78% of legal search queries now trigger a Google AI Overview, the highest rate of any industry Semrush tracks. Google's own head of retail says search isn't dying, it grew through its biggest quarter ever despite predictions it would shrink 25%. Both things are true at once, and for law firms they add up to the same conclusion: search hasn't gone away, it's stopped being about keywords and started being about whether AI trusts you enough to say your name.
The Death of Search That Wasn't
In 2026, a widely circulated Gartner projection said large language models would cause traditional search volume to decline by roughly 25%. Google's own global head of retail and consumer goods, Sophie Neary, tells a different story from inside the building: by the end of that year, Google recorded its biggest quarter of search usage ever. Not despite AI. Because of it.
That's not spin, it's a structural point worth sitting with. When AI makes it easier to ask a question, find an answer, and act on it, people don't stop searching, they search more, and differently. One in six Google searches is now voice or image rather than typed words. Conversational, multi-turn queries let someone build on a question instead of starting a fresh search every time. The box didn't disappear. It got a lot more capable, and so did the expectations of the person typing, or now, talking, into it.
For law firms, the practical translation is this: the volume of people looking for legal help online isn't shrinking. What's shrinking is the number of them who ever click through to a website to find it.
Why Legal Search Is Different From Everything Else
Every industry is living through this shift, but legal is experiencing it harder and faster than almost any other category tracked. Recent analysis of more than 10 million keywords found legal queries trigger a Google AI Overview 78% of the time, the highest rate of any industry Semrush measured. Separate research projects that by mid-2026, 70 to 80% of legal search queries will end in a zero-click outcome, the person gets their answer and never visits a single website.
There's a reason legal sits at the extreme end of this. Legal questions are exactly the kind of question people want summarised, not researched, they want to know what to do next, not to read six competing blog posts and form their own view. That makes law one of the categories AI systems are most confident answering directly, and one of the categories where a firm's absence from that answer is most costly.
It's Not About Keywords Anymore
Translate that to legal and the pattern is identical. Nobody searching for a lawyer types "divorce lawyer" anymore than they'd type "pink t-shirt." They ask, in full sentences, sometimes out loud: my husband and I separated fourteen months ago, we have two kids and a house, what happens now. That's not a keyword. It's a conversation, and it's a conversation the AI system is now equipped to have with them directly, pulling in whichever sources it trusts to answer accurately.
Owning "family lawyer Sydney" as a ranked phrase matters less than it used to. Being the source an AI system trusts enough to draw from, and name, when someone describes their actual situation in their own words, matters more. That's the entire discipline behind AEO: structuring content, credentials and authority so a model can find, verify and cite you, not just so a crawler can index a keyword-matched page.
Who's Actually Asking AI to Find Them a Lawyer
The assumption that AI-assisted search is a young person's habit doesn't hold up in legal. Consumers aged 45 to 60 lead every other age group in willingness to use AI to research which firm to hire, at 76%, and households earning $125,000 or more adopt at the highest rate of any income bracket. That's not a coincidence, it's exactly the demographic most likely to be dealing with a divorce, an estate, a commercial dispute or a serious injury claim, the higher-value matters every firm actually wants.
Separately, BrightLocal's 2026 Local Consumer Review Survey found 45% of consumers used AI to find a local business or service in the past year, up from just 6% the year before. AI is now the third most common local discovery channel, behind only Google and Facebook, and closing the gap fast.
The Reputation Trap Catching Established Firms
Here's the part that should concern the firms who feel least at risk. Research into AI visibility across law firms has identified what's been called the "white-shoe visibility paradox": the firms with the strongest historical reputations have often invested the least in active AI visibility, because for decades that reputation travelled through closed networks, referrals, reputation, word of mouth among other lawyers, rather than through content an AI system can actually find and cite.
The result is an AI footprint that under-represents some of the most genuinely distinguished practices in the country, while less established firms with better-structured digital content get named instead. A strong reputation built the old way doesn't automatically transfer into the new discovery layer. It has to be made legible to it.
