Frequently asked questions

Every cycle runs the same loop: observe, verify, advocate. We record what assistants answer to the questions your customers actually ask. We check every claim in those answers against your verified business facts. Then we fix the sources that fed the wrong ones — and re-run the same questions to show you what changed.

Each step leaves a paper trail: dated transcripts, a claim-by-claim accuracy report, and a monthly delta of what was fixed, what regressed, and what's new. The method, in full.

The assistants your customers actually use: ChatGPT, Google Gemini, Perplexity, Claude, and Meta AI. We run them the way real customers meet them — with web search on — and capture the sources each answer cites, because the citations tell us where every answer came from.

We also test one model with web access turned off, deliberately. It shows what AI "remembers" about your business without looking anything up — usually nothing. That's not a failure of the product; it's the reason the work matters: AI has no memory of your business, so the sources it reads have to be right. Our grounding explainer tells that story in plain terms.

Three things: whether AI finds you, whether it gets you right, and whose sources it trusts.

• Discovery Rate — how often you appear when a customer asks without naming you ("who does ceramic tint near me"). This is the number that feels like visibility.
• Answer Accuracy — when AI talks about you, is each fact correct, wrong, or missing? Hours, phone, services, prices — checked claim by claim.
• Citation Share — whose sources the assistants cite: yours, your competitors', or neutral directories. It's the early signal — it usually moves before Discovery Rate does; our Citation Share guide explains why.

Assistants vary their answers, so we repeat every question and report rates with confidence intervals. A change we report is a real change, not noise.

No — and nobody honestly can. An assistant decides each answer in the moment, from whatever it reads that day. Anyone who guarantees you a mention or a ranking is guessing.

What we promise instead is narrower and checkable: accurate sources, documented fixes, and a re-test that shows whether the answers moved — transcripts before, transcripts after. Whoever you hire for AI visibility, it's worth asking to see their transcripts. We wrote up why nobody can guarantee this, and how to buy safely.

Source fixes are usually verifiable within the first re-test cycle. A corrected listing, an installed schema block, a cleaned-up duplicate profile — those we can confirm quickly, with a citation.

Movement in the answers themselves comes after, and honestly: Citation Share — whose sources the assistants cite — typically moves before Discovery Rate does. How fast discovery moves depends on your competition and your starting position, so we don't promise a number. We show you the trend, with confidence intervals, and every improvement claim comes with a before-and-after transcript.

Usually some changes — and we keep them small and specific. The common ones: adding structured markup (schema) so machines can read your facts, making your name, address, and phone consistent everywhere they appear, and publishing answers to questions your customers ask that no page of yours answers.

On Advisory, we hand you the fixes ready to install — copy-paste complete. On Concierge, you approve and we install them for you.

Yes — through the sources assistants cite. When an assistant recommends someone else, we record who it named and which sources it read to get there. That's what Citation Share measures: whose material the assistants trust for your kind of question.

It turns a loss into a to-do list. You don't just learn that a competitor got the call — you learn which source earned it for them, and whether you can be on it too.

They're three layers of the same job: being findable. SEO — search engine optimization — helps your site rank in search results, and it's still the foundation. AEO — answer engine optimization — structures your content so assistants can quote you accurately. GEO — generative engine optimization — improves how AI models retrieve and recommend your business.

We use "AI visibility" as the plain-English umbrella for the AI layers, so you don't need to keep three acronyms straight to know what you're buying.

SEO works on your position in a list of links; we work on the answer itself. The job isn't to rank your page — it's to make sure that when an assistant answers a customer, the sources it reads describe your business correctly and completely.

The other difference is evidence. Every finding we report is cited to a dated transcript, and our metrics come with confidence intervals, not adjectives. We rebuilt our own measurement system when we found it was flattering us — that's the standard we hold your data to.

By repetition. One answer proves nothing, because assistants vary their responses — so we ask each question repeatedly, on each platform, and report rates with confidence intervals instead of cherry-picking single runs.

When we say something changed, the change is bigger than the noise. And every run is kept as a dated transcript, so you can check any claim we make yourself.

Your data stays confidential — never shared, never sold. Most of what we work with is public-facing anyway: listings, reviews, what assistants say about you. Anything you share beyond that stays between us, and we'll sign an NDA on request.

We're a small consultancy and won't pretend otherwise: your data lives with mainstream US providers, access is limited to the people doing your work, and the promise is confidentiality we can actually stand behind — in writing if you want it.