AI commerce readiness, measured
AI commerce readiness is how well an ecommerce store performs when the shopper is an AI agent rather than a person. Prefero scans your store the way ChatGPT, Gemini, Perplexity and browser agents do — crawl access, product structured data, policy signals, and head-to-head preference — and returns one score plus the specific fixes behind it.
AI commerce readiness is the degree to which an ecommerce store's catalogue, crawl access and policy data are machine-readable enough for an AI shopping agent to discover, compare and confidently recommend its products.
What AI commerce readiness actually measures
Traditional ecommerce metrics assume a human is looking at the page. AI commerce readiness assumes nobody is. An agent asked to "find a merino base layer under $120 that ships to Berlin in three days" never renders your hero image, never reads your brand story, and never scrolls. It fetches, parses, compares and answers.
That reduces the whole storefront to four questions, and readiness is the weighted answer to all four.
- Access
- Can the agent reach your product pages at all? robots.txt rules for AI user-agents, a sitemap that actually contains product URLs, and optional intent files like llms.txt and agents.md.
- Structured data
- Can it parse what it finds? Product JSON-LD with price, currency, availability, brand, identifiers and return policy — not just a name and an image.
- Policy signals
- Can it answer the objection? Shipping windows, return terms and warranty stated in a form a machine can quote, rather than buried in a PDF or a prose page.
- Preference
- Given a real alternative, does the agent pick you? Prefero simulates buyer personas head-to-head against a competitor URL you supply.
How the score is weighted
The readiness score is a weighted average of three category scores, each 0–100. Preference carries the same weight as structured data because being parseable and being chosen are different problems — plenty of stores are perfectly readable and still lose every comparison.
When no competitor URL is supplied, the preference category resolves to a neutral 50 rather than guessing, so an uncontested score is never inflated.
| Category | Weight | What moves it |
|---|---|---|
| Structured data | 40% | Required Product fields present; recommended fields (brand, GTIN, rating, return policy) present |
| Crawlability | 20% | AI user-agents not blocked, sitemap resolves to product URLs, llms.txt and agents.md present |
| AI preference | 40% | Share of simulated buyer personas that choose your product over a named competitor |
What a normal score looks like
We scanned 120 live ecommerce stores in May 2026 — a mix of well-known DTC brands and mid-market retailers — to establish a baseline. The distribution is wide and the median is not flattering.
The pattern behind those numbers is consistent: readiness tracks platform defaults far more than merchant effort. Shopify stores inherited crawl-access files and product markup they never configured; custom builds and older platforms mostly did not.
| Signal | Stores | Share |
|---|---|---|
| Machine-readable Product JSON-LD | 43 / 120 | 36% |
| — of which on Shopify | 37 / 63 | 59% |
| — of which on custom or other platforms | 6 / 57 | 11% |
| llms.txt served | 59 / 120 | 49% |
| agents.md served | 55 / 120 | 46% |
| Sitemap resolves | 98 / 120 | 82% |
| Sitemap contained zero product URLs | 26 / 120 | 22% |
Why this is worth measuring now
Agentic checkout is still early and still uneven — early implementations have been walked back, conversion data is thin, and nobody should rebuild a storefront on the assumption that agents will dominate next quarter.
But the readiness work is not speculative. Every fix that makes a store legible to an agent — complete Product schema, a resolvable sitemap, machine-readable return terms — is the same work that already improves rich results, feed quality and comparison-site accuracy. The downside is bounded; the upside is being present in a channel while it is still cheap to enter.
Frequently asked questions
- What is AI commerce readiness?
- AI commerce readiness is the degree to which an ecommerce store's catalogue, crawl access and policy data are machine-readable enough for an AI shopping agent to discover, compare and confidently recommend its products. It is measured across crawl access, structured data, policy signals and head-to-head preference.
- How is it different from SEO?
- SEO optimises for ranking a page a human will then read. AI commerce readiness optimises for an agent that never opens the page — it consumes the structured data, decides, and reports an answer. A store can rank well in classic search and still be unusable to an agent because its product data is rendered client-side or missing entirely.
- How long does a scan take?
- About 30 seconds. Prefero crawls robots.txt, the sitemap, a sample of product pages and the policy pages in parallel, then runs the buyer simulation. No signup, no install and no tag on your site are required.
- Does a low score mean my store is broken?
- No. A low score means AI buyers have less information to work with than your competitors give them. Most stores in our 120-store sample scored between 48 and 73 — the median store is legible to humans and partly illegible to machines.
- What do I get beyond the score?
- The free scan returns the score, the category breakdown and the top issues. The full report adds the ranked recommendation list, the persona-by-persona simulation transcript, the schema audit field by field, and a patch kit with copy-pasteable JSON-LD, robots.txt and llms.txt blocks.
Get your AI commerce readiness score
Paste your store URL. Add a competitor if you want the head-to-head. Free, no account, about 30 seconds.