# Scotland's biggest law firms have cracked SEO. Almost none are ready for AI search.

> Canonical: https://deliverbetter.co.uk/insights/ai-search-scotland-law-firms/
> Published: 2 Oct 2026. Last updated: 2 Oct 2026.
> Author: David Bain, Deliver better.

## In short

Search is starting to answer for people instead of sending them to you. We checked Scotland's ten largest law firms: nearly all have classic SEO sorted, but almost none have done the newer work that gets a firm named in an AI answer. The fixes are small, and the ground is open.

222 The front door moved
 

 
 For twenty years the job was to rank on Google. Get the page right, earn the links, climb the results. That still matters, but it is no longer the whole game.

People now ask a question and take the answer. “Do I need a solicitor to make a will in Scotland?” goes into ChatGPT, Gemini or Google’s AI Overview, and a written answer comes back with a few firms named in it. Most people never scroll to the ten blue links underneath. If your firm isn’t in that answer, you weren’t in the room.

That’s the shift. Search used to send people to your website. Increasingly it reads your website, decides whether you’re worth mentioning, and answers on your behalf. The firms that get mentioned aren’t always the biggest or the best. They’re the ones whose pages a machine can read, understand and trust. That’s the hard bit, and it’s where almost everyone is exposed.

 

 

 

 
 
 
 
222 What we checked
 

 
 We looked at the ten largest law firms headquartered in Scotland: Brodies, Burness Paull, Shepherd and Wedderburn, Morton Fraser MacRoberts, Harper Macleod, Anderson Strathern, Lindsays, Thorntons, Gilson Gray and Aberdein Considine. We deliberately left out the big UK and global firms with Scottish offices, like Pinsent Masons, CMS, DLA Piper and Addleshaw Goddard, because their digital estates are run from elsewhere.

For each firm we checked the homepage and one core service page, the kind of commercial page that actually wins work. We weren’t grading the writing. We were looking at the things that decide whether an AI engine can read a page and lift it into an answer: structured data (the schema that tells a machine “this is a legal service, here’s who provides it”), whether there’s any question-and-answer content, how much real content is on the page, whether that content is even visible to a crawler that doesn’t run JavaScript, and whether the site gives AI crawlers any guidance at all through an llms.txt file. All observed from the outside, so this is what a search engine or an AI model sees too.

 

 

 

 
 
 
 
222 What we found
 

 
 The good news first. The basics are largely sorted. Page titles, meta descriptions and heading structures were in decent shape nearly everywhere. This isn’t a sector that’s bad at SEO. The gaps sit almost entirely in the newer layer, the one built for AI answers rather than blue links.

 

 

 

 
 
 
 What we checkedFirms passingClassic on-page SEO (title, meta, headings)9 / 10Service content a non-JavaScript crawler can read8 / 10Service-level schema (tells a machine “this is a legal service”)3 / 10Any question-and-answer content on the service page1 / 10FAQ / answer schema (the format AI engines lift most readily)0 / 10Guidance for AI crawlers (an llms.txt file)2 / 10

 

 

 

 
 
 
 A few things stood out. Not one of the ten marks up its content with FAQ schema, the single format AI engines and Google most readily pull into an answer. That’s an open goal. Content depth swung wildly too, from service pages carrying four thousand words of genuinely useful detail down to one with barely seventy. And two well-known firms load their service content in a way that a crawler which doesn’t run JavaScript sees as a near-blank page, so the tools people now ask effectively can’t read them.

The firms doing it well show it’s not hard. Harper Macleod pairs deep content with proper legal-service schema and real question-and-answer sections, which is exactly why it turns up when you ask an AI “do I need a solicitor to make a will in Scotland”. Brodies and Morton Fraser MacRoberts are the only two giving AI crawlers any guidance at all. Everyone else has simply left the new front door shut, mostly without realising it’s there.

 

 

 

 
 
 
 
222 Five things worth fixing
 

 
 None of this is a rebuild. It’s a handful of changes that move you from invisible to quotable.

 

 

 

 
 
 
 *01*Answer the real questions in plain question-and-answer form, because that plain Q&A is the content an AI lifts into its answer.

*02*Mark it up so a machine understands it: FAQ schema on those questions, and your service pages tagged as a legal service with the people who provide it.

*03*Give the money pages real depth, because a seventy-word service page can't rank and can't be cited.

