AI discoverability

The two layers of discoverability

Brand-level discoverability is whether the engine knows the brand exists, what it sells, and where its canonical domain lives. Product-level discoverability is whether the engine can resolve a specific SKU — by name, variant, or brand-plus-model — to the correct PDP. They are tested separately, because a brand can be fully discoverable while half its catalog is not.

Both layers depend on entity cleanliness (consistent naming across the site's metadata, Wikidata, and third-party sources) and on crawler access — an engine cannot discover a page it is not allowed to fetch. Site structure, llms.txt, sitemaps, and robots rules all feed the signal; see OpenAI's crawler documentation for one engine's access rules.

Why it matters for ecommerce

Discoverability sits upstream of every other AI-search metric, which makes it the first thing to verify and the easiest to forget. Optimization spend on copy, schema, and reviews returns nothing if the underlying page is blocked to AI crawlers or unresolvable as an entity. In eCommerce Insights's scans, blocked AI bots and orphaned product URLs are among the most common silent failures on otherwise well-run stores.

Fast-growing catalogs feel this most: a freshly launched variant can take weeks to appear in AI answers simply because engines have not indexed the canonical URL or connected it to the brand entity yet. Launch checklists that include discoverability checks — sitemap entry, internal links from collection pages, structured data shipped on day one — shorten that lag.

A discoverability gap: an example

A dog-treats brand launches a freeze-dried beef liver SKU. Two weeks later, Perplexity finds the PDP when asked directly; ChatGPT responds that it has no information about the product. The gap is engine-specific (illustrative example). The team confirms the URL is in the sitemap, links the product from the brand's collection page to strengthen internal signals, and updates the brand's Wikidata entry. ChatGPT resolves the product correctly within two weeks. No copy was rewritten — the page was fine; it just could not be found.

How it relates to neighboring terms

Discoverability enables AI visibility, which GEO work then maximizes. Its mechanical inputs overlap with the agent-readability score — crawler admittance and entity clarity are checks in both — and llms.txt is its most direct lever: a curated reading list that points crawlers at the pages that matter.

How eCommerce Insights checks it

Per SKU: crawler admittance per bot, canonical URL reachability, entity resolution (does each engine connect the product name to the right PDP), and structured-data presence. Failures surface as specific fixes — a robots line, a sitemap entry, a naming inconsistency — before any optimization recommendations are made.

Related terms

• AI visibility — the downstream outcome discoverability enables.
• llms.txt — the curated reading-list signal for LLM crawlers.
• Agent-readability score — shares the crawler and entity checks.
• Product schema — the structured identity an engine resolves against.
• Hallucination detection — the failure mode when engines half-know a product.