By Scott Varland
The idea itself isn't new. Marketers have been trying to optimize for machine answers for more than a decade: featured snippets, voice assistants, knowledge panels, structured answers. The shift from SERPs to synthesized answers started years ago.
What's new is the measurement problem.
Search behaved like a list. You could rank #3 for a query, measure impressions and CTR, and build an entire performance discipline around that. The system was deterministic enough that dashboards made sense. So, as an industry, we built a ton of dashboards.
Large language models don't behave that way. Their outputs are probabilistic, contextual, and synthesized across vast networks of learned associations. The same question can produce different answers depending on phrasing, retrieval sources, session context, or how a given model works through the problem. Every output is frustratingly unique.
But most AI visibility tools still try to force this world into a search-style dashboard. Run prompts. Track mentions. Count citations. Generate a visibility score. Monitor it over time.
After using these tools extensively, I keep coming back to the same conclusion: they are measuring the symptom, not the understanding.
A better mental model is this: imagine sending a team of a dozen junior analysts to research your company. They spend a week reading the internet: your website, social channels, reviews, news coverage, competitor pages, forums, industry reports. Then you bring them into a room and start asking questions. What does this company actually do? Who are its real competitors? What is it known for? When would you recommend it, and when wouldn't you?
Large language models are doing that exercise at massive scale.
So the real strategic question for brands isn't whether an AI mentioned them in an answer. It's what the AI actually learned about them. Because that learning shapes answers about your brand whether or not your name even appears in the response.
Did the system understand the business correctly? Did it miss the thing the company most wants to be known for? What does it "trust" about you?
AI is serving as a helpful assistant to the consumer, not to the brand.
Once that picture forms, it starts shaping the solution space everywhere.
Most of the current market measures share of answer. What brands increasingly need to understand is share of meaning.
Not just whether the machine says your name, but what the machine thinks your name means.
The work starts with a structured audit — not of mentions, but of comprehension. What do the major models actually believe about your brand, your category, and your competitive set? The gap between what they believe and what's true becomes the strategic brief.
That is where the brief changes, the content strategy changes, and the competitive frame changes.
Stop measuring mentions. Start understanding influence. Talk to us.