Insights · GEO
How to get your consulting or professional service firm recommended by AI: a practical GEO guide.
The short answer
AI recommends the firms it can understand, check and cite. So say the same true thing about who you are everywhere, answer the questions your buyers actually ask, get named by sources other than yourself, and let the crawlers in. Then measure whether AI names you, because nobody can promise that it will. That work is GEO, short for generative engine optimisation.
01What GEO is
GEO gets your firm named inside the answer.
GEO is the work of making sure AI assistants such as ChatGPT, Perplexity and Google's AI Overviews understand your firm correctly and name it when a buyer asks who can help. It covers what you publish, how consistently the facts about your firm and its experts appear across the web, and what credible third parties say about you.
SEO works to earn a ranked link. GEO works to get your firm named inside the answer. Google's own guide says optimising for its generative AI features is still SEO, since they run on its core ranking systems. What changes is the weight of everything that is not your website: AI reads what the web says about you as closely as what you say about yourself.
02What AI looks for
We watched what AI searches for. It goes looking for proof.
Before an AI engine answers, it runs its own follow-up searches. We track 86 questions that partners at consulting and professional service firms ask AI when they look for help winning clients, and over three days in October 2026 the engines ran 975 follow-up searches to answer them.
80
of the 975 searches added the word “official”.
Asked how AI search works, the engines went looking for Google Search Central, Bing Webmaster Tools and OpenAI's own crawler documentation, not for opinions about them.
Source Fimiliar's own AI search tracking, 6 to 8 October 2026.Base: 86 tracked buyer questions, 975 follow-up searches. Our sample, not an industry figure.
101
searches went straight to one named website.
Consultancies, directories and publications, searched by name. If your site does not answer the question, a named site that does gets read instead.
Source Fimiliar's own AI search tracking, as above.Same base.
A third pattern: more than 60 of the searches asked for case studies, results, benchmarks or reviews. AI wants evidence it can repeat, and “trusted advisor” gives it nothing.
Cat's reading: if retrieval keeps moving toward named, official sources, your own site matters more, not less. In the audits she runs, sites are usually crawlable and AI can tell who the firm is. They fail because they answer nothing a buyer would actually ask.
It's like giving AI a blank book and then expecting it to know who you are.
Cathleen Vermaak-Nisbet, co-founder and CEO, Fimiliar
03The checklist
Make your firm easy for AI to understand.
Five jobs, one goal. This is the list we work through on our own site, in the order Cat sorts GEO work.
- 01
Identity Say who you are, the same way everywhere.
- One “who we are” paragraph, used word for word on your site, LinkedIn page, directory listings and bios.
- The firm's name never travels alone: always with what you do and who for.
- Profiles linked to each other and to your site, under one spelling of the name.
- Organization and Person structured data that matches the visible page.
- 02
Access Let AI reach and read it.
- robots.txt allows Googlebot, Bingbot and OAI-SearchBot, the crawler behind ChatGPT search.
- Key pages indexed, and eligible to show with a snippet in Google.
- Answers in the page text, not only in images, PDFs or animation.
- Site verified in Google Search Console and Bing Webmaster Tools.
- 03
Answers Answer what buyers actually ask.
- A list of the questions your buyers put to AI, in their words.
- A page or section for each, with the answer first.
- Specifics AI can repeat: who it was for, how long it took, what changed.
- Dates kept current.
- 04
Mentions Get named by someone other than you.
- Listed in the directories and review sites your buyers check.
- Quoted in the press, podcasts and publications they read.
- Included in lists and comparisons, only on strengths you genuinely have.
- 05
Experts Put real names on the expertise.
- A named author, with credentials, on every article.
- Partners posting in their own words.
- One bio everywhere, and each expert mapped to distinct topics so two partners do not compete for one answer.
“Nothing you write counts if the AI can't actually get to it,” as Cat puts it. We run the same list on our own site: one description of Fimiliar, word for word wherever it appears, a page written for AI assistants, AI crawlers named in robots.txt, and every answer in the page text rather than hidden behind motion.
