Health 69.7%, Finance 68.8%, Law 61.1%: The Industries Most Exposed to AI Overviews

Health queries trigger an AI Overview 69.68% of the time. Finance 68.80%. Law and Government 61.07%. Real Estate, at the other end, 22.27%.

That spread is 47.41 percentage points, and it comes from DataForSEO’s Search Insider: 2026 Edition, which sampled roughly 190 million US English keywords between June and September 2026 and grouped them by industry.

Two businesses can hold the same position one ranking and face completely different competition for the click. One of them has an AI answer above them seven times out of ten. The other has one twice out of ten.

The full table, all 23 industries

Overall trigger rate across the dataset: 56.7%.

The three most regulated sectors are first, second and eighth

We built this agency around financial services, legal and healthcare because those are the sectors where getting an answer wrong carries a consequence. It turns out they are also the sectors where Google is most willing to generate the answer itself.

I find that genuinely uncomfortable, and I think the industry should talk about it more than it does. The queries where a wrong answer costs someone money, a case or their health are the queries most likely to be answered by a summary rather than by a named, accountable source.

The report does not explain the pattern, so what follows is my reading rather than its finding.

These sectors are overwhelmingly informational at the top of the funnel. People ask questions for months before they ask for a quote. Informational intent triggers an AI Overview 70.44% of the time in the same dataset, up from 48.42% in Q4 2025, so a sector made of questions is a sector made of AI Overviews.

They also have the deepest reference corpus of any industry. NHS and NICE guidance, FCA handbooks, case law, professional body publications. The material that makes these fields answerable at all is the material that makes them summarisable.

Health backs this up in a second way. It has the highest trigger rate and the highest number of distinct domains in the top 1,000 most-cited list, at 122, more than any other industry. The citations are spread wide rather than concentrated in a handful of big names, which is the one genuinely encouraging number in this report for a specialist firm.

Trigger rate is not citation share, and the difference matters

These are two different measurements from two different datasets, and I have already seen them blurred together in coverage of this report.

Trigger rate is how often an AI Overview appears at all on results in your industry. It tells you how much of your click opportunity is being intercepted.

Citation volume is who gets named inside the answer. Arts and Entertainment dominates that league table with 135.7 million AI Overview citations, more than three times News and Media at 39.5 million, which tells you nothing useful about whether a Sheffield law firm can get cited.

High trigger rate plus wide citation distribution, which is what Health shows, is the most workable combination there is. It means the answer is nearly always generated and the sources it draws on are not locked up by five domains.

You cannot dodge it by picking different keywords

The standard response to this is to go after longer, easier, cheaper terms where the AI is supposedly less interested. The same report tests that three ways and it fails all three.

By search volume. Keywords with 201 to 300 monthly searches triggered an AI Overview 58.59% of the time in Q3 2026. Keywords with 601 or more triggered one 48.62% of the time, the lowest rate of any bucket. Lower volume meant more AI Overviews, not fewer.

By difficulty. Mid-difficulty keywords, KD 21 to 60, trigger at around 65%. The easiest, KD 1 to 20, at 60.64%. The hardest, KD 81 to 100, at 53.76%, the lowest of the five bands. Easy keywords are not a hiding place.

By commercial value. Keywords costing $5 or more per click trigger at 49.55%, almost identical to the cheapest at 50.54%. If you were hoping the expensive terms were protected because Google wants the ad revenue, they are not.

Industry is the only one of the four variables with a genuinely wide spread. Unless you are prepared to become an estate agent, keyword selection will not get you out of this.

What this changes for a regulated firm

Your forecast is wrong before your strategy is. If seven in ten of your target queries carry an AI Overview, and that overview takes an estimated 42% to 45% of the clicks, then the traffic model behind your content plan is overstating what a ranking is worth. Fix the model first. We wrote up what happened to desktop click-through rates separately, and the numbers there are worse than most people expect.

The compliance burden that slows you down also protects you. Every firm I speak to in these sectors treats sign-off cycles as the reason they can’t publish quickly. They are, and that’s real. They’re also the reason this opportunity has a longer life in financial services and law than it does in e-commerce. The barrier applies to your competitors too, and it keeps the thin, fast-produced content out of a market where it would otherwise flood in.

Specialism beats scale here more than anywhere. 122 distinct health domains in the top 1,000 cited list describes a market where being the clearest source on one narrow question is still a viable route in. Three giants haven’t cornered it.

If you want the sector-specific version of this, we keep running detail for financial services, legal and healthcare, including the Consumer Duty angle for advisers and the zero-click problem specific to regulated sectors.

One honest limit on all of this

United States, English language, DataForSEO’s own industry classification. Only the backlink portion of their research is global.

I am not going to present an American industry table as a British one. What travels across the Atlantic is the ordering, because it reflects query intent and how much reference material exists in each field, and neither of those is specifically American. What doesn’t travel is the exact percentage, and the gap between the UK and US on AI rollout has been real and persistent.

Our own UK figures are published separately, and they are thinner, because almost nobody is measuring this properly on this side.

Where to start

Take your 20 highest-value queries, the ones that precede an enquiry rather than the ones with the biggest volume, and search them. Count how many return an AI Overview and how many name you inside it. That gives you your own trigger rate and your own citation rate in an afternoon, which beats any industry average.

Then decide whether the gap is worth closing. For most firms in these three sectors it is, because the alternative is competing for a click that has already been taken.

The AI Visibility Score does the first part of that for you, free, if you would rather not do it by hand.