If You're Not in the Top 3 an AI Model Recommends, You're Invisible to the Customer
By Malin Gaertig · September 17, 2026
Brands outside the top three products an AI model recommends are effectively invisible at the moment of decision. The proof takes three questions and no software.
- —Outside the top three products an AI model recommends, a brand is effectively invisible at the moment of decision.
- —Search visibility and AI selection are different measurements — strong SEO does not guarantee you get chosen.
- —Ask a model to recommend the best product for a specific need and budget, then count who appears.
- —Asking the model why it chose them gives you a shortlist, your competitors and its reasoning.
- —MARK's Share of Model measures selection at the point of decision, not presence somewhere in an answer.
| Search results | AI selection | |
|---|---|---|
| Who decides | The engine returns ten options and the customer decides. | The model decides first and returns a shortlist. |
| Room to recover | A second page, and a lower ranking you can work up from. | Outside the shortlist there is no second page. |
| What you measure | Ranking, traffic and share of search. | Whether you are selected, and why. |
“If you're not in the top 3 products an AI model recommends, you're most likely invisible to the customer.”
That was the observation from Blake England, Head of Digital Product at JB Hi-Fi, on stage at the ARC AI Summit.
And it points to a bigger shift. The move from search to AI isn't just that the interface has changed. It's that the selection process has changed.
Google could show you 10 results and let the customer decide. An AI model increasingly does the deciding first.
It researches the category, compares the options, applies the customer's requirements and returns a shortlist.
If your product isn't in that shortlist, there may be no second page to rescue you. No lower ranking to work your way up from. Increasingly, you are simply absent from the decision.
What does visibility mean when an AI model decides first?
That changes what 'visibility' means. A brand can have strong SEO, a large catalogue and plenty of search traffic and still fail to appear when a customer asks an AI model to solve the problem its product solves.
The two are not the same measurement. Search visibility describes where you sit in a list a human scrolls. AI selection describes whether you are inside the set of options a machine hands to a customer who is already deciding.
How do you test whether an AI model actually selects you?
Try this simple test.
Ask ChatGPT, Claude or Gemini: “Recommend the best [product] for [specific need] under [$X].”
Count the products it recommends.
Then ask: “Why did you choose these?”
Now you have something much more useful than a visibility score. You have a shortlist, your competitors, and the model's reasoning for selecting them.
What is MARK's Share of Model?
That is the beginning of MARK's Share of Model: not whether your brand exists somewhere in an AI answer, but whether it gets selected when the customer is actually making a decision.
Existence is cheap. A model can name your brand on request and still never put you in front of a customer who is ready to buy. Selection is the commercial event, and it is the one most brands have never measured.
Is AI search just a new version of the rankings?
AI search is not creating a new version of the old rankings. It has created a new selection layer.
That's the question MARK was built to answer: who gets selected, who gets ignored, and what the model says when it explains the difference.