Measuring What AI Says About You

For years, tracking brand health meant surveys. Awareness, consideration, favourability, tracked quarterly, reported with a margin of error and a long lag between the question and the answer. That's still useful. It's also no longer the only place brand perception lives. A growing share of it now lives inside a chat window, in whatever ChatGPT, Gemini, or Perplexity says when someone asks who's good in your category.

Ask an AI assistant what your brand does, who its real competitors are, and why someone should choose it, and you get an actual answer today, one you can read, compare, and track over time the same way you'd track a survey result. The difference is you don't have to wait a quarter to find out. You can ask again tomorrow.

What actually gets measured

Answer engine optimization (AEO) is the discipline built around that question, and it's less exotic than it sounds. Done well, it means tracking five things on a recurring basis, whether your brand shows up in AI-generated answers, how accurately and consistently it's described when it does, how your competitors show up in the same conversations, which channels and sources are actually feeding those answers, and where the real gaps are between how you want your brand perceived and how it's actually perceived by a model.

That fourth point, channels and sources, is easy to undervalue and probably the most useful one. An AI model isn't inventing an opinion of your brand from nothing. It's pulling from specific pages, reviews, and mentions across the web, and some of those sources carry far more weight in the answer than others. Seeing which sources are actually shaping what AI says about you, where you're showing up and where you're not, and where you should be, is a layer of insight traditional brand tracking can't always provide. That said, AEO is a genuine complement to a survey-based tracker, not a replacement for one.

An insights tool, not a downstream report card

There's a recent, uncomfortable example of what happens when this kind of insight gets pulled after the fact. When Callaway and creator group Good Good released an ad showing one creator shoving another to the ground over a new golf driver, the fallout was fast. Retailers pulled the merchandise, Good Good withdrew as a PGA Tour sponsor, and Callaway pledged $1 million to organizations addressing violence against women. Both companies spent the aftermath arguing about who signed off on the video. The more useful question, raised afterward by Zappi CMO Nataly Kelly, was why no process caught it before it aired.

Once the ad was already pulled, Zappi ran it past 150 AI synthetic respondents trained on real consumer response data. It scored the lowest possible mark, 1 out of 5, and the AI's own recommendation reads almost like an accusation in hindsight, to "soften or clarify the physical moment so it reads as playful rather than awkward." That take was available in minutes. It just arrived long after the ad had already run.

It's the same logic that should apply to AEO. It's not something that only gets checked once a campaign is out in the world. A look at how AI describes your brand, concept, or a piece of creative is a fast, repeatable signal worth having at every stage (brand strategy, creative development, media planning, comms planning) and it's largely wasted if it only ever gets pulled up as a postmortem.

Where this fits at Rule3

We run our AEO/GEO Audits through AI Acuity, the Undercurrent Group's shared capability for reading how brands show up across AI engines. The output isn't a dashboard of scores to admire. AI Acuity is built around actionable solutions, specific, prioritized recommendations for what to fix, where, and in what order, refreshed on a schedule you set rather than something you check once and forget.

The organizations getting ahead of this aren't the ones with the flashiest AI strategy. They're the ones who started measuring how AI perceives their brands and consistently build their marketing approach around it.

Sources

• Callaway and Good Good's Ad Was So Bad, Even AI Audiences Sounded the Alarm, Nataly Kelly, Adweek, 31 August 2026

• Services, Rule3