Who gets cited in AI Overviews, and does it match who ranks?

01. AI Overviews and “near me” did not appear together

Twelve of the fifteen queries returned an AI Overview. Three did not, and the split was not random. All three “near me” queries returned a local pack and no AI Overview. All twelve research queries returned an AI Overview. Three out of three, and twelve out of twelve.

On this sample, Google does not put an AI Overview in front of someone who has already decided to hire a solicitor and is looking for one nearby. It puts a map.

This is the finding with the clearest commercial consequence, so it is worth being blunt. If your enquiries come from people searching “family solicitor near me”, the AI Overview is not currently taking that click. The local pack is the surface that matters, which means Google Business Profile work, reviews and proximity are still doing exactly the job they were doing two years ago.

If your enquiries come from people researching cost, process, or whether they need a solicitor at all, the AI Overview sits above everything and answers the question before anyone scrolls.

Most firms have both. The point is that they are different problems with different fixes, and treating “AI is eating our traffic” as one undifferentiated worry sends the budget to the wrong place.

02. One in four cited sources is not in the organic top ten

Across the twelve AI Overviews I could see seventy-one cited domains. Fifty-two also appeared in the organic top ten for the same query. Nineteen did not. Twenty-seven percent of visible citations came from pages a searcher scrolling the normal results would never reach.

The rate swings hard by query. Three of the five cited sources on “no win no fee solicitor uk” do not rank in the top ten. On “executor duties uk” and “do i need a solicitor to make a will”, every cited source also ranks.

What separates them is what kind of source the AI Overview reaches for. The queries where citation and ranking match are the ones citing institutions — gov.uk, Age UK, the Law Society, The Gazette, MoneySavingExpert, Legal and General. Those bodies rank anyway, so there is nothing to diverge. The queries where citation and ranking come apart are the ones citing individual law firms, and those firms are frequently not the ones winning the organic results.

That is worth sitting with if you run a firm that has given up on outranking the national brands.

03. The citation mix flips on the type of question, not the practice area

I expected practice area to predict who gets cited. It does not. The question does.

Seven private client queries, all probate or wills, sorted by how many of the six visible citations were law firms rather than institutions: “do i need a solicitor for probate” six firms and no institutions. “How long does probate take uk” six and none. “Do i need probate if there is a will” five and one. “What does a probate solicitor do” five and one. “How much does a will cost uk” four and two. “Do i need a solicitor to make a will” two and four. “Executor duties uk” two and four.

A clean gradient inside a single practice area.

Questions that imply engaging a professional are answered mostly from law firm websites. On “do i need a solicitor for probate”, every visible cited source was a firm: DMH Stallard, Bishop’s, Evolve Family Law, the Probate Bureau, DMA Law and Welland Valley Legal. gov.uk ranks first organically for that query and was not among them.

Questions that are purely definitional are answered mostly from institutions. On “executor duties uk”, four of six were gov.uk, Age UK, The Gazette and Legal and General.

The same pattern shows on “how long does probate take uk”, where gov.uk holds organic positions one and four and appears in none of the visible citations, while six commercial legal sites do.

This revises something I published in May. The UK Law Firm AI Visibility Report found firms taking under a third of citations. Three months later, on decision-shaped questions, firms take most of them. I would not read one report against another as a trend without re-running the original method, but the direction is the opposite of what the May data showed, and it is worth saying so plainly rather than quietly updating the number.

If your enquiries come from “near me” searches, nothing has changed yet. Keep doing local properly. Reviews, proximity, categories, a maintained profile. The map pack is intact and the AI Overview is not competing for that click.

If you publish content answering research questions, the AI Overview is both your competition and your opening. It will answer the question whatever you do. The citation is the only thing left to win.

And the citation is winnable without the ranking. Twenty-seven percent of cited sources were not in the top ten, and on some queries it was three in five. A firm that cannot realistically outrank Irwin Mitchell or Slater and Gordon can still be one of the sources the answer is built from. Smaller firms are already doing it: Welland Valley Legal, Bishop’s, Scott Bailey, Pellys, Poole Alcock and Heald Law all appear as cited sources without a corresponding top-ten ranking.

Write for the question that implies hiring someone. The clearest pattern in the data is that “do I need a solicitor for X”, “what does an X solicitor do”, “how long does X take” and “how much does X cost” are answered mostly from law firm websites. Those four question shapes, for your own practice areas, are where a firm’s content has the best chance of being the source rather than the also-ran.

The catch is that a citation is not a visit. It puts your name in the answer a prospect reads. That is worth something, it is not worth the same as a click, and any consultant telling you otherwise is selling.