Agikod methodology
How Agikod analyses a Shopify store
Agikod measures how well AI agents can read, understand and compare the public information of a Shopify store. This is exactly how the diagnosis is produced.
Method in force since 7 October 2026.
Public information, observed as an AI agent sees it
The analysis uses only publicly accessible store pages: home, catalogue, priority products, identity, contact, shipping, returns, terms, FAQ and officially linked help centres.
The crawler respects robots.txt rules, reading limits and safeguards against unsafe addresses or redirects. It never accesses a private area of the store.
26 criteria, worth a total of 100 points
- 9 common criteria: identity, activity, category, contact, served areas, structured data, AI-agent accessibility and external presence.
- 17 commerce criteria: catalogue, products, prices, availability, variants, images, reviews, shipping, returns, payment, terms, warranty, FAQ, copy quality and target market.
- A criterion that cannot be measured is never counted as a failure: it leaves the denominator. An insufficiently measurable analysis remains partial and is not proposed for indexing.
What is deterministic and what uses AI
Most checks are deterministic: whether information exists, price consistency, availability, structured markup, sitemap, robots.txt, commercial policies or product attributes.
AI is used only for three editorial assessments: clarity of the activity, product-copy quality and value proposition. Its answers must be tied to excerpts actually found on the analysed pages; unverifiable quotations are rejected.
Score caps prevent a misleading high score
- A store blocked to major AI agents or unreadable without JavaScript cannot score above 30/100.
- A store with no readable price on the sampled products cannot score above 50/100.
- A store with no verifiable legal identity and no accessible terms cannot score above 60/100.
What the public profile shows
The public profile shows the score, measured axes, three priorities and up to three observed products. It displays the analysis date and remains a limited projection of the same diagnosis used in the dashboard.
Detailed evidence, excerpts, full recommendations, corrections and monitoring data remain in the owner’s private space.
Publishing and quality control
A profile may be accessible from the link sent to the store without being proposed to search engines. Only complete, sufficiently supported and technically stable Shopify V2 profiles pass the Agikod indexing gate.
Partial analyses, legacy profiles, fictitious templates and claim pages remain outside the company sitemap.
What the score does not guarantee
The Agikod score measures the quality and readability of the observed information. It does not measure a position in ChatGPT or any other third-party service.
Agikod does not guarantee indexing, citation, recommendation or sales. The diagnosis identifies conditions the store can improve to be more easily understood and compared by automated systems.