Brand DNA worksheet for AI writing

Fill this in before the first generated draft

Complete the worksheet in plain language. If a sentence could apply to any consultancy in your category, it is not DNA. The Living Brand Graph can store it, but it will not help a model choose.

1. 01Promise: In one sentence, what does the brand make possible for a specific audience?
2. 02Proof boundary: What outcomes, industries, or methods are you willing to cite, and what must never be implied?
3. 03Audience: Whose job, constraint, and success metric should the writing serve first?
4. 04Point of view: What do you believe that competent peers still debate?
5. 05Voice: Which words you use, which words you refuse, and how direct you are about uncertainty.
6. 06Non-goals: What the brand is not, even if a prompt could make it sound larger.

Turn DNA into writing rules

Rules should be testable. 'Be bold' is not testable. 'Do not claim quantified ROI unless a locked proof node supports the number' is testable. 'Prefer concrete nouns over category poetry' is testable. Put those rules where strategy and creation can both see them.

• Always: mention the audience's job when teaching a method.
• Never: invent customer names, partner logos, or statistics.
• Prefer: short sentences, sourceable examples, and claims a reviewer can check.
• Avoid: borrowed thought-leadership tropes that could have been written for anyone.

Keep DNA alive

When the offer changes, update DNA and lock the new claims after review. When a draft feels off, ask whether the packet was missing DNA or whether the DNA is now outdated. Persistent brand context includes these decisions, not only entities extracted from a website.

If Brand DNA and verified knowledge disagree, stop generating. Resolve the graph first.