When a customer asks ChatGPT or Google’s AI for “a reliable electrician near me,” the AI doesn’t reach into a stored memory of businesses it knows. It runs a search, reads what it finds, and names the businesses it can piece together into a confident answer. That last step decides most of the outcome. A business the AI can’t assemble into one clear picture, because its name, location, category, and claims don’t agree across the web, is a business the AI leaves out of the answer. The work that fixes this looks technical. It’s the oldest work in brand strategy.
How does AI decide which business to recommend?
AI recommendation runs on retrieval, not memory. When a customer asks an assistant for a recommendation, the system generates a set of related searches, pulls current pages from a search index, and builds its answer from what it retrieves in that moment. Google calls this grounding, and describes a “query fan-out” technique that issues “concurrent, related queries” to gather more sources. The business that gets named is the one the system can retrieve and trust right then, not the one with the best-looking website on file.
Google’s own guidance is blunt about the continuity with everything that came before: “optimizing for generative AI search is optimizing for the search experience, and thus still SEO.” The floor to be considered is low. A page has to be indexed and able to appear with a snippet, and Google states there are “no additional technical requirements” beyond that. Clearing the floor gets you into the pool of candidates. It does not get you chosen.
Picture two heating and cooling companies in the same city. Both have decent websites. One presents the same company name, service list, and service area everywhere a machine looks. The other reads as an air conditioning contractor on its site, a general handyman on one directory, and a different trade name on another. When an assistant assembles an answer, the first company is easy to describe with confidence. The second is a question mark, and assistants don’t recommend question marks.
AI doesn’t recommend the business with the best website. It recommends the business it can retrieve and resolve.
What does it mean for a business to be “resolvable” to AI?
A resolvable business is one an AI can assemble into a single, consistent identity from everything it finds. The assistant reads your website, your Google Business Profile, the directories you’re listed in, and the platforms where people review you, then tries to reconcile all of it into one entity: this company, in this place, offering these services, for these customers. When the sources agree, the model builds a confident picture and can put your name in an answer. When they conflict, the picture blurs and your name drops out.
The entity an AI describes is stitched together from many hands, not just yours. Google says a Business Profile is compiled from “crawled web content” such as your official website, “licensed data from third parties,” “users who contribute factual information,” and Google’s own interactions with a place. You author some of that. You influence more of it. You control none of it completely. Resolvability is what happens when all of those inputs point to the same answer.
| Element | Content |
|---|---|
| Term | Resolvability |
| Plain definition | How easily an AI can combine everything it finds about your business into one consistent identity. |
| Why it matters | An assistant only recommends businesses it can describe with confidence. Low resolvability means it stays quiet about you. |
| Common confusion | Owners think the goal is more content. The goal is agreement across the content that already exists. |
An AI can only recommend a business it can resolve. If it can’t tell who you are, it won’t risk telling a customer.
Why does inconsistent information keep you out of AI answers?
Conflicting core facts make a business illegible, and an illegible business is a risky one to recommend. When your website lists one business category, your profile lists another, and a directory lists a third, the model can’t be certain the listings describe the same company or what that company does. The facts that have to match are the load-bearing ones: name, address, phone, category, hours, service area, and the claims you make about what you offer. Marketing voice can vary from page to page. These facts cannot.
This is why the local-search world has treated citation consistency as non-negotiable for years, and why it matters more now that a machine is doing the reading. As Whitespark’s Miriam Ellis put it in late 2025, “NAP consistency on structured citations is simply vital,” and AI is “bringing citations of all kinds back into the digital marketing limelight.” A model reading two slightly different versions of your address can’t be sure the records describe the same business, and it resolves that doubt by leaving you out.
A common failure looks like this: a company rebrands or moves, updates its website, and leaves a dozen old listings untouched. The fix is an audit of every place the business is named, and a correction pass until the core facts read the same everywhere. If a fact helps an AI decide whether to recommend you, it has to say the same thing in every place it appears.
Why does AI name a business but link a directory instead of its website?
For local businesses, the AI often names the business but cites a third party, because that’s where the corroborated evidence lives. The answer recommends a plumber and links a review platform or the Business Profile rather than the plumber’s own site. That pattern reflects where your strongest, most independent proof lives, the platforms that establish who you are and whether you’re any good.
