TL;DR
- Everyone measuring AI search is defaulting to the same KPIs, but there is no one-size-fits-all AI search KPI
- New research shows 54% of AI search categories lack a clear category owner, and those categories represent 89% of all AI search demand
- If someone else owns your category, you should stop chasing visibility and instead focus on accuracy: when you do show up, does the model describe you the way you actually want to be described?
- Accurately measuring AI search KPIs requires a brand canon, a living document of your brand’s claims, positioning, facts, and more
Your Visibility Score Can Be Perfect and Still Be Wrong
AI can mention a brand in every prompt you track, but that brand can still be losing.
The industry has a default KPI set for AI: visibility, share of voice, mention rate, citation count. Except they only answer one question: did your brand show up?
Earlier this year, our team was presenting AI visibility numbers to an enterprise B2B client. The CMO asked something we’d never heard before: “We’re showing up, but is our brand being recommended?”
We didn’t have the answer then, so we created a new metric that looks for recommendation language when a brand is mentioned. Think words like “recommend,” “best,” “top,” or “ideal.”
Along the way, we realized we had to think differently about AI visibility and the metrics that matter.
For more on why you should stop blindly chasing AI rankings, check out my teammate Alisa’s blog on how to fix how LLMs see your brand.
Most of AI Search Still Lacks Brand Leaders
New research from Kevin Indig's Growth Memo puts a number on something a lot of us have felt anecdotally: most categories in AI search do not have a settled answer yet.
The study tracked 1,094 categories across 5 prompts from January to June 2026, analyzing more than 50,000 brands and 600,000 citations in ChatGPT. It sorted every category into one of three tiers:
If you analyze the data by actual demand instead of category count, the lack of ownership is even clearer: 89.3% of AI search volume sits in categories with no clear owner.
Indig's own conclusion is a sounding alarm for investment today because ownership compounds and it's expensive to unseat later and I agree. The longer you wait, the easier it becomes for other brands to claim leadership in categories that haven't been settled yet.
How fast will that happen? We don't know.
But we do know that every moment you wait is a missed opportunity to be visible where your customers are already searching.
And where does that leave brands operating in a category with clear leaders?
If you’re not the category leader, gaining visibility in AI search will take patience and consistency. Brands aren’t built overnight. The only way to eventually win here is to look for small pockets of opportunity.
The problem is this strategy takes time that marketing leaders can’t afford, especially if you’re operating on the wrong KPIs to begin with.
The other issue Indig states is that his study does not measure sentiment, recommendation quality, user trust, or purchase impact. But brands need those details to understand their position and make accurate bets.
Visibility Metrics Mean Nothing Until You've Defined What’s "Correct"
Visibility tells your brand only that you were mentioned; not whether you were mentioned the way you want to be.
To answer the second part of that equation, you need something to compare to. Not a vibe or a hunch, but an actual reference document that includes:
- Your positioning
- Your messaging
- The facts about your business
- The story about who you are and why someone should choose you
Seer calls this document a brand canon.
Without a brand canon, every AI mention is unlabeled data. ChatGPT calls you "budget-friendly" and you have no way to know if that's a compliment or a demotion. Perplexity attaches three caveats to your sentiment score, and you can't tell if that's drift or just noise.
You need two points to analyze the distance between them, and most brands only have one: whatever the model happens to be saying right now, with nothing to measure it against.
This is the part of GEO measurement that's being treated as optional, when in reality it's the prerequisite that gives meaning to every other number.
The Right KPI Depends on Where You Stand
Put the category research and the canon problem together, and a clearer measurement framework takes shape. The KPI that deserves your attention changes based on which tier you're in and where you stand in the tier.
Once you know what tier you fall in, you can figure out which KPIs to focus on. Seer's three core AI search KPIs, developed by our VP of Analytics John Lovett, give each tier a specific number to own instead of a vague directive to “measure accuracy” or “keep pushing visibility.”
Those core AI search KPIs are:
- Be Seen: AI Signal Rate - How often your brand is mentioned in AI generated answers for queries in your category
- Be Believed: Answer Accuracy Rate - How accurately AI systems represent your brand, measured through a structured rubric
- Be Chosen: AI Influenced Conversion Rate - The conversion rate among users or sessions influenced by AI surfaced content
The second piece of the puzzle is determining whether you’re a category owner or not, as that will inform your priorities.
What to do if you aren’t a category owner:
- Pinpoint your top competitors and focus on outperforming them. Instead of chasing visibility wins against the category owner, bucket your competitors into three groups: Category Leaders (who by far and large own the space), Core Competitors (who we compete with more directly), and Emerging/Trailing Set (who we're already outperforming).
- Then focus on defending the leads you have against your Core Competitors. Your goal is to pick up new wins within that same bracket, while still keeping an eye on where the Category Leaders win. You can show measurable wins in the near term, while working toward the longer-term goal of overtaking the Category Leaders.
