GEO and AEO for UK law firms. Generative and answer engine optimisation.
Generative engine optimisation and answer engine optimisation are the same job under two labels: getting your firm named and cited when a client asks ChatGPT, Google’s AI Overviews, Gemini, Claude or Perplexity to recommend a solicitor. I measure where your firm appears across those assistants now, then fix the pages, the markup and the proof that decide it.
What is the difference between GEO, AEO and SEO?
GEO and AEO describe the same work, and most of that work is SEO. Generative engine optimisation is the academic term for influencing what a generative model produces. Answer engine optimisation came out of featured-snippet practice and means writing so a machine can lift a direct answer. Both come down to being the clearest, best-evidenced source on the question a client asked.
Google is on record about the gap between the two. Its own documentation on AI features and your website states that “there are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary”, and that you do not need new machine-readable files or markup for them.
So when someone sells you GEO as a separate discipline with a separate retainer, ask which part of it is not SEO for solicitors. The honest answer is the measurement, and a shift in what the work is aimed at. The techniques are the ones that earned featured snippets, pointed at systems that quote you instead of linking to you.
How do I find out whether AI assistants are recommending my firm?
You ask them, on a fixed set of questions, on a schedule, and you write down what comes back. There is no dashboard that reports this reliably, so the measurement has to be built rather than bought.
The set I run holds the questions a client would actually type. Some are decision questions: best conveyancing solicitors in Manchester, who should I use for a commercial lease dispute. Some are check questions on the firm’s own name. Each one goes to ChatGPT, Google’s AI Overviews and AI Mode, Gemini, Claude and Perplexity, and each answer is recorded two ways: whether the firm was named, and whether its site was cited as a source. Those are different results and they have different fixes.
One reading is not a trend. Assistants give different answers to the same question on the same day, so a firm that appears in three of six answers this week and two of six next week has not moved. That is why the measurement is a share read over months, and why I will tell you when a change is inside the noise rather than presenting it as progress.
What actually decides whether an assistant cites your firm.
Six things do most of the work, and they hold across AI Overviews, ChatGPT, Gemini and Perplexity.
A complete answer in the opening 40 to 60 words
Before any context. If a system could lift your first paragraph and use it as its whole response, it will.
Sections that stand on their own
Each heading has to make sense to someone who has not read the rest of the page.
Structure a parser can walk
Headings in order, short paragraphs, numbered lists for processes, tables for comparisons.
Specific sourced claims instead of adjectives
A number with a link behind it beats a paragraph of confident prose.
A visible date that matches the markup
A date that agrees with the page’s own structured data, and figures that are current.
One primary subject per page
Named the same way throughout. A page about three things is cited for none of them.
None of that is new. It is the clarity, structure and evidence that always earned featured snippets. What has changed is the cost of ignoring it, because the systems that used to send you a click now answer the question themselves.
Do all the AI assistants work the same way?
No, and the difference is the most misunderstood part of this work. Being cited by one earns you nothing with the others, because they retrieve from different places.
I can show you that from my own site rather than from a case study. My main service page is cited in Google’s AI Overview for “seo for solicitors” on most of the captures I have taken. On the checks I have run, ChatGPT has not named it for the same query. Nothing about the page explains the gap, because the page is the same page. The retrieval is different.
What follows for your firm is that “get us into AI search” is not one job. Google’s AI features read the live web and lean on what already ranks, so they respond to the work on your own site. ChatGPT leans on Bing’s index and on third-party sources: directories, trade press, review platforms, forums. Presence off your own site matters more there. Perplexity cites heavily and visibly, so it is the cheapest place to see whether your material is being picked up at all. I work them separately and report them separately, and the platform pages set out what each one responds to.
Is this work compliant with the SRA Code of Conduct?
Yes, and the compliance question is narrower than firms expect. What an assistant says about your firm is not your publicity, but what it quotes from your site is, and Rule 8.8 applies to it exactly as it applies to the page itself.
Two things follow in practice. Claims have to be ones you can stand behind in the form a machine will lift them, which rules out the comparative superlatives some firms put in a hero and nobody reads too closely — an assistant will quote them as fact, attributed to you. And reviews and testimonials used as proof have to be genuine, attributable and properly handled, because a fabricated one quoted by ChatGPT is still a fabricated one.
I flag anything on a page that a regulator could object to before it ships, and I would rather lose a citation than put a firm in front of the SRA for a sentence I wrote. If you want the compliance position established before any work starts, that is what the free SRA-compliant AI visibility audit is for.
What this engagement actually does.
The work divides into four parts, and most firms need them in this order.
