How AI search is changing brand visibility (and what to do about it)

There's a specific moment that convinced me this shift was real. I asked ChatGPT to recommend tools in a category I know well, a category where I could name the top eight products from memory. The answer named four. Two of the four were, by any reasonable measure, weaker products than three it left out. But the answer was confident, well-structured, and completely sufficient for someone who didn't already know the market.

That buyer isn't going to open ten tabs to double-check. They got a shortlist. The research phase is over.

If you're responsible for a brand, this is the new visibility problem, and most of the tooling and habits you built for the last era can't see it.

What actually changed

Search used to be a two-step process with you in the middle. The buyer typed a query, got a list of links, and clicked. Every step was measurable. Impressions, positions, click-through rates. Your job was to be present and attractive in that list.

AI-mediated search collapses the list into an answer. Someone asks Perplexity "best email tool for a newsletter with 5,000 subscribers" and gets three named products with a paragraph each, sources footnoted. Someone asks ChatGPT to compare two vendors and gets a verdict. Google itself now answers many queries directly in AI Overviews before a single organic link appears.

Three properties of this shift matter for brands:

It's zero-click by default. The answer is the destination. When a buyer does eventually visit your site, they often arrive as direct traffic or a branded search, and the AI's role in sending them is invisible to your analytics.

It's a shortlist, not a ranking. Position seven on Google still got seen. An AI answer names three to five brands and the rest don't exist for that query. The difference between "included" and "excluded" is binary and brutal.

It's opaque. There's no Search Console for ChatGPT. No impression counts, no position data, no query reports. Unless you go measure it yourself, you have no idea what these engines say about your category.

Why your dashboards are quiet about this

Your rank tracker can show green while AI answers exclude you completely, because ranking on Google and being cited by an AI engine are correlated but distinct. Engines like Perplexity retrieve live web content, but they favor certain kinds of sources: community threads, comparison articles, documentation, review sites. A page can rank fourth on Google and never get cited, while a two-year-old Reddit thread gets quoted in half the answers in your category.

Meanwhile, models like ChatGPT and Claude answer many questions partly from training knowledge. Whether you exist in that knowledge depends on how much was written about you, where, and how clearly, over the past several years. No metric you currently track approximates this.

The referral traffic you can see from AI tools understates the effect badly, because reading an answer doesn't produce a referral. The visible trickle is the shadow of a much larger volume of decisions being shaped upstream.

What to do about it

The good news: this is tractable. Not gameable, but tractable. Here's the sequence I'd run, and it starts with measurement because everything else is guessing without it.

1. Measure your baseline before you change anything

Write down 15–20 questions a real buyer in your category would ask. Not keywords. Questions: "best X for Y," "alternatives to [leader]," "is [competitor] worth it," "how do I solve [the problem you solve]." Run them across ChatGPT, Claude, Perplexity, and Gemini. Record which brands get named, which get cited as sources, and where you appear. Note your competitors' hit rates next to yours.

This takes an afternoon by hand and it will reframe your priorities faster than any audit you've paid for. (Measuring this continuously and automatically is a core part of what Brandlism does, but the manual version is free and you should do it either way.)

2. Make your own site retrieval-friendly

Engines that pull live content favor pages that state facts plainly. Somewhere on your site, in plain declarative sentences, it should say what your product is, who it's for, what it costs, and how it differs from the obvious alternatives. Clever positioning copy that never quite says what you do retrieves badly. Structured data helps. Consistent naming helps more than you'd think; if your brand is spelled three ways across the web, you're splitting your own identity.

A basic site audit catches most of this. It's unglamorous and it compounds.

3. Show up in the sources engines actually cite

Trace the citations in your baseline measurement. In most categories they cluster: a handful of comparison articles, some Reddit and Hacker News threads, review platforms, and documentation. That cluster is your target list.

For comparison articles, that's classic outreach with unusually precise targeting; you know exactly which pages feed the answers. For community threads, the only durable play is genuine presence: a product people recommend unprompted, and a founder or team who participates honestly where their market hangs out. Do not fake this. Astroturfed recommendations get detected, get deleted, and poison the well.

4. Publish answer-shaped content

The queries AI engines handle are questions, so content that directly answers questions gets retrieved. Honest comparison pages (including against competitors, including where they win), "best X for Y" roundups you're credibly part of, and clear how-to material tied to the problem you solve. This overlaps heavily with good SEO content strategy, which is convenient: most of the work pays off on both surfaces.

5. Keep measuring, because the ground moves

Models update. Retrieval sources shift. A competitor ships a comparison page and takes over three answers they didn't own last month. A one-time snapshot decays in weeks. Whatever cadence you can sustain, monthly at minimum, re-run your question set and track the trend. The brands that treat AI visibility like they treated rank tracking, as a continuous measurement discipline rather than a one-off audit, are the ones who'll notice changes while they're still cheap to respond to.

The realistic take

AI search isn't replacing Google tomorrow, and your SEO work isn't wasted; much of it feeds both layers. But the share of buying decisions shaped by AI answers is growing every quarter, it's currently invisible to standard tooling, and the brands being recommended today are accumulating an advantage that gets harder to displace as these answers calcify into defaults.

You don't need to believe any grand prediction to act on this. You just need to answer one empirical question: when the engines answer your category's questions today, are you in the answer? An afternoon of manual checking will tell you. If you'd rather have it measured continuously, across engines, with your competitors tracked alongside, that's what we built Brandlism for. There's a 14-day trial, no card required, at brandlism.com.

Start with the measurement. Everything else follows from what you find.