The AI visibility score is a number from 0 to 100 that says how ready a product is to be found, understood and recommended by AI assistants. It adds up 12 checks in two halves: Listing quality, what you write, and Agent readiness, what Shopify’s AI catalog returns for the product.
What are the two halves of the score?
- Listing quality asks: once found, how well does the listing answer a shopper? It covers the description, the title, the category, the category fields and the variant attributes. The description and title are judged by a language model; the rest is read straight from your Shopify data.
- Agent readiness asks: can an AI assistant find this product and trust what it reads? It covers whether the Shopify catalog returns the product at all, and whether that copy carries the core fields, priced variants, policy links, correct availability, media and reviews. These checks are live: the catalog is asked on every scan.
The catalog is the one AI agents read through Shopify’s agentic commerce tools. That is why the second half matters as much as the first: a perfect listing the catalog cannot see is still invisible to an assistant.
What does each check look for?
A product’s Score breakdown shows every check with what it found, and how to earn the missing points.
Listing quality
| Check | How it is measured | How to pass it |
|---|---|---|
| Description | Judged by the AI rater: long enough, says who the product is for and how it is used, and carries concrete facts an assistant can quote. | Write for a shopper who is comparing: audience, use cases, materials, dimensions, what is in the box. |
| Title | Judged by the AI rater: names the product type and one distinguishing attribute, without all caps or keyword stuffing. | Merino wool crew sweater, navy, rather than SWEATER or a string of keywords. |
| Category match | Whether the product’s category fits what the text describes. A product with no category does not pass. | Assign the most specific category that fits. |
| Category fields | The share of the category’s attributes (the fields Shopify adds for that category) that have a value. | Fill colour, material, size, pattern and whatever else the category defines. |
| Variant attribute completeness | Whether variant option values are specific enough to match a shopper’s request. | Rename Option 1, Style A or Default Title to what they are: Navy, 42, Left handed. |
Agent readiness
| Check | How it is measured | How to pass it |
|---|---|---|
| Catalog eligibility | Whether the Shopify catalog returns the product at all when searched for in your store. | Keep the product active and the online store public, then give Shopify time to sync. |
| Core fields present | Whether the catalog’s copy carries a title, a description, a price, media and variants. | Fill in whatever is missing on the product in Shopify. |
| Variant data structure | Whether every variant the catalog returns has a price. | Give every variant a price, including the ones you rarely sell. |
| Policy clarity | How many of the four policy pages Shopify exposes to agents are published: refund, shipping, privacy and terms of service. | Publish the missing pages under Settings → Policies in Shopify. |
| Availability hygiene | Whether what the catalog says about availability agrees with your admin. | Usually a sync delay after a stock change; scan again after a while. |
| Media quality | At least three images, all with alt text, passes in full. Fewer images or missing alt text passes in part. | Add images and write alt text that says what is in each one. |
| Review signal | The review count the catalog reports for the product. | Collect reviews, and make sure your review app passes them to Shopify. |
What do the rating bands mean?
| Rating | What it means |
|---|---|
| Strong | Well optimized. |
| Good | Solid visibility. |
| Weak | Missing key signals. |
| Critical | Needs attention. |
The Products page colours each score by its rating and lets you filter by it. The thresholds are the same for every store.
Which model scores, and why does it never change?
The two text checks are judged by one language model, the same for every store, chosen by Octolift. A score is only comparable if one rater produced it: models disagree with each other by a wide margin on the same listings, so a rater that changed from store to store would make scores meaningless.
A score is only produced when the rater has judged both text checks. If it does not answer, you see We couldn’t score this product and can try again, rather than a half verdict.
When is a product scored?
- When you analyze it. The analysis page opens on the product’s last analysis, with its date, for as long as the product is unchanged. Press Reanalyze for a fresh scan. Text verdicts are reused while the title, type, category and description have not changed; the catalog checks run live every time.
- After you save a change. The product is analyzed again on its own. The catalog may take a while to pick up the change, so the Agent readiness half can follow a little later.
- Never in the background. Octolift does not crawl your catalog on a schedule. Only the first scan of a product counts against your plan; every later one is included.
What do Not discoverable and Catalog unreachable mean?
Not discoverable means the catalog did not return the product when Octolift searched for it in your store, so every Agent readiness check fails. Common reasons: the storefront is password protected, the product is new or just changed and has not synced, it is not active, or it is not available in the checked market.
Catalog unreachable means the catalog could not be reached at all. The scan is marked incomplete and not saved, so a temporary outage never overwrites a real score. Press Try again in a moment.
To raise a score, the fastest route is Improve with AI.