The Median Brand Appears in Just 12% of AI Answers About Its Own Market

GEOforge Research · AI Visibility

We measured how often AI names real brands in their own categories. Half of them show up in fewer than 1 in 8 answers.

By Paris ChildressGEOforge6 min read

There's a comfortable assumption baked into most conversations about AI search: that your brand is at least in the conversation. The data says otherwise.

What AI Share of Voice actually means

AI Share of Voice is the share of AI-generated answers in which your brand is named when people ask buying questions in your category. It's the GEO equivalent of organic visibility: not where you rank on a page, but whether you make it into the answer at all, and how prominently, relative to everyone else competing for the same recommendation.

It's the number that matters now, because the answer is increasingly the destination. Buyers ask, the model responds with a shortlist, and brands that aren't on it never get considered. So the first honest question isn't "are we winning?" It's "are we even there?"

We measured 21 brands across 65,478 answers

Across 21 tracked brands and 569 real buyer prompts, ChatGPT named the brand in 28.1% of answers, with an average Share of Voice of 5.3%. That already sounds low. But the average flatters the picture, because a few well-established names pull it up. Look at the brands individually and the floor drops out.

The median brand was named in 12.4% of answers about its own market. Nearly half the brands we track, 8 of 17, appeared in fewer than one answer in ten. The gap between the most and least visible brand was enormous: 0.6% at the bottom, 62.7% at the top. This isn't a story about brands losing the AI conversation. Most of them are simply absent from it.

Why is the gap between brands so wide?

Because AI visibility isn't earned the way page rankings were. The brands at the top of that range tend to have a dense footprint in the places models actually read: third-party reviews, analyst coverage, comparison content, reference pages, and consistent mentions across the wider web. The brands at the bottom usually have a polished website and very little else. The model has nothing to pull them into the answer with.

That's the uncomfortable part. A great product and a great site are not enough to be visible in AI. Presence is built off-site, in the sources the model trusts, and most brands have never deliberately built it.

What a healthy AI Share of Voice looks like

The honest benchmark isn't zero and it isn't 100%. It's relative. The right question is how your mention rate compares to the leader in your category and to the two or three competitors AI keeps naming instead of you. A brand sitting at 5% in a category where the leader holds 40% has a clear, measurable gap to close, and a clear set of sources to go and earn.

How to find your real baseline

• Measure mention rate across a real set of buyer prompts, not one flattering query.
• Report it per brand and benchmark against the category leader, not against zero.
• Separate the prompts where you never appear from the ones where you sometimes do.
• Track the gap to the two or three competitors AI names in your place.
• Re-measure on a schedule so the baseline becomes a trend.

Knowing your real number is uncomfortable, and that's the point. You can't close a gap you've never measured. Tracking it is table stakes. Closing it is the work that actually moves a market.

Sources & method. All figures verified against the GEOforge measurement database on 16 June 2026. Corpus: 65,478 ChatGPT answer-runs across 21 tracked brands and 569 categorised buyer prompts, measured April–June 2026. Per-brand figures cover the 17 brands with sufficient run volume. Figures are ChatGPT-specific; cross-engine results may differ.