What to do if you are a category owner:
- Strengthen your position. The same logic applies in reverse for category owners. Visibility is no longer a risk, so your focus should be on making sure the story AI tells about you is what you want it to be.
Real-World Example 1: Winning Visibility But Losing the Recommendation
We ran this framework for a decking brand client who was a Category Owner. They were winning the visibility game outright:
- This brand was the default answer
- Mentioned more than any competitor
- First in the AI response more often than not
- 78% of mentions were positive, 1% negative, and the rest were mixed (not inherently positive or negative)
- Cited 3x more often than competitors with language like “industry standard” and “go-to choice”
Yet in 35% of the responses that mention this brand, the AI model hands the "best overall" ranking to a competitor.
We saw this pattern repeat across hundreds of responses: the competitor gets credited as the true overall winner, while our client gets the runner-up label of "best traditional composite." The brand's advantage rides on distribution, color range, and contractor familiarity. Its competitor gets credited with the true driver of premium positioning: performance. Personalization adds another layer to this, but that's a topic that needs its own article.
A visibility dashboard would call this a resounding win, while a canon-comparison would flag it immediately. Despite winning attention, the brand is being boxed into a specific, lesser story while a competitor gets the growth narrative.
Combatting this starts with knowing what's true, and that's what the brand canon gives us. It tells us where to actually compete and where to concede, because not every race is ours to win. That's not our positioning, and it's not how we want to be known.
The goal isn't to be known for everything. It's to be represented clearly and favorably, so when an LLM lists you next to competitors, you're the easy choice.
Define the canon first. Then reinforce it everywhere: our own site, earned media, social, and UGC.
Real-World Example 2: Comparing to the Wrong Opponent
The second example involves a client that was benchmarking their performance against a Category Owner. Our vacation-rental brand client tracks a strong 42.5% share of voice against its 12-brand competitive set.
But compared to the top competitor, they lose every single time. You can probably guess who that top competitor is, the most well-known brand in the world for vacation rentals: Airbnb.
Here’s how AI visibility for our client compared to Airbnb:
Airbnb shows up in nearly every non-branded response about vacation rentals, on every platform tested. This is what an Owner category actually looks like.
Chasing visibility against Airbnb would be a waste of time and an easy way to burn through budget without results.
In this situation, the better questions for our client are: on the mentions the brand does get, is the model describing the brand the way its own canon says it wants to be known? And who are their core competitors we should actually benchmark against.
We’re helping them analyze which gaps they can close with core competitors while building their brand to better compete with the category owner in the future.
This starts with understanding how competitors are winning their visibility. Is it mention driven or citation driven? What does that look like at the topic level? This breakdown lets us target both short term and long term gains.
A citation play might be solved with new or refreshed content. A mention play is more likely an earned media effort that takes longer to build. Either way, both work toward the same goal: durable visibility over time.
Visibility Still Matters, But It’s Not the ONLY Thing That Does
None of this is an argument to stop tracking visibility. You have to show up before anyone can judge whether you showed up well, and in a genuinely unsettled category, the fight for attention is the whole game right now.
My recommendation is narrower than "ignore visibility." You should know which fight you're in before you decide which number proves you're winning it. A brand in an Owner category that only reports visibility is measuring the wrong risk. A brand in an Unsettled category that only reports accuracy is measuring a risk it hasn't earned the right to worry about yet; it hasn't won enough attention for accuracy to be the binding constraint.
It Was Always Important to Document Who Your Brand Is, But the Cost Is Greater Now
The brand canon isn’t a new requirement or something GEO invented. A brand without a documented, centralized point of view on who it is and why customers choose it has always been operating at a deficit. No effective marketing initiative, in any channel, has ever run well without that foundation.
What AI search changes is the cost of not having one. It used to be possible to run a fragmented brand story across channels and mostly get away with it. No single system was reading all of it back to you in one place, at scale, and repeating it to your prospective customers as fact. Now something is.
What to Do Before Your Next AI Visibility Report
Before you present a visibility or share-of-voice number to a client or a leadership team, ask two questions:
- Does this category already have an Owner, and is it us?
- If a model described us perfectly accurately to our own canon right now, would visibility even move?
For most brands, the honest answer is no to both. Visibility only counts how often you're mentioned, and it has no idea how well you're described. A glowing, on-message mention and a lukewarm, off-message one both count toward your visibility score. Making the description perfect wouldn't move that number, because it doesn’t tell you the whole story.
You can use the framework in this blog and John’s blog on how to build a brand canon to create your own canon and start measuring beyond visibility.
Or you can leverage our team’s expertise to move faster: we build the canon from your official sources, derive the tiered claim set, and score how every major AI platform describes you against it. Chat with us to learn more.
.png)
.png)