Be the brand ChatGPT recommends, not the one it has never heard of.
Buyers now ask ChatGPT, Perplexity and Google's AI Overviews the questions they used to type into a search box. The answers name two or three brands. In the categories we measure, AI Overviews already appear on most buyer queries, and the brands cited are rarely the ones with the biggest ad budgets. AI citation is the fastest-growing component of the Demand Control Index, and the one where a challenger can still overtake the leader.
When you need this
- 01
AI Overviews show on most of your buyer queries and the sources cited are not you.
- 02
A competitor is named in ChatGPT answers for your category and you are at zero. We saw exactly this: 6 of 10 answers for the leader, 0 of 10 for the challenger.
- 03
The models say something about your brand that is outdated or wrong.
What we do
A fixed set of buyer prompts run against ChatGPT, Perplexity, Gemini and AI Overviews every month. Which brands are named, which pages are cited, how that changes.
We audit what each model says about your brand and your competitors, then fix the sources it learned from.
Pages structured so a model can lift a clear, attributable answer: definitions, comparisons, numbers with sources.
Organization, product and person markup, consistent names and the sameAs links that let systems connect your brand across the web.
We reconstruct the prompts people use before they ever reach your site and map them to pages that answer them.
A single brief per topic that serves Google rankings and AI answers. Not a separate strategy per engine.
How we measure
In the yarn category, 59% of buyer search volume already shows an AI Overview. The leader is cited in 6 of 10 ChatGPT answers; the challenger in none. That gap is the opportunity.
Of the answers to your category's buyer prompts, how many name you. Tracked per model, monthly.
Visits arriving from ChatGPT, Perplexity, Copilot and Gemini, and what they do on the site.
Citations turn into people typing your name into Google. We watch that line move.
Things we learned the hard way
From six months of AI-bot logs, our own monitoring experiments and the research. Each one changed how we work.
A brand already in the answer gets cited about five times more often than one that is not. Being known across the web matters more than being findable on one page.
How often a brand is talked about predicts AI visibility far better than its link profile. The correlation with backlinks is weak.
In roughly two thirds of B2B "best X" answers, the engine cited a vendor's own page and then recommended a competitor named on it. Your comparison page can sell for someone else.
Users arriving from ChatGPT hit invented paths hundreds of times. One redirect per guessed address turns each 404 into a citation that lands.
The same prompt swung from first place to absent within hours. We run every prompt several times, with web search on, and extract brands in a separate step so the model never knows who we are measuring.
Find out what the models say about you
Send us your site and two competitors. Before the call we run your category's buyer prompts through the models and bring the answers.