How do I build a don't-say list and check every draft against it?

A don't-say list is a short, written list of phrases the brand must not use, each with the reason and a safe alternative. Build it from the claims your team is tempted to make, not from a generic list of banned words, and check every draft against it before approval.

The entries that matter most are the almost-true ones: the claim that was true last year, true for one customer, or true only with a condition attached.

Most brands already know the obvious prohibitions. The list earns its keep on the subtler ones, the claims a well-meaning writer makes because they sound like the brand's position.

• Superlatives you cannot prove: the best, the first, the only, number one.
• Guarantees: guaranteed results, risk-free, never fails.
• Claims with a hidden condition: free when there is a paid tier, instant when it takes a day, unlimited when there is a limit.
• Out-of-date facts: last year's customer count, a feature that has changed, a price that has moved.
• Anything regulated in your sector: health, financial, environmental or legal claims your compliance team has a view on.
• Words the brand simply does not use, for voice rather than risk.

An entry that only says what not to write gets ignored or argued with. Give each one four parts, so a writer who hits it knows what to do next.

1. 1The phrase, and its variants. Write the phrase and the forms it takes. "Free" also appears as "for free" and "free of charge". List them under one entry so one rule is one finding.
2. 2Why it is off limits. One sentence. "There is a paid tier, so free alone overstates it." The reason is what lets a writer judge a case the list did not foresee.
3. 3What to say instead. The approved alternative, word for word: "free to start", "no card required". Writers use an alternative far more readily than they obey a ban.
4. 4Whether negation is allowed. Some phrases are fine when denied. "We never sell your data" uses the words "sell your data" and is the brand's position. Mark which entries may appear in a negated sentence, so the check does not flag your own reassurances.

Keep the list in a shared document or a spreadsheet tab with four columns: phrase and variants, why, say instead, allowed when negated. Keep it short. A list of two hundred entries is one people learn to skim; twenty that are genuinely tempting is more useful.

To check a draft, search the caption, the first comment and the text on every slide or frame for each phrase. In a spreadsheet, a formula using SEARCH across the caption cell will flag matches; for artwork, someone has to read it. When a phrase matches, read the sentence around it before deciding: if the entry allows negation and the sentence denies it, it passes.

Pair the list with a claims log: every factual claim the campaign makes, where it came from, and whether it has been verified. The don't-say list catches the wrong words; the log catches the right words with no evidence behind them. Figures in the log also feed the ledger of spent figures.

Here is what a finished entry looks like, for an imaginary software company. Yours will be specific to your product, which is the point.

• "Free", "for free", "free of charge". Why: there is a paid tier, so free on its own overstates it. Say instead: "free to start". Allowed when negated: no.
• "Guaranteed", "guarantee". Why: we cannot promise results for a customer's own campaign. Say instead: "designed to" or describe what the feature does. Allowed when negated: yes, as in "we never guarantee results".
• "Trusted by thousands". Why: we have no count we can stand behind. Say instead: name a specific customer who has agreed to be named, or say nothing. Allowed when negated: no.

• Review it when the product, pricing or legal position changes, not on a calendar.
• Add an entry every time a reviewer or a client catches a claim in a draft. That catch is the list telling you what it is missing.
• Remove entries that no longer apply. A stale rule teaches people the list is not worth reading.
• Keep one owner. A list everyone can edit and nobody owns drifts within a quarter.

Your don't-say list is part of the brand brief: onboarding asks what you must never claim. Every draft is then checked against it, so a claim that is almost true is caught before a reader catches it. The same check covers spent hashtags, repeated figures, platform limits and unbacked claims.

The linter is not metered on any plan, and it runs on one-off posts as well as campaign posts. See the features page for the full set of checks, or how to run client approval for the sign-off that follows.