UK Legal Services AI Impact Report 2026 | Somuna
How AI is changing client acquisition for UK law firms in 2026 – click-through rates, practice area risk, consumer trust, and PPC costs, with sources.
What this report is
This is Somuna’s analysis of how AI is changing the way people find and choose UK legal services. It draws together published research from Ahrefs, Pew Research Center, SparkToro, and consumer trust surveys, alongside Somuna’s own commercial interpretation of what these findings mean for law firm marketing.
This is a synthesis and analysis report, not a primary study. We have not independently surveyed law firms or consumers ourselves for this edition. Every statistic is sourced to its original study, listed in full at the end of this report, so you can verify and go deeper on anything that matters to your firm. Where original UK-specific primary research exists on this topic, we reference it directly and encourage you to read it in full.
We’ve built this because, in reviewing UK law firm websites through our own audit work, we kept having the same conversation with managing partners: traffic is falling, nobody’s quite sure why, and the explanations on offer are either too vague (“Google keeps changing things”) or too alarmist (“AI is replacing lawyers”). Neither is accurate. This report is our attempt at something more useful: a clear, sourced picture of what’s actually happening, and what to do about it.
Executive summary
AI is not replacing lawyers. Trust in AI for high-stakes legal matters remains very low — only 17% of people would trust AI to help with a divorce, and just 11% for criminal defence. What AI is doing is sitting between potential clients and law firm websites at the discovery stage, answering questions directly in search results and increasingly being asked directly to recommend a solicitor.
The scale of this transition is now well documented. AI Overviews correlate with a 58% reduction in click-through rates on top-ranking search results. Nearly 7 in 10 Google searches now end without any click at all. This is not a future risk. It is showing up in law firm analytics right now, and the firms noticing the sharpest declines tend to be the ones who haven’t yet started measuring their AI visibility specifically.
The risk is not evenly spread. It concentrates in transactional, lower-complexity work — wills, straightforward conveyancing, simple personal injury claims — where consumers are more comfortable with AI involvement. It’s structurally lighter in high-stakes, emotionally complex work — family law, criminal defence — where human judgment remains non-negotiable to the public, even though AI increasingly influences how those clients find their solicitor in the first place.
The response doesn’t require abandoning what already works. It requires building on it: monitoring AI visibility specifically (not just traditional rankings), strengthening the same E-E-A-T signals that support both traditional and AI-driven search, and treating this as an extension of good SEO rather than a separate discipline.
Section 1: The click-through rate decline — how big, how fast
The core trend is well established across multiple independent studies, and it has accelerated since early 2025.
An Ahrefs study published in February 2026, analysing 300,000 keywords, found that pages ranking in the top organic positions see a 58% reduction in click-through rate when an AI Overview appears above them — nearly double the 34.5% decline the same research house measured in April 2025. The trend is not levelling off. It’s compounding.
Pew Research Center’s study of real user behaviour found that only 8% of users click through to a traditional search result when an AI Overview is present on the page, compared with 15% when no AI Overview appears — roughly half the click-through rate.
The broadest measure comes from SparkToro’s analysis, published in June 2026: 68% of all Google searches now end without any click at all, up from roughly half that a few years earlier. That figure covers all search, not just legal queries, but it sets the baseline against which any individual firm’s traffic decline should be judged. If your traffic is falling and you assume it’s something you did wrong, it’s worth checking this baseline first.
What this means for law firms specifically
Informational and question-style legal queries — “do I need a solicitor for X,” “how does Y work” — are exactly the type of query most likely to trigger an AI Overview. These are often the queries that bring first-time visitors into a firm’s site at the top of the funnel. That’s the traffic most exposed to this decline.
Section 2: Risk by practice area
AI’s impact on legal services isn’t one number. It’s the product of two separate variables that move independently of each other: how often a query in that practice area triggers an AI-generated answer in the first place, and separately, how much the public actually trusts AI to be involved in that kind of legal matter. A practice area can score high on one and low on the other — and that mismatch is where the real strategic risk sits, because it’s the least intuitive pattern and therefore the one firms are least likely to have already planned for.