Independent proof carries weight because reputation isn’t something a business can assert about itself. BrightLocal’s 2026 analysis of local ranking factors weights the Google Business Profile at roughly 17 percent, reviews at 14 percent, and citations at 9 percent of what drives local visibility, and notes that “many of the same signals that influence traditional local rankings” also shape how a business appears in AI-powered search. The pattern underneath the numbers is old and human. A GatherUp survey cited by Whitespark found that 55 percent of consumers trust what customers say about a brand over what the brand says about itself. An AI weighs corroboration the same way, because it’s drawing on the same web.
Your own site is necessary but rarely sufficient. In local search, other people’s words about you often outweigh your own.
What actually makes AI more likely to recommend you?
The strongest signals are authority, corroboration, content depth, freshness, consistency, and speed, and every one of them is something that has always marked a credible business. An analysis of 216,524 pages by SE Ranking, reported in Search Engine Journal in late 2025, found referring domains to be the single strongest predictor of whether ChatGPT cites a page, with domain traffic next and domain trust also a strong signal. Depth and freshness moved the needle. Even page speed acted as a gate. None of it was a markup trick.
The gaps between strong and weak are wide. In that analysis, the highest-trust domains averaged 8.4 citations, against 1.6 for low-trust ones. Long pages beat short ones, 5.1 citations past roughly 2,900 words versus 3.2 for pages under 800 words. Freshness told the same story: 6 citations for pages updated within three months, 3.6 for stale ones. Speed worked as a gate, with pages that rendered their first content in under 0.4 seconds averaging 6.7 citations against 2.1 for slow pages. Separately, a 2024 study from researchers at Princeton and IIT Delhi (published at KDD as “Generative Engine Optimization”) found that adding citations, statistics, and direct quotations to a page were among the most effective ways to lift its visibility in AI answers, by up to 40 percent, with lower-ranked sources gaining the most.
If you want a place to start, work in this order:
- Make your core facts agree everywhere. The cheapest, highest-return work. It’s the foundation every other signal sits on.
- Earn real corroboration. Reviews and accurate listings on the platforms that get cited in your category.
- Write pages that answer the decisions a buyer faces, with evidence, depth, and dates, not thin pages built for a search engine.
- Confirm you’re reachable and fast, so the machines that read you can finish the job.
As a rule of thumb, the things that make you credible to a person are the same things that make you retrievable to an AI.
Why is this brand strategy and not just SEO?
Deciding what’s true about your business and saying it the same way everywhere is brand coherence, and coherence is the oldest job in brand strategy. Every signal in the last section rests on one thing underneath it: a single, governed set of facts about who you are, expressed consistently across every surface a machine can read. That governance is not a technical chore bolted onto marketing. It’s the core discipline of building a brand. An assistant reading your business is a new kind of audience, one with no patience for contradiction.
Subverse has always held that the fundamental job of a brand is to be understood, clearly and consistently, over time. AI didn’t change that job. It made the job legible and, for the first time, mechanical. A person encountering three slightly different versions of your business fills the gaps and gives you the benefit of the doubt. A model encountering the same three versions lowers its confidence and moves on to a competitor it can describe cleanly. Coherence used to be judged by people, who forgive. Now it’s also judged by a machine that can’t.
This is also why a dashboard that tracks your “AI visibility” can tell you whether you’re appearing but not why, and can’t fix what it finds. The cause of a low number lives upstream, in whether your signals agree. A score is a symptom. Coherence is the treatment. Brand coherence used to be a matter of taste. Now it’s the thing that decides whether an AI can recommend you at all.
Conclusion
The businesses AI recommends are the ones it can resolve into a single, credible identity. That resolvability is built from agreement, your core facts saying the same thing everywhere, backed by real corroboration and pages worth citing. Start with the facts, because they’re the foundation every other signal depends on and the cheapest thing to fix. The pitfall to avoid is treating this as a technical project handed to whoever manages the website, when the real work is deciding what’s true about your business and holding that line across every place it appears. That decision, and the discipline to keep it consistent, is brand strategy. AI just raised the cost of getting it wrong.