Measurement first
The prompt set is built for your practice areas and locations, run against every assistant, and recorded as a baseline before anything changes. Without that you cannot tell later whether the work did anything.
Then the pages
The answer in the first 60 words, sections that stand alone, the comparisons a client is actually making, and the questions they ask phrased the way they ask them. This is where most of the movement in Google’s AI features comes from.
Then the machine-readable layer
Complete and valid Organization, Service, FAQPage and LocalBusiness markup, checked against schema.org and, where Google offers a rich result for it, in Google’s Rich Results Test, with the visible page and the structured data saying the same thing. Not because markup earns citations on its own, but because disagreement between the two is a reason to distrust the page.
Then the proof
Real figures, a credential a reader can check, an author who exists, and client evidence that is genuine and handled properly. Legal services are judged harder than most subjects by every system in this list, so a claim a reader can verify carries more weight than one they have to take on trust.
How long before anything changes?
For Google’s AI features, weeks rather than days, and only after the page has been recrawled. For ChatGPT, longer and less predictably, because the change has to reach a different index and then be retrieved.
The honest version is that AI visibility work has a slower and noisier feedback loop than ranking work. Positions move in Search Console and you can see them. Citation share moves in a measurement you have to run yourself, on answers that vary between runs. I plan on a quarter before the direction is readable, and I will not tell you a fortnight’s movement means anything.
What it costs, and who it is for.
This is part of the monthly senior retainer, which covers SEO and AI search visibility as one piece of work, rather than a separate line with its own fee. Current rates and what each level includes are on the pricing page.
It suits firms that already rank reasonably and are not being named when a client asks an assistant instead of typing a search. It does not suit a firm whose site cannot be found at all yet. That is ordinary SEO for solicitors and it comes first. If search is only part of a wider question about where the marketing budget goes, start with law firm marketing.
GEO and AEO FAQs.
FAQ 1Is GEO just SEO with a new name?
Mostly, yes. Google’s own documentation says no special optimisation is needed for its AI features. What is genuinely new is the measurement, because the old instruments do not read citations, and the aim, because a cited answer can serve the client without ever sending you the click.
FAQ 2Can you guarantee my firm will be cited by ChatGPT?
No, and nobody can. There is no submission process, no paid placement and no ranking factor list. What I can do is measure where you stand now, fix the things that demonstrably decide it, and show you the change over time.
FAQ 3Our marketing agency says GEO needs its own budget. Are they wrong?
Ask them which activity in the proposal is not SEO. If the answer is measurement across assistants and work on third-party sources, that is fair and it is real work. If it is rewriting pages and adding schema, you are being quoted twice for one job.
FAQ 4Do we need an llms.txt file?
No. Google states plainly that no new machine-readable or AI text files are needed for its AI features, and I have not seen evidence that any major assistant depends on one. This site has one because it costs nothing to maintain. It is not the reason a firm is or is not being cited.
FAQ 5Does any of this help if a client never clicks through?
It is the point rather than a side effect. A partner who asks an assistant for a recommendation and hears your firm named has been referred, whether or not they visit the site. That is why the measurement records naming and citation separately, and why enquiries matter more here than traffic.
FAQ 6We are a small firm. Can we compete with the national names on this?
Often more easily than in ordinary search. Assistants answer a lot of narrow, local and practice-specific questions where the national firms have nothing specific to say, and a well-evidenced page from a firm that actually does the work in that town can win the answer outright.
FAQ 7How is this different from the AI visibility audit?
The audit is free and one-off: a report on where you stand across the assistants, what is holding you back and what to do about it, turned round in 72 hours. This is the ongoing work that acts on it and keeps measuring.
FAQ 8Should we be doing this instead of ordinary SEO?
No. Google’s AI features read from what already ranks, so the ranking work is the foundation rather than the alternative. A firm that abandons SEO for solicitors to chase AI citations usually loses both.
Related work.
AI search visibility
What each assistant responds to, platform by platform.
SRA-compliant AI visibility audit
A free one-off picture of where the firm stands, in 72 hours.
Schema markup for law firms
Structured data that agrees with the visible page.
SEO for solicitors
The ranking work the AI features read from.
Law firm marketing
Where search sits in the wider marketing question.
Content strategy for law firms
Practice-area content planned around real search demand.
Want to know whether AI assistants are naming your firm?
Book a 30-minute intro call and I will run three prompts for your practice area before we speak, so the conversation starts from what the assistants actually say about you rather than from a pitch.
Or start with the free SRA-compliant AI visibility audit, turned round in 72 hours, for the full picture across every platform.