We keep an llms.txt file too. Cat rates it near-useless but harmless, and Google agrees from its side: its Search does not use the file, so it neither helps nor harms there.
04Before and after
A “who we are” paragraph AI can actually use.
AI checks a firm against five readings: who you are, what you do, what you know, who says you know it, and whether credible sources support it. Here is a made-up firm, before and after.
Before · unclear
We are a boutique, full-service consultancy that partners with ambitious organisations to drive transformational change and deliver lasting value through a people-first, innovative approach.
Who for? Doing what? Says who? There is no name, no category and nothing checkable, so there is nothing for AI to repeat. Swap the word “consultancy” and it describes a thousand firms.
After · clear
Quillmere Sourcing is a procurement consultancy for mid-sized UK manufacturers. We cut supplier costs and rebuild sourcing teams. The firm is led by a former head of procurement, publishes a yearly benchmark of supplier price rises, and is quoted in the trade press.
Every circle answers one of the five readings. Name and category together, a buyer and a job, the expertise, and proof someone else can confirm. A made-up firm, to show the shape.
Ours, for comparison: Fimiliar is a trust-led lead generation company for consulting and professional service firms. That sentence opens the same paragraph wherever we describe ourselves, word for word.
Changing it from page to page, Cat says, is “like telling your friends at a braai that you do this one week and then next week you do something else.” (A braai is a South African barbecue.) A thin picture makes AI guess: asked who Cat is, one follow-up search an engine ran was her name plus “biography or profile or researcher”.
05The rule
Don't lie to AI about who you are.
You can lie to it, and it will probably work in the short term. Long term, it will not. So don't lie to AI about who you are.
Cathleen Vermaak-Nisbet
AI cross-checks what you say against reviews, other people's sites and the awards your competitors actually hold. When it finds a gap, the way it describes you cools. A claim has to be true on every surface you put it on, including any list you place yourself in. Google says much the same: chasing inauthentic mentions is less helpful than it might seem.
06Beyond your site
Put the same expertise everywhere AI reads.
Buyers already ask AI before they ask you: 94% of business buyers use AI during buying, then check what it says with people they trust (Forrester, 2026, nearly 18,000 global buyers, across purchase types). The answer about your firm has to hold up in both places.
Cat ranks the surfaces AI weighs, weakest first. Your own website, because you own it and AI knows it. A founder's LinkedIn: still owned, but a person writing. Strongest is anything other people publish about you: PR, reviews, directories, podcasts, community. The more rungs carry the same true claim, the more AI trusts it.
- Turn talks into text. Give every podcast, webinar and video a page with a transcript, so the expertise exists in words.
- Keep formats in step. Text, images and video should describe the same people and services; Bing names this as a way to reduce ambiguity.
- Structure for reuse. Headings, tables, FAQ sections, data and cited sources, which Bing says help AI reference a page accurately.
- Keep a rhythm. Cadence feeds GEO, so a steady month beats one loud week.
At Fimiliar, GEO runs as part of one system with content, outbound and engagement, so what AI says about your firm matches what buyers hear everywhere else.
07Measure it
How to measure AI visibility.
Put your buyers' questions to AI on a schedule, and write down what comes back. Cat tracks four numbers.
- 01
Visibility
Out of a fixed set of buyer questions, how often AI names you at all.
- 02
Share of voice
When you are named, how much of the answer is you. Third of three beats third of ten.
- 03
Position
Where in the answer you land. The first three names are the ones a buyer reads.
- 04
Sentiment
How warmly AI describes you, and whether it has proof. It moves on reviews and case studies with real numbers.
| Where | What it shows | Keep in mind |
|---|---|---|
| A prompt tracker | Whether AI names you for your buyers' questions, who else it names, and the searches it runs first | A sample of your questions, not the whole market |
| Google Search Console | The Generative AI performance report: how people find your content through Google's AI features | Google says no third-party tool sees its internal systems |
| Bing Webmaster Tools, AI Performance | How often Copilot and Bing's AI summaries cite your pages, which pages, and the grounding queries behind them | In preview, sampled, and citations rather than clicks |
| Your first calls | Whether new buyers say AI pointed them to you | Only if you ask, every time |
We run the same tracking on Fimiliar, and it takes time: outside questions that already use our name, AI still rarely names us. Answers move, competitors move, and AI sometimes gets a fact wrong that needs correcting, so GEO is maintenance, not a project you finish. What the follow-up searches keep asking for, case studies, benchmarks and sales-cycle data, is a fair guide to what to publish next.
