Executive Brief 2.0 — AI Shopping Readiness

Management decision

**Recommended decision:** fix now: close the few high-confidence gaps, then use the package as a premium sample or sales proof after Jan approves external use.

**Current score:** 89.22 / B

**Audit tier:** ecommerce

**Target:** `https://www.blivakker.no`

What this means commercially

The sampled site already has strong machine-readable ecommerce evidence. The commercial opportunity is to turn that base into cleaner AI shopping handoff: less crawler friction, clearer owner-authored guidance, and safer exact-product purchase paths.

Top 3 fixes

• **Publish an evidence-backed /llms.txt guidance file:** Create /llms.txt with brand summary, category map, priority collections/products, shipping/returns facts, canonical links, explicit agent policy, token-budget guidance, and freshness/source-of-truth notes.
• **Low content-to-boilerplate ratio:** Move important answers into crawlable text and reduce boilerplate.
• **Unclear H1 structure:** Use one descriptive H1 and hierarchical H2/H3 sections.

Expected lift after fixes

• Cleaner summaries and fewer wrong-page inferences once boilerplate and heading issues are fixed.
• Higher confidence product/category routing once llms.txt and canonical guidance point agents to owner-approved URLs.
• Better conversion handoff once exact-product URLs, cart links, Storefront API, or scoped agent tools are verified.

Agent access posture

• **Recommended posture:** SELECTIVE
• **Confidence:** high
• **Why:** This is an ecommerce target: product/search discovery may be commercially useful, but training access, checkout automation, account pages, and sensitive inventory/pricing surfaces should be controlled separately.

Platform commerce readiness

• **ChatGPT/OpenAI Product Feed Readiness:** `partial` — Prepare a feed-grade product/variant export before claiming ChatGPT Shopping readiness.
• **Perplexity Merchant Program Readiness:** `partial` — Use this as an onboarding/apply checklist; do not claim Perplexity inclusion or ranking without live evidence.
• **Cloudflare Markdown for Agents Readiness:** `not_tested` — If Cloudflare is used, run an approved Accept: text/markdown check and compare product/policy fact preservation.
• **MCP/API Maturity Assessment:** `checklist` — Do not build MCP first for most shops; prove feed/schema and handoff maturity, then design scoped product-search/cart tools with evals.
• **Retest & Before/After Delta:** `checklist` — After fixes, rerun the same audit target and compare score, findings, recommendation cards, and raw evidence snapshots.

Buyer-agent scenario summary

• **14/14** deterministic buyer-agent proxy scenarios passed in the sampled evidence.
• These are readiness proxies, not live ChatGPT/Perplexity/Gemini behavior claims.

Browser journey evidence

• **Status:** pass
• **What it showed:** A Webwright-style browser pass loaded the approved URL, captured visual evidence, and recorded visible text/control evidence across 4 named browser scenario(s).
• This is visual/browser evidence only; it does not claim live answer-engine ranking or visibility.

Competitive benchmark status

Populated competitive benchmark included: Blivakker vs Nordicfeel, Vita, and Lyko across 7 dimensions (see Competitive Benchmark.md).

What we are not claiming

• Not a live ChatGPT/Gemini/Claude/Perplexity ranking or visibility test.
• Not a full-catalog crawl unless explicitly run against sitemap/product URLs.
• Not checkout/payment execution.
• Not an affiliate or commission recommendation.
• Not a public endorsement or commissioned statement unless separately approved.

Next decision

If this is going external, approve the named publication action separately. If this is staying internal, approve competitor URLs/artifacts next so the Premium benchmark can be populated with real comparisons.