*04*Check a crawler can actually read the page: if the content only appears after JavaScript runs, many AI crawlers see a blank page, so server-render the important stuff.

*05*Open the door to AI crawlers with an llms.txt, and check your robots.txt isn't quietly blocking the likes of GPTBot, ClaudeBot and Google-Extended.

 

 

 

 
 
 
 
222 If you're not cited, you can still buy your way in
 

 
 Getting cited organically is the goal, and it’s worth the most, but it takes time. The useful news is that the AI tools are now opening up paid placement, so there’s a faster route while the organic work beds in.

Google is the most straightforward. It’s placing ads inside and directly beneath its AI Overviews, and they run through the Google Ads account you already have, with the search and shopping formats you already use. If you’re advertising on Google today, appearing alongside AI answers is largely a matter of settings, not a new platform.

The others are moving quickly. OpenAI has been building advertising into ChatGPT, with sponsored units kept separate from the answer itself, and Perplexity ran its own sponsored-placement programme. The picture changes month to month, so the right move is to treat this as a live, emerging channel: see what’s actually available in your market now, test small, and watch where your clients are asking their questions.

The principle holds either way. Organic citation earns you the mention. Paid placement buys you the slot while you earn it. The firms that win will do both, deliberately, rather than waking up to AI search in a year’s time.

 

 

 

 
 
 
 
222 Where to start
 

 
 Start by finding out what the tools say about you now. Ask ChatGPT, Gemini and Google the ten questions your best clients would ask before they hire a firm like yours, and see who gets named. If it isn’t you, that’s your brief.

Then fix the money pages first. Real answers to real questions, marked up so a machine understands them, on pages a crawler can actually read. It’s not a huge job, and right now almost nobody in the sector has done it, so the ground is there to take.

 

 

 

 
 
 
 
222 Questions law firms ask about AI search
 

 
 What's the difference between SEO and AI search?SEO gets your page into the ranked list of blue links. AI search is when a tool like ChatGPT, Gemini or Google's AI Overview reads the web and writes a single answer, naming a few firms in it. You can rank well on Google and still never appear in that answer, because the two jobs depend on different things.

Why would a firm rank on Google but not show up in AI answers?Classic SEO is largely about ranking: titles, headings, links. Getting named in an AI answer depends on whether a machine can read, understand and trust the page: structured data, real question-and-answer content, enough depth, and content a crawler can read without running JavaScript. Most firms have the first and not the second.

What is schema, or structured data, in plain terms?It's a bit of hidden code that labels what's on a page, so a machine knows ‘this is a legal service, provided by these people’ or ‘this is a question, and here's the answer’. Visitors never see it. AI engines and Google use it to understand your page and quote it with confidence.

What is FAQ schema, and why does it matter?FAQ schema is structured data that marks up your questions and answers so a machine can lift them straight into an answer. None of the ten firms we checked use it, which is why it's one of the easiest wins available right now.

What is an llms.txt file?It's a short file on your website that gives AI crawlers a bit of guidance about what's there and what matters, a little like a welcome note at the front door. It's new and optional, and almost nobody in the sector has one yet.

How do I check whether AI tools can actually read my website?Two quick checks. View one of your service pages with JavaScript turned off: if it goes near-blank, many AI crawlers can't read it. And check your robots.txt isn't quietly blocking AI crawlers such as GPTBot, ClaudeBot and Google-Extended. You can't be cited by a tool you've locked out.

Can we pay to appear in AI search?Increasingly, yes. Google is placing ads inside and beneath its AI Overviews, through the Google Ads account you already have, and OpenAI and Perplexity are building their own sponsored placements. Organic citation is worth more, but paid is a faster route while the organic work beds in. It changes month to month, so test small.

How do I find out whether my firm is being cited right now?Ask ChatGPT, Gemini and Google the ten questions your best clients would ask before hiring a firm like yours, and see who gets named. If it isn't you, that's your brief. It's not a huge job to fix: real answers to real questions, marked up so a machine understands them, on pages a crawler can read.

 

 

 

 
 
 
 
222 Want to know how your firm shows up in AI search?
 

 
 This is the kind of work we do at Deliver better, the practical side of AI and digital, for people who have actual jobs to do. If you want to know where you stand, or where to start, ping me a message or let’s grab a coffee.

 

 
 
 [
 
 Start a conversation
 
 ](/contact)
 

 
 [
 
 Try the free tools
 
 ](/free-tools)
