Every merchant I work with can tell me where they rank on Google for their main keywords. Almost none of them can tell me where they stand when a shopper asks ChatGPT or Shop for a product recommendation. That is a strange gap, because the second question is the one growing, and it turns out to be measurable. You just have to look in the right place.
There is a place to look
When an AI assistant answers a shopping question against Shopify's ecosystem, it is searching the Global Catalog: a cross-merchant index of products you can query directly. You can ask it the same questions your customers ask, and read back which products it surfaces, in what order, and from which sellers. That is the raw material for an honest answer to "where do I rank in AI shopping."
I have been running exactly this against real stores, and three numbers come out of it that actually matter.
The three numbers worth tracking
Presence and rank. For a given query, do your products appear at all, and how high? This is the closest analogue to a search ranking, and it is the first thing to check. Most stores discover they are present for their own brand name and absent for the generic category searches that bring new customers. That single finding reframes the whole problem.
Share of agentic voice. Across a set of queries that matter to you, in what fraction do you appear at all? One store might show up for eight of ten relevant searches and another for one. That percentage is a blunt but useful health score, and watching it move tells you whether your work is paying off.
The buy box. When several stores sell the same product, the catalog clusters them and the assistant picks one offer to recommend. If you are one of those sellers, the question is not whether you appear but whether you are the chosen offer, and if not, what beats you. Price, shipping, availability, and ratings all feed that choice. You can see the competing offers and where yours lands.
Why a one-time check is not enough
The catalog reflects live pricing, availability, and merchant preferences, so a snapshot is a moment, not a trend. A competitor drops their price, your variant sells out, you rewrite a product title, and your position moves. The useful version of this is tracked over time: the same queries, run on a schedule, so you can see a rank improve after you fix a product, or catch the week a reseller started undercutting you on your own item.
That is the difference between a screenshot and a rank tracker. The screenshot tells you today. The tracker tells you the direction.
What to do with the numbers
The numbers are only useful if they point at an action. Absent on category queries usually means brand-only product titles and missing taxonomy. A weak buy-box position usually means price, shipping coverage, or ratings. A rank that slipped usually traces to a product edit or a stockout. In every case the metric tells you where to look, and the fix lives in your catalog data, which is the part you control.
Where we land
You would never run a store blind to your Google rankings, and AI shopping is heading the same way: a channel you can measure, with positions you can move. The mechanics are not a black box. The data is queryable, the signals are knowable, and the work that improves them is the same clean-catalog work that helps you everywhere else.
This is the next thing we are building into AgentReady: not just making your catalog legible to assistants, but showing you where you actually stand inside the Global Catalog over time, on the queries that matter to your store, so you can see whether you are winning instead of guessing.
Comments
Every comment here comes from a verified email. Write yours, confirm from your inbox, and it's live.
Loading comments…