---

name: brand-iq description: Brand IQ analysis METHOD — what evidence to read from the store, how to shape a voice / audience / style-guide proposal from it, and what never to invent. Read this ONLY when task-brand-iq-setup is working a node that needs it. It has no entry of its own and must never be entered directly — a merchant asking to set up Brand IQ enters that task, which then reads this. tags: [internal, marketing] license: MIT-0

Brand IQ setup mode

task-brand-iq-setup owns the sequence; this skill owns the method. When the merchant asks for the store to be analysed and Brand IQ set up, that task's node graph decides the order and you report each node as you enter it. Everything here is how to do the work inside those nodes.

Brand IQ is your memory of the merchant's brand — Brand Voice, Audiences, and Style Guide — applied to every piece of content she later generates. This skill covers auto-setup: reading the real store and proposing drafts the merchant confirms, rather than making them fill three forms by hand.

Auto-analysis loop (the core flow)

When the merchant asks to set up or auto-analyze their Brand IQ:

1. Call ANALYZE_STORE_BRAND. It is READ-ONLY and returns brand evidence (product naming, category mix, price band, vendors, image-style hints) — it does NOT write anything and does NOT decide the brand voice for you.
2. Open with ONE short sentence naming your read of the brand (e.g. "Bold, design-led snowboard brand spanning entry-level to premium"). Do NOT restate the evidence back — product counts, vendors and price band are already known, and repeating them just wastes time. Then go straight to three TIGHT drafts:
   • one Brand Voice — a name (e.g. "Warm & confident") plus ONE short guideline paragraph (≤ 80 words) on tone and do's / don'ts;
   • Audiences — as many as the evidence honestly supports (typically 2–3), each a name plus a one-line profile that includes a demographic or lifestyle cue (age range, values, or buying motivation) AND what they respond to (e.g. "Eco-conscious millennials, 25–35 — value sustainability, respond to transparency and provenance");
   • one Style Guide — a name plus the concrete terminology / grammar / punctuation rules the evidence supports (typically 4–5) — the essentials only, never a filler rule to reach a count. Ground every draft in the real evidence — actual categories, price positioning, vocabulary that recurs across DESCRIPTIONS and store copy. A product's NAME is catalog data, not voice evidence: "Aqua Bomb" names what the store sells, never how the brand speaks, so a name's wording must not become a tone, a guideline or a style rule (measured W1 2026-10: a voice proposal built its register off one product name, and the merchant read the result as not their brand). Never generic filler, never padding. Be concise: the merchant is waiting on this, and shorter drafts generate faster.
3. Call SHOW_BRAND_ANALYSIS_PROPOSAL with FLAT arguments (no nested objects): voice_name, voice_content, style_guide_name, style_guide_content, and audiences as a list of plain strings, each in the form "Name: description" (e.g. "Eco-conscious millennials: value sustainability, respond to transparency"). It PERSISTS NOTHING — it only surfaces a review card. STOP after calling it and wait for the user.
4. Do NOT save anything yourself — no tool of yours can. The user reviews, edits, and approves on the card, and the frontend writes each approved draft back to Brand IQ. Asked in chat to apply or save it, point them to the card's approve button: their click is what saves, and nothing is saved until they press it.

If the store has no products, ANALYZE_STORE_BRAND returns an error — tell the user to add products or set up Brand IQ manually; never fabricate a brand.

Re-analyze

• "Re-analyze" / "re-scan my store" runs the same loop from step 1 and produces a fresh proposal. Treat any existing Brand IQ entries as context to improve on, not something to silently overwrite — the user confirms the new proposal before anything changes.
• (Reserved) Merge-vs-replace semantics against existing entries — to be specified with the frontend Re-analyze flow.

Reading the current Brand IQ — BRAND_IQ_GET

• Your SOUL's Brand Voice block is a provision-time snapshot; the merchant may have edited Brand IQ since. Whenever current brand settings matter — drafting content, choosing tone, picking keywords or target markets, or improving on existing entries during a re-analyze — call BRAND_IQ_GET and treat its answer as the truth.
• Pass fields to fetch only what you need: brand_voice, audiences, style_guides, product_keywords, target_markets, primary_category. Omit it for all six.
• A null / empty value means the merchant has not configured that item — that is an ANSWER (e.g. propose setting it up), not an error.
• If the tool ERRORS, the settings could not be loaded. Do NOT fall back to guessing or to the SOUL snapshot as if nothing happened — say the settings could not be loaded.
• In the re-analyze loop, fetch brand_voice / audiences / style_guides first — the existing entries are the context you are improving on.

Using the primary category (reserved)

• (Reserved) Once a Brand IQ primary category is set, use it to bias SEO keyword selection and channel-strategy recommendations. Detailed guidance lands here alongside the SEO / category work.

Boundaries

• ANALYZE_STORE_BRAND and SHOW_BRAND_ANALYSIS_PROPOSAL have no side effects; they never touch the live store or the merchant's saved Brand IQ.
• Persisting a brand voice / audience / style guide is the frontend's job after the user approves — do not assume anything is saved just because you proposed it.