Guides · The AI shelf

How to measure your share of the AI shelf

A practical method for measuring how often AI shopping assistants recommend you: choosing the questions, asking them properly, scoring the answers and reading the trend.

Key takeaways

  • Choose 10 to 25 shopping questions your customers really ask, mixing broad, budget and use-case questions.
  • Ask each question several times on each assistant, because answers vary from one run to the next.
  • Score how often you’re recommended and where you rank, and record who is recommended instead and the reasons given.
  • Watch the trend week to week, and look at what competitors changed when your share moves.

When a shopper asks an AI assistant what to buy, it answers with a short list. Your share of the AI shelf is how often you’re on that list for the questions that matter to you. Here’s how to measure it in a way you can trust.

Choose the right questions

Start from how your customers shop, not from your product names. Aim for 10 to 25 questions, in a mix like this:

  • Broad: “What’s the best waterproof hiking jacket?”
  • Budget: “Best waterproof hiking jacket under $200?”
  • Use case: “What jacket should I take for hiking in Scotland in spring?”
  • Comparison: “Is [your product] better than [competitor product]?”
  • Problem: “My rain jacket keeps wetting out. What should I buy instead?”

Your search data, customer service questions and on-site search terms are good sources. Keep the list stable, so you can compare week to week.

Ask each question several times

AI answers aren’t fixed. The same question can return different products on different runs, and answers can depend on location and on what the assistant knows about the person asking. A single answer is an anecdote. To measure:

  • Ask each question several times on each assistant (we use 10).
  • Ask from a clean session without personal history, and from the country you sell in.
  • Keep the wording identical each week.

Score the answers

For each answer, record:

What to recordWhy
Whether you’re recommendedThe core of your share.
Your rank in the listThe first pick gets far more attention than the fifth.
Which competitors are recommendedShows who is winning the shelf instead of you.
The reasons givenPrice, reviews, a feature: tells you what to change.
The sources citedShows which review sites and pages shape the answer.

Your share for a question is simply the number of answers that recommend you, divided by the number of answers. If you’re picked in 7 of 10 answers, your share is 70%.

Read the trend, and find the cause

One week’s number means little; the trend is what matters. When your share moves, look at what changed just before it:

  • Did a competitor cut its price, launch a product or run a promotion?
  • Did a review site the assistant cites publish a new list?
  • Did your own pages change: price, stock, product details or reviews?

This is why it helps to measure the AI shelf in the same place as competitors’ prices and promotions: the cause is often a competitor’s move, and it’s much easier to spot side by side.

What you can influence

You can’t buy a place in organic AI answers, but you can make it easier to be chosen: accurate, detailed product information, clear prices and policies, genuine reviews, and presence on the independent sites assistants cite. See how AI shopping assistants choose products.

Doing it by hand? Ten questions, four assistants, ten runs each is 400 answers a week. That’s why most brands automate it. Ghost Watch asks the questions you choose and records all of the above, next to your competitors’ moves.

Terms in this guide