Why does ChatGPT recommend your competitor?
AI answers often name just a few businesses, like the real ones below. Every month we run your buyers' questions in Perplexity and Google AI Overviews, 3 times each, and check up to 10 of them in ChatGPT and Gemini by hand where possible. You get a report: who was named instead of you, which pages the answers cite, and what changed since last month, with the raw answers.
We can't promise an AI will name you. We can show you, every month, whether it does.
The free ChatGPT check is one question about your area, asked once. We send you the full answer as a sample, not a measurement.
Real answers, captured 26 Sep 2026. Each one names a handful of businesses. The rest are not in the answer.
Fewer clicks reach any website. AI keeps the answer.
Lower average click-through rate for the No. 1 organic result once an AI Overview appears.
Of 3,000 sites in the study got at least one visit from an AI chatbot; Perplexity sent just under a third of that AI traffic.
Google searches that led to a click on a normal result, with an AI summary shown vs without one (900 US adults, March 2025).
Of B2B buyers already have a preferred vendor before they first talk to sales.
Whether AI names a business decides who gets the customer
ChatGPT became Tally's No. 1 referral source, with over 2,000 new users signing up via AI tools that they could track.
Of new Vercel sign-ups came from ChatGPT in one month of 2025, up from 1% six months earlier.
Brands well known to UK buyers that Perplexity never named across six runs of two buying questions in their category (one account, 2 Oct 2026).
New to GEO? GEO explained in 2 minutes, in diagrams.
Four questions about GEO
One paragraph each. The first sentence is the answer.
What is GEO?
GEO (Generative Engine Optimization) is the work of getting a brand named, cited or recommended in answers from AI engines such as ChatGPT, Perplexity, Gemini, Google AI Overviews and Claude. The term comes from a 2023 research paper by Aggarwal et al. In practice it covers three things: whether AI crawlers can read your pages, whether those pages state facts clearly enough to quote, and how other sites describe you.
How is GEO different from SEO?
SEO aims for a high position in a list of links; GEO aims for a mention or citation inside a written answer. The foundations overlap: Google says there are no additional requirements to appear in AI Overviews beyond being indexed and eligible for a snippet. The differences are what you measure (answers, not rankings) and how much third-party descriptions of your brand matter.
How do I get my brand cited by ChatGPT?
Let ChatGPT's search crawler read your pages, then make those pages the clearest answer to your buyers' questions. OpenAI documents OAI-SearchBot as the crawler that surfaces sites in ChatGPT search, separate from GPTBot, which collects training data (OpenAI documentation). After access: answer-first pages, tables, dated facts, and correct descriptions of you on the sites ChatGPT already cites. Full guide.
How do you measure AI visibility?
Run a fixed set of buyer prompts on each AI engine 3 times, and count how often the answers name your brand, cite it, and recommend it compared with your competitors. Record the date, the engine, the surface, and the market for every answer, report each engine separately, and mark a break in the trend when a model is updated. Our method.
The Citation Loop. Five steps, every month.
Each cycle ends with a measurement. That measurement sets the next cycle's work. Full methodology.
- 01
Probe
Run your buyers' questions through Perplexity and Google AI Overviews, 3 times each. Record who is named and who is cited.
- 02
Diagnose
List the sources the engines cite for your topic. Compare what they say about you with what is true.
- 03
Build
Fix crawl access and page structure. Publish definitions, clear comparisons, and facts an engine can quote.
- 04
Earn
Get your brand described correctly on the third-party sites the engines already cite.
- 05
Re-measure
Run the same prompt set again. Compare with the baseline. Pick the next cycle's priorities.
- 01
Generative Engine Optimization
Visibility in AI answers, measured in Perplexity and Google AI Overviews.
- 02
Answer Engine Optimization
Pages where one passage answers one question in full.
- 03
SEO and technical SEO
Crawl access for search and AI bots, rendering, indexation, speed, structured data.
- 04
Content for citability
Definitions, comparison tables and dated facts an engine can quote without changing the meaning.
- 05
Entity and brand signals
The same facts about your brand on your site, profiles, directories, and the review sites that mention you.
How we measure AI visibility
The full method is public. You can check it before you talk to us.
| Metric | What it counts |
|---|---|
| Mention rate | Answers that name your brand, divided by all answers. |
| Citation rate | Answers that link to or list your domain as a source, divided by all answers. |
| Share of voice | Your mentions divided by mentions of you plus your tracked competitors. |
| Accuracy | Mentions where the facts about you are correct, divided by all mentions. |
Each prompt runs 3 times per engine, in a clean session with a fixed language and region. One run is not enough: the same prompt can name different brands on different runs. Read the measurement method.
GEO for your trade.
- 01
GEO for accountants: what Perplexity cites in the UK
- 02
GEO for architects: what Perplexity cites in the UK
- 03
GEO for conveyancing solicitors: what Perplexity cites in the UK
- 04
GEO for physiotherapists: what Perplexity cites in the UK
- 05
GEO for private dentistry: what Perplexity cites in the UK
- 06
GEO for solicitors: what Perplexity cites in the UK
- 07
GEO for vets: what Perplexity cites in the UK
No logo wall. Open methods instead.
No logo wall: we are a new agency and we withhold client names. What you can check instead: our methods, and anonymised case studies with a source behind every figure.
- 01
The audit method
Every step, prompt type and platform. Audit method
- 02
The measurement method
What counts as a mention, how often we run prompts, what happens when a model changes. Measurement method
- 03
This site
AI crawlers allowed in robots.txt, schema.org markup on every page, a full llms.txt.
- 04
Small clients
A local business gets the same method as a large brand, with a smaller prompt set.
Common questions
Can you guarantee that ChatGPT or Perplexity will mention my brand?
No. No agency controls what a model outputs, and anyone who guarantees placement is guessing. We improve the signals AI engines rely on and measure the change with a published method, so you can judge the result yourself.
How long does it take to see results?
It depends on the engine. Engines that search the web live can reflect changes once your pages and third-party mentions are recrawled. What a model knows from training changes only when the model is retrained. We record a baseline first and report the change every month, including when nothing has moved.
How many prompts do you track, and how often?
The prompt set is built from your customers' real questions and agreed with you before the first run. Its size depends on your products, markets, and the languages you sell in. Defaults: the free audit uses 10 prompts, each run 3 times on Perplexity and Google AI Overviews, which gives 60 recorded answers; the paid audit uses 30 prompts, 180 answers. Monitoring and the monthly plans rerun the 30 agreed prompts once a month, 3 times each on Perplexity and Google AI Overviews, and check up to 10 of them by hand in ChatGPT and Gemini.
What do you do when an AI model is updated?
We re-run the full prompt set as soon as we detect the update, mark a break in the trend line so results before and after are not compared as if nothing changed, and review which sources the new model cites.
Do you work with small businesses?
Yes. The method is the same for a local business and a multinational; only the size of the prompt set and the number of markets change.
What is llms.txt, and do I need one?
llms.txt is a proposed convention (see llmstxt.org): a Markdown file at the root of a site that summarises it for language models. It is not an adopted web standard and AI engines do not all read it. We treat it as a cheap, low-risk addition, not a lever. Ours is at /llms.txt.
Find out what AI answers say about you and who they cite instead.
We run your category's buying questions through Perplexity and Google AI Overviews and show you where you appear, where competitors appear, and what to fix first.