Brand Intelligence

When to use this

Use this first. Every other skill in the system assumes a diagnosis exists.

Use it when the brief names a solution instead of a problem, when stakeholders disagree about what is wrong, when previous brand work underperformed without explanation, or when you are about to spend money on expression before anyone has established what needs to change.

Do not use it to produce a market research report. This skill exists to narrow, not to survey. If the output is longer than the brief and no more decided, it has failed.

Inputs

Gather before starting. Ask the user directly for anything missing rather than assuming it.

Required: what the business sells and how it makes money; the growth constraint leadership believes it faces; who the priority customer is; what triggered this engagement now.

Valuable where available: customer interviews or support transcripts, sales call notes, win/loss records, analytics on the paths customers actually take, existing brand materials, prior research, and the language customers use unprompted.

Where a previous cycle exists, the post-launch learning agenda from release-report.md is evidence — what the last brand change was expected to do, and what actually happened. It is usually the strongest evidence available and it is almost always ignored. Ask for it.

If almost none of this exists, say so plainly and mark the diagnosis as assumption-led. Do not compensate with confident prose.

Mission

Find the problem that is worth solving, and establish that branding can materially influence it.

Most brand engagements fail here, not in execution. They solve a real problem that branding cannot move — a distribution gap, an unclear product, a pricing objection, a sales process — by making it more attractive. Your first obligation is to say when that is happening.

Method

1. Establish business reality. Model, margin, growth constraint, sales motion, what leadership is accountable for this year. Brand decisions that ignore the commercial engine get overruled.
2. Map audience progress. For each priority audience: their situation, the progress they are trying to make, the barrier, the alternative they use today, and what triggers a decision. Situations and barriers, not demographics and personas.
3. Trace the decision journey. Where the brand is actually encountered, in what order, under what conditions, and what question the person is holding at each point.
4. Read the category. Its promises, visual and verbal codes, proof conventions, clichés, and comparison frames. Name what everyone says so you can find what nobody can say.
5. Inventory existing equity. What the brand already owns in memory — assets, associations, behaviors, trust, distinctive cues — and the liabilities alongside them. Equity is expensive to build and easy to discard by accident.
6. Build the perception gap. Current understanding versus required understanding, per audience and per dimension, with the evidence and consequence for each.
7. Write causal hypotheses. For each candidate problem: the mechanism, the supporting evidence, at least one rival explanation, and the observation that would separate them.
8. Rank by consequence and confidence. Not by how interesting the finding is.
9. State one strategic problem. The smallest problem whose solution would matter most.

What must be established

Every conclusion carries an evidence class — observed, reported, inferred, assumed, unknown — and a note on which decision it affects. Conclusions that affect no decision are cut, however interesting.

Rank evidence by consequence before reliability. Direct behavior, unprompted customer language, product usage, and commercial data outrank internal opinion, including the founder's. Where they conflict, the conflict is the finding.

Ask what would have to be true for the proposed brand intervention to work, and check whether anyone has established it.

Strong versus merely thorough

The diagnosis is working when it makes some previously plausible options obviously wrong.

Failure modes

• Treating demographics as insight. Age and income are not a barrier to progress.
• Treating competitor visuals as strategy. What they look like is not what they mean.
• Treating founder adjectives as audience truth.
• Promoting an interesting observation to an opportunity with no causal link.
• Recommending a rebrand because the identity looks dated. Dated is not a business problem unless you can name what it costs.
• Diagnosing a brand problem when the evidence points at product, price, distribution, or sales. Say that instead. It is more valuable and harder to say.
• Producing a research scrapbook. Volume is not diligence.

Never claim without evidence

Never invent a customer quotation, need, or behavior. If you are reasoning about what customers probably think, label it inferred and say what would confirm it.

Never state a market size, share, growth rate, benchmark, or competitor result you have not sourced. Never attribute a number to a study you cannot name.

Output

Produce intelligence-brief.md using the structure in references/intelligence-brief.md.

Verification

• For every major conclusion, name its evidence class and the decision it affects. Cut what affects nothing.
• Attack your top opportunity with the best rival explanation you can construct. If it survives only because you stopped looking, it is not the top opportunity.
• Confirm the final problem is narrower and more consequential than the brief you started with. If it is broader, you surveyed instead of diagnosing.
• Counterfactual pass. For each recommendation: if it were removed, what meaningful outcome becomes less likely? No credible answer means cut it, downgrade it, or reframe it.
• Fresh-eyes pass. Read the output as someone who must act on it and was not in the process. Cut process debris, duplicated findings, unearned certainty, and any sentence that impresses without changing a decision.

Complete when

Leadership can state what is true, what is uncertain, what matters most, why it matters commercially or behaviorally, and what the next strategic decision has to be.

Handoff

Brand Strategy consumes the strategic problem, audience progress map, evidence ledger, equity inventory, and constraints. Anything Strategy will need to defend a choice must be in the brief with its evidence class intact — a conclusion that arrives downstream without its class attached will be treated as fact by someone who was not in the room.