08Questions
What consulting and professional service firms ask about AI search.
What is generative engine optimization?
Generative engine optimisation, or GEO, is the work of making sure AI assistants such as ChatGPT, Perplexity and Google's AI Overviews understand your firm correctly and name it when buyers ask who can help. It covers what you publish, how consistently the facts about your firm and its experts appear across the web, and what credible third parties say about you.
How do you get a consulting firm recommended by ChatGPT?
Make your firm easy for it to understand, check and cite. Allow OAI-SearchBot, the crawler OpenAI uses for ChatGPT search, in your robots.txt. Use one true description of your firm everywhere. Publish specific, answer-first pages for the questions your buyers ask, with evidence AI can repeat, and earn mentions from credible third parties. Nobody can promise a recommendation; this makes one more likely.
How do you optimise for AI search engines?
Start with what the engines publish. Google says foundational SEO still applies: let it crawl, keep answers in text, and make structured data match the page; special AI files are not needed. Bing adds clear headings, tables, FAQ sections, evidence and fresh pages. Then make sure the rest of the web describes you the same true way.
How do buyers use AI to choose a consulting firm?
As one source among several. 94% of business buyers use AI during buying and check what it says with people they trust (Forrester, State of Business Buying 2026, across purchase types). In the 975 follow-up searches we saw AI engines run for 86 buyer questions, they kept asking for case studies, benchmarks, reviews and official sources. So a consulting or professional service firm needs evidence AI can find.
Do I need an llms.txt file to show up in AI answers?
No. Google says its Search does not use llms.txt, so adding one neither helps nor harms your visibility there. It is harmless to keep one, and Fimiliar does. The work that changes AI answers is elsewhere: one consistent description, pages that answer real buyer questions, and mentions from other people.
How do you measure visibility in AI search?
Put a fixed set of your buyers' questions to the AI engines on a schedule and record four things: how often you are named, how much of the answer is you, where you appear, and how warmly you are described. Add the Generative AI performance report in Google Search Console and the AI Performance report in Bing Webmaster Tools, which shows how often Copilot and Bing's AI summaries cite your pages.
Sources
- Google Search Central, "Optimizing your website for generative AI features on Google Search", updated 10 July 2026 Generative AI search is still SEO; no special files, markup or chunking needed; Search does not use llms.txt; inauthentic mentions; the Generative AI performance report; third-party tools.
- Google Search Central, "AI features and your website", updated 10 December 2025 No additional requirements for AI Overviews or AI Mode; pages must be indexed and eligible for a snippet; structured data must match visible text.
- Bing Webmaster Blog, "Introducing AI Performance in Bing Webmaster Tools Public Preview", 10 February 2026 Citations across Copilot, AI summaries in Bing and partners; grounding queries; guidance on structure, evidence, freshness and consistency across formats.
- Bing Search Blog, "New AI Visibility Insights in Bing Webmaster Tools: Intents, Topics, Citation Share, Compare", 16 June 2026 Citation share and the other preview additions to AI Performance.
- OpenAI, "Overview of OpenAI Crawlers" OAI-SearchBot surfaces websites in ChatGPT's search features; allow it in robots.txt.
- Forrester, "The State of Business Buying, 2026", 21 January 2026 The 94% using AI figure. Base: nearly 18,000 global business buyers, across purchase types.
- Fimiliar, own AI search tracking, 6 to 8 October 2026 86 tracked buyer questions and the 975 follow-up searches AI engines ran to answer them. Our sample, not an industry figure.
Next step
Make your expertise easier for AI to understand, trust and cite.
Thirty minutes on your firm and the clients you want to win, to see if you qualify.