Why AI Assistants Check Your SRA Registration Before Recommending Your Firm

The short answer: when someone asks an AI assistant for a solicitor, the assistant first has to satisfy itself that the firms it is about to name are genuinely regulated. "Solicitor" is a protected title in England and Wales — only SRA-regulated people and firms can use it — so a recommendation is only safe if the firm can be matched to the regulatory record. The assistant does this by comparing what your website says about you against authoritative public data, chiefly the SRA register and Companies House. Where those sources agree, it is confident enough to recommend you. Where they conflict — a trading name that does not match your registered name, an out-of-date address, practice areas the register does not reflect — it cannot safely confirm you are the regulated firm the user asked for, so it hedges or drops you. None of this is visible to you, and there is no notification when it happens.

This is one of the most common, most fixable reasons a competent UK law firm never appears in AI recommendations. Below is how the check actually works, the four mismatches that cause exclusion, and what to do about each.

"Solicitor" is a regulated query, not a generic one

When a buyer types "best property solicitor in Cardiff" into Google, they get a list of links and decide for themselves. When they ask ChatGPT, Claude or Perplexity the same question, the assistant returns a small number of named firms as recommendations — and naming an unregulated business as a "solicitor" is a factual error the model is trained to avoid.

That changes the assistant’s behaviour. It is no longer matching keywords; it is trying to confirm that each candidate is a real, identifiable, regulated entity before it will put its name in an answer. The SRA register is the authoritative source for that confirmation. Your firm’s relationship to that register — how cleanly your public information maps onto it — is therefore a gating signal, applied before the quality of your website content is ever considered.

How AI actually uses the register — entity resolution, not a live lookup

It is worth being precise here, because the popular shorthand is misleading. Most AI assistants do not call the SRA’s database in real time on every query. What they do is build a confident picture of your firm — an entity — from everything they have seen about you across the public web: the register, Companies House, your own website, legal directories, review sites. The more these sources agree, the more confident the model is that you exist, that you are regulated, and that you do what you claim. That confidence is what earns the recommendation.

For browsing-enabled answers — Perplexity, Google AI Overviews, and ChatGPT when it searches — register-derived sources can also be retrieved and cited directly in the response. In both cases the principle is the same: consistency between your public information and the regulatory record is the trust signal. Inconsistency is the failure mode.

We call this regulated-entity resolution: the process by which an AI assistant decides whether the firm in front of it is the same regulated firm named in the official record. Firms that resolve cleanly get recommended. Firms that resolve ambiguously get hedged.

The chain looks like this — four authoritative sources feeding a single resolution check, with one binary outcome on the other side:

The four mismatches that get firms excluded

In our analysis of UK regulated firms, the same four discrepancies account for the large majority of avoidable exclusions. Each is a gap between what your website tells an AI assistant and what the SRA register tells it.

1. Trading name versus registered name

Your firm trades as "Cardiff Property Law" but is registered with the SRA and at Companies House as "C P L Solicitors Limited." To a human this is obviously the same firm. To an AI assistant building an entity, it can read as two weakly-connected things — neither of which it can confidently recommend. The fix is to state the registered name and the SRA number plainly on your site and link the trading name to it, so the two resolve to one entity.

2. Address and contact inconsistency (NAP)

Name, address and phone number that differ between your website, your Google Business Profile, the SRA record and the directories you appear in is the single most common consistency failure. An old office address on one source and a current one on another forces the model to choose which to trust — and uncertainty pushes you down or out. Every public mention of your firm should carry the same address and phone number, matching the register.

3. Practice areas the register does not support

If your site leads with "specialist medical negligence solicitors" but your SRA record and wider footprint give no signal that you practise in that area, the assistant has no corroboration for the claim — so it will not stake a recommendation on it. Make sure the practice areas you want to be recommended for are stated clearly, consistently, and in a form the register and your other sources reinforce.

4. Regulated status that is not machine-readable

Many firms display "Regulated by the Solicitors Regulation Authority" as an image, a footer logo, or buried text with no SRA number nearby. A person sees it instantly; a machine often cannot extract it. Your SRA number, in plain text and ideally in structured data, is what lets an assistant connect your website to the register with confidence. Without it, you are asking the model to take your regulated status on trust — which, for a recommendation it has to stand behind, it will not do.

Why you cannot see this happening

The hardest part of this problem is that it is silent. When a buyer asks an AI assistant for a solicitor and your firm is excluded on a consistency check, there is no impression logged, no bounce, no analytics event — nothing in any dashboard you own. The enquiry simply goes to the two or three firms the assistant could confirm, and you never know the conversation took place. A firm can rank well on Google, have a strong reputation, and still be quietly filtered out of AI recommendations because its registered name and its trading name have never been connected in a way a machine can read.

How to pass the check

The work is unglamorous and entirely within your control:

• State your registered name and SRA number in plain text on your site, and connect your trading name to them.
• Make your name, address and phone number identical across your website, Google Business Profile, the SRA record and every directory you appear in.
• Express the practice areas you want recommendations for clearly and consistently, reinforced across your public footprint.
• Add structured data (schema) that encodes your firm as a regulated LegalService entity with its SRA number, so assistants can resolve you to the register without guessing.

Done properly, this turns four sources of doubt into four sources of confirmation — and moves you from the "cannot safely recommend" pile into the "named firm" pile.

The scale of the gap

The opportunity here is large precisely because so few firms have closed it. TendorAI maintains a dataset of more than 63,000 UK regulated firms drawn from the official registers — SRA, ICAEW, FCA and Propertymark. Within the SRA segment, 1,458 registered law firms have no website at all, which makes them effectively invisible to every AI assistant. Thousands more have a website but carry exactly the consistency gaps above. In most towns, the firms that get recommended are not the best firms — they are the firms whose regulatory data an assistant can confirm.

That is a fixable position, and being early to fix it is the advantage.

TendorAI is the AI visibility platform for UK regulated professional services firms. We maintain verified profiles for more than 63,000 SRA, ICAEW, FCA and Propertymark-registered firms, install structured data on your own domain, and track how AI assistants describe and recommend you. Check your firm's AI visibility score.