It’s not a ranking, and it isn’t random. AI builds a picture of each business from signals across the web, then names the ones it can read, identify and trust. Understand that, and the rest follows.
AI doesn’t rank businesses — it builds a picture of each one from signals across the web, then names the ones it can read, confidently identify, and trust. Three things decide whether you’re in the answer: can it reach your information, can it resolve you as a specific business, and do independent sources back up what you claim.
There’s no ranked list inside an AI answer. The assistant assembles a reply from sources it trusts and names a few businesses. So the question isn’t “how high do I rank” — it’s “does the model trust me enough to put my name in the answer.” Everything below is what that trust is built from.
Your information has to be accessible — real HTML, AI crawlers allowed in, pages that don’t hide behind heavy scripts. If an assistant can’t read you, nothing else matters.
It has to be sure which business you are. Consistent name, location and services across the web — plus structured data — let it lock onto you as one specific entity. Inconsistency reads as uncertainty, and uncertainty gets left out.
The model leans on corroboration. Reviews, directories and reputable mentions that agree with what your own site claims are what turn “a business that exists” into “a business worth naming.”
AI recommends on attributes, and attributes come from being specific. “Accounting for hospitality businesses in Auckland that use Xero” gets matched to a real, narrow question far more readily than “accountant.” The same applies to “mortgage broker for first-home buyers in Orewa” over “finance.”
Generic positioning is invisible positioning — in both Google and AI. The businesses that win are the ones specific enough that the model can confidently say “this one fits.”
It’s structural legibility plus corroboration. Slower than a trick, and the reason it actually holds.
Each assistant sources its answer differently, but all of them reward reachable, resolvable, trusted businesses. The how-to changes per engine:
When the signals are right, a business shows up in more than one place. Mortgages with JJ is named by both ChatGPT and Google’s AI Mode for “mortgage broker Orewa” in logged-out searches (June 2026). Karin Blaauw is named by ChatGPT among Hibiscus Coast agents and ranks organically on Google for her suburbs. Neither is a national brand — they’re local businesses whose reachable, consistent, corroborated signals let multiple engines reach the same conclusion.
As always, engine-specific and dated — what’s true today is verified, not assumed to hold everywhere forever.
What you control — and what you don’t. No one controls how a model phrases an answer, and no one can guarantee a mention. What you control is the inputs: whether you’re readable, whether you’re resolvable as one specific business, and whether independent sources back you up. Get those right and you’re in the running every time the question comes up.
Run the free AI Readiness Check — a plain-English snapshot of whether AI can read your business, and whether it names you today.
Get your free AI snapshot →