The table below sets out where each of the seven practice areas we track currently sits on both measures, alongside current average cost-per-click as a proxy for competitive and commercial pressure.
| Practice area | AI answer frequency | Public trust in AI for the matter | Current avg. CPC | Primary pressure |
|---|---|---|---|---|
| Family law | High | Low (17% would trust AI for divorce) | £5.50 | Discovery interception, not substitution |
| Wills & probate | High | Medium-high (47% would trust AI to draft a will) | £5.70 | Service substitution for simple cases |
| Personal injury | Medium | Medium (comfortable with AI for paperwork) | £15–£50 for competitive terms | Auction inflation, aggregator competition |
| Conveyancing | Medium | Low (only ~19% favour chatbots for property matters) | £6.20 | Moderate; high transaction value favours human oversight |
| Employment law | Medium | Low-medium | £5.30 | AI-drafted grievances increasing case complexity, not reducing enquiries |
| Immigration | Low-medium | Low (high-stakes, error-sensitive) | £4.80 | Government automation on simple visa queries only |
| Criminal law | Low | Very low (11% would trust AI for defence) | £8.20 | Minimal; scarcity and urgency keep costs high regardless of AI |
Two things stand out when the data is laid out this way rather than practice area by practice area.
First, AI answer frequency and consumer trust are inversely related almost everywhere except one place: family law. That’s the practice area where the two measures pull in opposite directions hardest, and it’s worth dwelling on. A family law client is highly likely to have their initial questions answered by AI before they ever reach a solicitor’s website, yet that same client wants nothing to do with AI when it comes to the actual matter. The commercial risk here isn’t losing the case to a machine. It’s losing the client at the discovery stage, before the human-versus-AI trust question is even relevant, because the firm simply wasn’t visible when the AI-mediated search happened.
Second, cost pressure and AI disruption aren’t the same axis. Personal injury has moderate AI answer frequency but the highest CPCs on the table by a wide margin, driven by claims management companies and insurance-side automated bidding rather than by AI substitution itself. Criminal law sits at the opposite end on AI exposure but still commands a high CPC, because scarcity and urgency — not competition from AI — set the price. A firm evaluating its own exposure needs to look at both columns separately, not assume that “high AI risk” and “high acquisition cost” always travel together.
Section 3: Why the published averages understate the problem for individual firms
Section 1’s figures are aggregate. An average CTR decline of 58% is a useful signal that something industry-wide is happening, but it flattens a lot of variation underneath it — and the firms sitting below that average are the ones least likely to know it.
Two structural reasons explain why an individual firm’s decline can be sharper than the published average suggests. The first is competitive density. Aggregate studies pool queries across every sector Google indexes; legal queries, particularly local ones, sit in a more competitive and more heavily AI-summarised category than the average query in the dataset. The second is content thinness. AI Overviews are more likely to appear — and to fully answer the question themselves, eliminating the need for a click — on pages where the underlying search results are shallow or repetitive. A local market with five firms running near-identical, thin practice area pages is exactly the environment where Google’s AI has enough to build a complete answer without sending the searcher anywhere.
This is a measurement problem before it’s a strategy problem. The standard way most firms check their own visibility — searching their own name or practice area terms manually — is structurally unreliable for this purpose. Personalisation, location, search history and login status all shape what any individual sees when they search, which means a managing partner checking their own firm’s position gets a result that doesn’t match what an actual prospective client, searching cold, would see. Proper measurement requires tools built for this specifically: rank-tracking software with AI Overview and AI Mode detection, tracking not just position but whether the firm’s content is cited within the AI-generated answer itself, and click-through rate trends isolated by query rather than blended across the whole site.
Once that measurement is in place, the response becomes tactical rather than reactive. A newly-appeared AI Overview on a previously reliable keyword is a signal to revisit that specific page — tightening the direct answer near the top, adding the kind of specific detail AI systems tend to cite, or building a dedicated FAQ-style page if the query is common enough to warrant one. Firms without this monitoring in place typically only notice a problem once enquiry volumes have already fallen, at which point diagnosing the cause — AI interception versus a ranking drop versus a seasonal dip — is far harder to do with confidence.
Section 4: How AI systems decide what to cite — and what that means for existing SEO work
It helps to understand, at a basic level, what an AI Overview or an AI chat tool is actually doing when it answers a legal query. It isn’t generating an answer from nothing. It’s summarising and synthesising a small number of sources it has judged to be relevant and trustworthy for that specific question — sources drawn heavily from what already ranks well in traditional organic search. This is why “AI visibility” and “SEO” aren’t two separate disciplines competing for budget. The second is largely the input the first depends on.
