newer failure belongs on this list: the case with no discernible author. Grammatically clean, structurally correct, entirely without a point of view, because it was generated rather than written and nobody added one back. That failure is the subject of the rest of this article, and it is the one most likely to affect organizations otherwise doing everything right.

Keeping Your Voice on the Page

Voice is not a decoration applied at the end. It is the accumulated effect of hundreds of small choices about rhythm, vocabulary, how much you concede, whether you use contractions, and what you find worth saying. A model asked to write in a warm, authentic voice produces what it has learned warmth looks like, a recognizable performance of warmth rather than yours. Adjectives do not transmit voice. Examples do.

So build a voice guide from your own best writing. Gather eight or ten pieces that genuinely sound like you: the appeal that overperformed, the executive director's best letter, the program update people quoted back to you. Ask a model to describe the voice in operational terms rather than adjectives. Average sentence length. Whether contractions appear. How the writer handles bad news. Which words recur. What it never does. The result is a working voice specification, more useful than anything a branding exercise produces because it comes from evidence rather than aspiration, and keeping it current is the discipline covered in our piece on brand voice consistency. Then put two or three full pieces of your own writing in the context alongside it, because models imitate demonstrated patterns far more reliably than they follow described ones.

Draft section by section. Whole-document generation is where voice goes to die, because the model establishes a register in the first paragraphs and maintains it mechanically for eight pages, producing a uniformity no human writer would produce. Working one section at a time lets you intervene while the piece is malleable and preserves the natural variation a real document has.

Use AI more for structure and diagnosis than for finished prose. This is the shift separating organizations getting real value from those producing generic material. The highest-value uses are analytical: find every sentence where the organization is the subject, list the claims the attached data does not support, identify the weakest paragraph and explain why, rewrite this section three ways so I can see the options, tell me what a skeptical reader would ask here. Those uses improve your writing. Asking for a finished case replaces it.

Then read it aloud. The oldest editing technique available remains the most effective detector of machine-assisted prose, because generated text is optimized for how it looks rather than how it sounds. Anything you stumble over, or feel embarrassed saying out loud, is a place to rewrite, and since much of a case will eventually be spoken by board members, this doubles as a check on the derivative you care most about. Finally, the last pass belongs to a human. Not a review, a pass. Someone who knows the organization rewrites into the document the detail only an insider knows, the sentence that takes a position, the specificity no model could have produced because the fact lives in somebody's memory rather than in any file you uploaded.

Testing the Case on Three Kinds of Reader

Internal review tells you almost nothing, because everyone reviewing the case already knows the answers to the questions it is supposed to answer. They read past the gaps by filling them from memory. The only reliable test is reading it to people whose knowledge differs from yours, and three readers catch three different classes of failure. Read it aloud rather than emailing it, because you learn as much from where they shift in their seat as from what they say afterward.

A board member catches the ownership failure. You are not testing whether they approve, since they will. You are testing whether they can say it back. Read them the case, wait a day, then ask them to describe the campaign as if you were a prospect. Wherever they hesitate, paraphrase into something vaguer, or default to old talking points, the case has failed to give them usable language. That is a document problem, not a board member problem, and it usually means the argument is correct but not memorable.

A lapsed donor catches the credibility failure, and this is the most uncomfortable and most valuable of the three. Someone who gave before and stopped has a specific, informed skepticism no internal reviewer can simulate. They will know whether the urgency is real, because they remember the last time you called something urgent. They will notice if the vision resembles one you described five years ago without noting what became of it. Ask them directly what makes them doubt it, and be prepared to hear about your last campaign rather than this one.

A stranger catches the assumption failure. Find someone who knows nothing about your organization or your field, ideally with no particular sympathy for the cause. They will stop you at the jargon you no longer hear yourself using. They will not understand the acronym. They will ask a basic question about why the problem exists, which you will discover the document never answers because everyone in your office already knows. They will also tell you when they stopped caring. Strangers are the only readers who can find the hole you cannot see, because your expertise is what conceals it.

A model supports this without replacing it: ask it to read the draft as a skeptical donor and list objections, or to name the sentence where a reader is most likely to disengage. What it cannot do is notice the pause before someone answers a question, and that pause is frequently the most informative data you will get.

After the Draft: Using the Case and Keeping It Current

A finished case is a beginning, not a deliverable. The most common waste in campaign communications is an excellent case sitting in a folder while staff and volunteers keep improvising from memory. Produce the derivatives at once, from the approved source, so they agree. Then train the people who will speak from them. Board members do not need the full document, they need three sentences they can say naturally and a clear sense of what to do when asked something they cannot answer. Rehearsal beats briefing, and a model makes a useful rehearsal partner: have it play a skeptical prospect and let a volunteer practice until the argument stops sounding memorized. Organizations running a broader annual program will find the same source material feeds the annual fund, provided someone maintains the link rather than letting the two sets of messaging drift apart.

Then keep it alive, because a campaign runs for years and a case written in month one describes a world that no longer exists by month thirty. Costs change. Phases complete. A lead gift lands and the remaining need shifts. A case still citing the original figures after two years quietly signals that nobody is paying attention, and prospects in the later public phase are reading a document written for the quiet phase, for a different audience, at a different moment.

Treat the case as a versioned document with a named owner and a quarterly review during an active campaign. Update the facts: numbers, progress, completed phases, new commitments. Check the argument: does the urgency mechanism still hold, does the vision still match the plan, has anything you promised failed to happen in a way the document should acknowledge. And check the voice, because updates accumulate and each one is a chance for someone in a hurry to paste in a generated paragraph that does not sound like the rest. AI handles the first part efficiently, comparing the current case against updated financials and listing every claim now out of date. The second and third parts need the human attention the original draft did.