For twenty-five years, "search" meant the same thing: type a few words, scan a page of blue links, click one, land on a website. Whole industries, including how an accountancy firm gets found, were built on that single motion. In 2026 that motion changed. This briefing sets out what happened, how far it has gone, and the one part of the story that is genuinely good news for firms willing to act. It is written to be checkable: every external statistic links to a primary source you can click, and there is a full Sources list at the bottom. Where a number is our own, we say so and give the denominator. Where we could not stand a popular claim up against a credible source, we left it off, and we say which ones, and why, at the end.
Part one
The shift: search itself changed
In May 2026, at Google I/O, Google announced that AI Mode, its conversational, answer-first experience, is becoming the default way Search works, worldwide.
According to Google's own announcement, AI Mode is now powered by Gemini 3.5 Flash as its default model, rolling out globally; the company also unveiled "the biggest upgrade to our Search box in over 25 years," an intelligent box that expands for longer, more conversational queries.[1] Google reported that AI Mode had surpassed one billion monthly users, with queries "more than doubling every quarter" since launch.[1] Google also introduced background "information agents" that monitor the web for you 24/7: search that does work without you watching.[1]
The detail that matters most for any local service business is simple: the answer now comes first, and the list of websites comes second, if at all. The interface most of your prospects open is no longer a list of ten links you can climb. It is a synthesised answer that names a few businesses and moves on. The front door moved, and it moved fast.
Why this is the start of the story, not the end of it. A default is not a niche feature you can ignore until it matures. When AI answers are the default experience for a billion-plus users, the question for a firm stops being "should we worry about AI search yet?" and becomes "when a prospect asks, is our firm one of the names the answer contains?"
// Source: Google, "Google Search's I/O 2026 updates," blog.google, May 2026. blog.google/products-and-platforms/products/search/search-io-2026 (accessed 6 Jun 2026).
Part two
The magnitude: most searches never reach a website now
People sense this is happening. The data shows how far it has gone, and it is further than most firms assume. These are independent, named studies, not vendor marketing.
// The magnitude · 01 · zero-click
83% of AI-Overview searches end with no click at all
The Pew Research Center analysed the real browsing of 900 US adults in March 2025. When a search returned a Google AI Overview, 83% of those searches ended without the user clicking any result, compared with 60% of searches that had no AI summary. Users clicked a link inside the AI summary just 1% of the time.[2] In other words, when the AI answers, the website rarely gets the visit.
// The magnitude · 02 · zero-click, confirmed
Around 60% of all searches now end without leaving for another site
An independent line of evidence reaches the same place. Bain & Company's consumer research found about 60% of searches now end without the user progressing to another destination site, and that around 80% of consumers rely on these "zero-click" AI results for at least 40% of their searches (Bain–Dynata survey, December 2024, n=1,117).[3] Two different methods, Pew's browsing data and Bain's survey, point to the same shift.
// The magnitude · 03 · the forecast
Gartner forecasts traditional search volume down 25% by 2026
It isn't only behaviour inside a search results page that's shifting; it's the volume of traditional search itself. Gartner predicted that traditional search engine volume will drop 25% by 2026, as AI chatbots and virtual agents become substitute answer engines.[4] Gartner has been clear this figure is scenario modelling rather than certainty; we label it as a forecast, not a measured fact.
// The magnitude · 04 · the click drop
When an AI Overview appears, the top result loses a large share of its clicks
For the pages that do still rank, AI answers eat into the clicks. Ahrefs, studying 300,000 keywords, found that the presence of an AI Overview was associated with a 34.5% lower click-through rate for the top-ranking page in its April 2025 analysis and, in a December 2025 update, a 58% lower average click-through rate as AI Overviews expanded.[5] The number you cite depends on the month; the direction has only ever been one way.
What "the magnitude" adds up to: the page of ten blue links is no longer where most searches end. The answer is the destination. For a firm, the practical consequence is blunt: you can no longer rely on ranking your way to a click, because for most searches there is no longer a click to win. Being mentioned in the answer is the new visibility.