That said, ranking well organically is necessary but not sufficient. A page can rank on page one and still never get cited in an AI-generated answer, because citation depends on a narrower set of trust signals than ranking does. Three are worth a firm auditing for specifically, because they’re commonly missing even on otherwise well-optimised sites.
1. Named authorship.
Expertise needs to be attributabed to a real, named, credentialled person, not to the organisation as a whole. AI systems weigh authorship as a trust signal in a way traditional ranking algorithms historically haven’t emphasised as heavily — content with a visible, qualified author behind it reads as more trustworthy input for a system that’s ultimately trying to avoid citing something wrong. A firm’s own content audit is a reasonable starting point: pull up the ten highest-value practice area pages and check how many have a named solicitor attached, rather than “our team” or no attribution at all.
2. Structured data
Structured data is a machine-readable layer underneath a webpage that tells a crawler or an AI system precisely what it’s looking at, rather than leaving it to infer. For a law firm, the relevant schema types describe the organisation itself, the individual solicitors and their credentials, and any direct question-and-answer content on the page. Most law firm sites either have none of this beyond whatever a generic SEO plugin auto-generates, or have it only partially implemented — which matters, because incomplete structured data can be as unhelpful as none at all if it creates inconsistency between what the schema claims and what the page actually says.
3. Social proof
Third-party trust signals, particularly reviews are a crucial success factor. A high review count is a weaker signal on its own than reviews that name a specific practice area and location, because that specificity gives an AI system concrete, checkable context rather than generic sentiment. This is a low-cost, high-leverage area for most firms to improve — it requires a consistent ask at the point of case closure, not a technical project.
None of this displaces the fundamentals a firm should already have in place: a technically healthy site, genuinely useful content, and a good mobile experience. What changes is the marginal work worth prioritising next. For a firm with solid SEO foundations already in place, the three signals above are usually the fastest route to AI visibility specifically. For a firm without those foundations, they’re worth building in tandem rather than treating AI visibility as something to bolt on separately.
Section 5: What this means commercially
For firms that lean heavily on paid search to fill their pipeline, there’s a longer-term financial argument here as well as a visibility one. Every high-intent keyword a firm can rank for organically — and be cited for in AI-generated answers — is a keyword it no longer needs to pay for. Over time, that reduces dependency on paid campaigns and builds something more durable than a media budget: visibility that doesn’t disappear the moment spend stops.
The gap between firms that are visible in AI-mediated search and firms that aren’t is widening, not narrowing, and it’s happening faster than most internal marketing conversations have caught up with. But the response is identifiable and, for most firms, achievable without a wholesale strategy change: measure AI visibility specifically rather than assuming traditional rank tracking covers it, strengthen the authorship, structured data, and review signals outlined above, and treat content decisions as an ongoing response to a search landscape that’s now changing month to month rather than year to year.
The reader your business most wants — the one with budget, with experience, with discernment — has heard this language before, from the last three providers who didn’t deliver. They’ve learned to read it not as a promise but as a tell. A site that promises an outcome without acknowledging the work, the trade-offs, or the conditions for success is signalling one of two things: the writers don’t know how the outcome is actually achieved, or they don’t want the buyer to know. Neither is reassuring.
If your current marketing partner isn’t already talking to you about AI citation share alongside traditional rankings, that’s worth asking about directly.
Additional research used in this report:
Robin AI / Perspectus Global, 2025: Consumer trust survey — 47% trust AI for wills, 17% divorce, 11% criminal defence
Leapfrog Internet Marketing, April 2026: 28% of consumers using ChatGPT to research solicitors by late 2025
This report will be reviewed and refreshed periodically as the underlying research develops. If you’d like to discuss what any of this means for your firm specifically, Somuna offers a free 30-minute website review called The First Look — no preparation needed, no sales pitch.
Need a hand?
If you’re not sure where your firm stands, that’s worth finding out before the gap gets any wider. At Somuna, we offer a free 30-minute website review called The First Look — we open your site for the first time, live, and give you honest feedback on what’s working and what isn’t, including how your firm appears in AI search. No preparation needed. No sales pitch. Just a straight conversation about where you are and what might need to change.
John Anderson has 15+ years experience in online growth and SEO. He helps organisations get found and make sense to customers through clear messaging, sustainable acquisition strategies, and a grounded, practical approach to AI.
