How the process works

I check what ChatGPT and Gemini tell your customers about you today, score your site's readiness along seven dimensions, and hand you a prioritised, vendor-independent plan. The work has four steps and I do most of it in the background. From you I need about an hour at the start, a few access credentials, and your decisions on the proposals. I don't dress up the numbers: what I can't measure, I don't claim.

One thing up front. This is not a chatbot, and not a button I press so your company sits at the top of ChatGPT tomorrow. It is measurement, a plan, and follow-through — human judgment over dated data. The steps build on each other, each has its own deliverable, and you can slow down or stop at any point.

The four steps

The first one is free and already shows where you stand. The rest follow only if you decide it's worth going deeper.

  1. Free mini-check.

    I put one real buyer question — for your trade, in your town — to two AI models and show you the answer: who gets named, and who is left out. Alongside it I check three quick technical signals on your site: can the AI crawler reach it, is the content readable without JavaScript, is there basic structured data. You get the result within two business days, with an explanation. Free, no obligation.

  2. Full audit with competitor comparison.

    The detailed measurement. I score your site along the seven dimensions with dated, traceable data, and put your buyers' questions to several models, not just one. Competitor positions are named, concretely, in a private report — it is your decision material, not public content. The method is documented on the methodology page.

  3. Action plan.

    The measurement becomes a prioritised plan: what to fix first, what next, each with an estimated effort. The plan is vendor-independent — it doesn't tie you to any agency or tool, and your own developer can deliver it. The steps are concrete work packages a contractor can quote against, not abstract advice.

  4. Follow-through and re-measurement.

    After the fixes I re-measure with exactly the same rubric, monthly. You see progress on the same seven dated dimensions — not on promises. Where a number gets worse, I print that too; that is the whole point of follow-through.

When should you expect results?

The most common question, and the most important to settle up front. The three workstreams move at very different speeds.

WorkstreamTimeWhat it delivers
Technical fixes2–4 weeks Unobstructed AI-crawler access: lifted robot blocks, readable content, basic structured data.
Content restructuring4–12 weeks Answer-ready pages, question-centred structure — content an AI can lift a quotable answer from.
External presence and reputation6–12 months Reviews, independent mentions, directories and sources. The slowest band — and the one with the most weight.

Why does the third band take so long? Because it isn't decided on your website. The named sources behind AI answers are mostly not companies' own pages: in an analysis of 7,000+ citations, Wikipedia alone accounted for 47.9% of ChatGPT's citations, and forums supplied nearly half of Perplexity's (source: Digital Bloom's citation report) — reviews, directories and independent mentions decide whether a model names you. That presence cannot honestly be built in weeks.

Anyone promising AI recommendations within weeks is either wrong or selling something else.

Technical fixes are genuinely fast. Whether ChatGPT or Gemini names you to a buyer, however, depends on external presence — and that is measured in months. I say this up front because I won't promise what your customers could disprove a few weeks later. Readiness and recommendation are two different things.

One more number for context: AI is far more selective than classic search. SOCi's 2026 Local Visibility Index — built on hundreds of thousands of business locations — found ChatGPT recommends only 1.2% of local businesses, Gemini 11%, Perplexity 7.4%, while the same brands appear in Google's local 3-pack 35.9% of the time (source: SOCi Local Visibility Index, 2026). There is no published fixed review-count threshold — what operates is a confidence threshold: AI-recommended locations average 4.3 stars, and highly visible brands answer reviews within days, not weeks (same study). Six to twelve months isn't slowness — it's what the real work takes.

What I need from you

  • Access. Your website's admin (or your developer's contact if you'd rather not share it), your Google Business Profile if you have one, and Search Console if it exists — if not, that's fine, it can be set up.
  • One hour at the start. A single conversation covering what your company does, who your buyers are, and what questions they would ask. This produces the buyer questions for the measurement.
  • Decisions on the proposals. The plan gets made; what gets implemented, and in what order, is your call.

Experience says the most common delay isn't the work — it's missing access. That's why I ask for these up front.

Ready to see what the AI says about you today?

The mini-check is free, takes two business days, and commits you to nothing. If you want to go deeper afterwards, pricing is published up front.

Request a free mini-check See pricing