"Data Strength" Isn't a Google Buzzword, It's Your Foundation
Neary's least glamorous point in the conversation is also her most important one. She calls it data strength, structured, first-party information that powers Google's systems on your behalf. Every retailer gets access to the same AI-powered advertising tools; the differentiator is whose underlying data is actually clean, complete and trustworthy enough for the system to use well.
For a law firm, the equivalent foundation is structured, accurate, consistent information about the firm itself: named lawyers with real credentials and admission dates, practice areas mapped to genuine expertise, consistent name-address-phone details across every directory and profile, and clean schema markup tying it all together. It's the same unglamorous groundwork behind our technical SEO work, and it's the layer everything else, citations, recommendations, AI-generated answers, actually gets built on. Skip it and even excellent content has nothing solid to sit on.
Trust Compounds the Same Way in Every Category
Neary's favourite example from the podcast is Patagonia, the most-recommended fashion brand across large language models, not because of a clever prompt or a single viral campaign, but because of consistent brand storytelling, content and creative partnerships sustained over many years, generating the training data an AI system could actually learn from.
The mechanism is identical for a law firm, just built from different materials. Consistency over years in verified accreditation, peer-review directory listings, current client reviews, and named-lawyer content tied to a specific jurisdiction, that's the legal equivalent of Patagonia's brand consistency. It's the same review-engine and authority-building sequence behind our authority building and reputation management work, and it's not a project you finish, it's an asset you keep compounding.
What Agentic Search Means for a Law Firm
The next layer past conversational search is agentic AI, systems that don't just answer a question but take action on your behalf. Neary's own example is asking a shopping agent to buy a jacket automatically the moment it's back in stock. The legal parallel isn't far off: a prospective client's own AI assistant researching their situation, shortlisting firms based on credentials and reviews, and drafting the enquiry, potentially before a human ever visits your site directly.
Neary predicts that within three or four years, today's static websites, a picture, a button and some words on a page, will look as dated as the earliest e-commerce sites do now. The firms best positioned for that shift are the ones building websites and AI-powered client experiences today that can actually converse with a prospective client's own AI agent, not just wait for a human to click through.
Run This Check This Week
You don't need a platform subscription to get a first read on where you stand. This takes twenty minutes.
- Describe a real client situation to ChatGPT, Gemini and Perplexity, not a keyword.
Use the exact language a distressed or confused prospective client would use, not "divorce lawyer Sydney." See who gets named.
- Check whether your named lawyers' credentials are verifiable online.
Accreditation, admission details and specialist status need to be checkable by a model, not just claimed on your own site.
- Audit review recency, not just volume.
A hundred reviews from four years ago carries far less weight than twenty from the last quarter.
- Check your name, address and phone details across every directory.
Inconsistency between listings weakens the entity signal every AI system is trying to verify.
- Ask the AI directly why you weren't named.
The answer is sometimes wrong, but it's often a genuine, immediately fixable gap.
What to Actually Do About It
Search isn't dying for law firms any more than it's dying for retail. It's the same conclusion Neary reaches from inside Google: the product got better, so people use it more, just not in a way that rewards the old playbook of chasing keywords and hoping a blue link converts.
Being found is no longer the finish line, being trusted enough to be named is. That trust is built the unglamorous way: clean, verifiable data about your firm and your lawyers, consistent authority signals sustained over years, and content structured for a model to cite, not just for a person to scroll past.
None of that happens by accident, and none of it happens overnight. The firms still relying on reputation alone, or on last decade's keyword-ranked website, are the ones most exposed to the reputation trap already catching established names.
This is exactly the work we build for clients: SEO and AEO built around how AI systems actually verify and cite a law firm, websites engineered to be legible to both humans and models, and the underlying AI and technology layer, including custom GPTs trained on a firm's own knowledge, that keeps a firm's expertise consistent everywhere a prospective client, or their AI agent, might go looking for it.