How scoring works
Updated
The AI Visibility Check tells you whether AI recommends your store. The Website Audit tells you why. It scores your store against 40 criteria, grouped into the four factors that decide whether an assistant picks a store for a "where to buy" question, and gives you a score per factor so you can see which one is holding you back.
What are the four factors?
| Factor | What it looks at |
|---|---|
| Trust & Authority | What the rest of the internet says about you: reviews, mentions, reputation. Most of this lives off your own website. |
| Discoverability | Whether an assistant can reach and read your product pages at all: access, readability, structured data. |
| Content Quality | Whether your pages actually carry the facts a shopper asks for, in a form a machine can lift. |
| Merchant Competitiveness | Whether your price and shipping stand up against the stores AI recommends alongside you. |
The four are not equal, and the audit weights them by how much each one really moves an assistant's choice. What that means in practice is that two stores with the same total score can need completely different work, which is why the report always shows you the per-factor breakdown rather than a single number.
Why is so much of it about things outside my website?
Because "where should I buy this" is a question about stores, not products. Every reseller carrying this product has roughly the same photos and the same description, straight from the brand. None of that separates you.
What separates you is whether you look like somewhere safe to send a customer, and an assistant mostly learns that from places you do not own: reviews, forums, press, listings. You can write a perfect product page and still lose to a store with a worse page and better standing. That is not a flaw in the scoring; it is what the assistants are doing.
How do 40 criteria become one score?
Each criterion is scored on its own, criteria roll up into their factor, and the factors combine into your total. Nothing is a coin flip: every criterion is either measured from your store, measured from public sources, or excluded.
More useful than the total: each finding is ranked by impact, so the report tells you what to fix first rather than handing you a list of forty things and wishing you luck.
What if a criterion does not apply to my store?
It is taken out of the calculation, not scored zero.
This matters more than it sounds. If your store has no videos, a video-related criterion is not a failure, it is a question that does not apply to you, and scoring it zero would punish you for a choice that is none of the audit's business. So it is removed and the remaining criteria in that group carry the weight between them. The rule throughout: rather not score something than score it wrong.
Does the language I checked in change my score?
Yes, and in your favour rather than against it.
Some trust signals live on platforms that are English-language by nature. If you ran your check in another language, those criteria are skipped, not failed: the audit scores what it can genuinely measure for your market and leaves the rest out of the calculation. You are never marked down for a platform your shoppers do not use. More languages are being added.
Why does the audit need a check first?
Because it reuses what the check found. The audit judges your price and shipping against the actual competitors AI named in your answers, not against a generic benchmark, and it reuses the language you picked. Without a check there is nothing to compare you to, which is why the audit unlocks once your first check has run.
Next step
See what is available now and what is coming: Roadmap.
