GEO: AI search
Your site and your listings elsewhere, changed so ChatGPT, Perplexity, Gemini, Claude, Copilot and Google AI Overviews are more likely to name you and describe you correctly.
An AI answer names a few companies, cites some sources, and the buyer may never visit a website. The assistants build that answer from your site and, often more, from what others say about you: review sites, directories, comparison articles, forums and press. Answers also change between runs, so a single screenshot proves nothing. We measure on the same questions, asked repeatedly.
For Heads of marketing and founders at B2B software, services and professional firms whose buyers ask AI assistants for a shortlist before they contact anyone.
You need this when
- 01
Buyers ask ChatGPT for a shortlist, and you’re not in the answer.
- 02
When an assistant does mention you, it gets what you do or who you serve wrong.
- 03
Competitors are cited from review sites and comparison articles you’ve never seen.
What’s included.
Buyer question set
A fixed list of the questions buyers ask at each stage, agreed with you, which becomes the measure for the work.
Baseline across assistants
Each question run in ChatGPT, Perplexity, Gemini, Claude, Copilot and Google AI Overviews, recording who is named, what is said and which sources are cited.
On-site changes
Plain statements of what you do, for whom and where, plus comparison, pricing and FAQ pages, structured data and AI crawler access set in robots.txt.
Off-site footprint
Consistent company descriptions on directories and review sites, corrections where facts are wrong, and data or expert comment offered to the publications the assistants cite in your category.
Repeat tracking
The question set rerun every month in each assistant, several times per question, with changes logged against the baseline.
Referral tracking
Visits from chatgpt.com, perplexity.ai and other assistants tracked in GA4, and AI assistants added as an answer to your forms’ source question.
How it runs.
Question set and baseline in a Sprint of 1–4 weeks; after that, we report trends over months rather than single runs.
- 01 · Questions
We agree with you the buyer questions that matter, from first problem to shortlist.
- 02 · Baseline
We run each question across the assistants and record who is named and which sources are cited.
- 03 · Fix on site
We rewrite and add the pages and structured data that answer those questions plainly.
- 04 · Fix off site
We correct and complete your listings and offer material to the sources the assistants rely on.
- 05 · Track
We rerun the question set each month and report what changed and what we’ll do next.
Deliverables
- Buyer question set
- Baseline report: mentions, accuracy and cited sources per assistant
- Source gap list: where competitors are cited and you aren’t
- On-site changes, made or briefed
- Off-site action list with progress log
- Monthly tracking report
- AI referral tracking in GA4
Works with
What buyers ask.
Can you promise ChatGPT will recommend us?
No. Nobody controls what an assistant says. We can make a mention more likely by fixing what the assistants read, and we show the effect on the same questions over time.
How is this different from SEO?
SEO aims for a ranked link a buyer clicks. GEO aims for your name, described correctly, inside the answer itself. The site work overlaps, but more of GEO happens off your site, on the sources the assistants read.
Should we let AI crawlers read our site?
It’s your decision, and each crawler is a separate choice. Blocking training crawlers such as GPTBot keeps your pages out of model training. Blocking search crawlers such as OAI-SearchBot or PerplexityBot can keep you out of live answers. We list each one, what it does and what we recommend.
Pairs well with
Offers, landing pages and campaigns aimed at the companies you want to sell to.
Let’s build your pipeline.
Tell us what you sell, who buys it and roughly what a customer is worth to you. We reply with questions, then a written proposal.