// Sources: Pew Research Center (2025); Bain & Company (2025); Gartner (2024); Ahrefs (2025). All linked in full in the Sources list below. Several are US-based studies; the behavioural shift they describe is global, but exact UK figures will differ.
Part three
The opportunity: being the name AI recommends is now the best position in search
Here is the part most firms miss while they're worrying about lost clicks. If most searches no longer produce a click, then the clicks that do happen are worth far more, because the visitor arrives already pre-qualified by the AI. They didn't scroll a list and guess. They asked for a recommendation and got one, and your firm was the recommendation.
// The opportunity · 01 · conversion
AI-referred visitors can convert dramatically better than organic search
In a case study by the agency Seer Interactive, covering a single client from October 2024 to April 2025, traffic that arrived from ChatGPT converted at 15.9%, against 1.76% for the same client's Google organic traffic, roughly a nine-fold difference.[6] Perplexity converted at 10.5% in the same study. AI traffic was a tiny fraction of total volume (around 11,000 sessions), but it converted like traffic that already knew what it wanted.
The honest caveat, stated up front: this is one client, one agency, a small number of AI sessions, and a US dataset, not an industry average, and not a promise of your numbers. We cite it because it is a named, primary case study you can read in full, and because it illustrates a pattern many others report: AI sends fewer visitors, but more ready-to-act ones.
The logic is simple: a recommendation-primed buyer is the most valuable visitor in search. Search has moved from "here are some links, you decide" to "here is who you should use." Being one of the names in that answer is the single highest-value position there is, and, unlike a #1 ranking, almost nobody in a typical local accountancy market is competing for it yet.
// The opportunity · 02 · it's optimisable
Being named is something you can influence: academic research shows it
This isn't luck. The foundational academic paper on the topic ("GEO: Generative Engine Optimization," from researchers at Princeton and collaborators, presented at KDD 2024) showed that targeted optimisation methods can boost a source's visibility in generative-engine answers by up to 40%, with tactics like adding statistics, citing sources and including quotations among the biggest levers.[7] Whether an AI names you is shaped by signals you control.
Our own evidence: the firms aren't ready for this yet
Everything above is the macro picture, drawn from credible third parties. The local picture (are the firms competing for these recommendations actually ready to be named?) is something we measured ourselves. In June 2026 we ran our own free Foundations Check across the public websites of 93 UK accountancy firms across three groups, to see whether the machine-readable plumbing AI assistants rely on is even present.
| Signal | General (n = 34) | IR35 / contractor (n = 28) | E-commerce (n = 31) |
|---|---|---|---|
| Says "I am an accounting business" in machine-readable schema (of readable firms) | 17% 4 / 24 | 19% 4 / 21 | 33% 8 / 24 |
| FAQ schema present (of readable firms) | 0% 0 / 24 | 24% 5 / 21 | 46% 11 / 24 |
| Blank / unreadable to a simple crawler (of full sample) | 18% 6 / 34 | 25% 7 / 28 | 23% 7 / 31 |
Renownly first-party research, n = 93 (34 + 28 + 31), 3 June 2026, three convenience samples. Schema percentages are over the firms whose homepages we could read (denominator beneath each figure); blank/unreadable is over the full sample. Aggregate and anonymised; no firm is named. Foundations / readiness only, not live AI rankings; our reader does not execute JavaScript. Full method and limitations in the 93-firm report and the contractor cut.
The pattern: the AI front door has arrived, but the firms haven't walked through it. In our IR35/contractor sample, only 4 of the 21 readable firms (19%) carried machine-readable "I am an accounting business" markup, and a quarter of the full sample served a simple crawler nothing readable at all.[8] The opportunity in Part three is real precisely because so few firms are ready to claim it. Today, the group getting the foundations right is genuinely small.
What this means for an accountancy firm in 2026
- The shift is a default, not a trend. AI answers are now the front door for a billion-plus searchers, so treat them as the primary discovery channel, not a side experiment.[1]
- Ranking for clicks is a shrinking game. Most searches no longer produce a click at all.[2][3] Plan for being mentioned, not just listed.
- Being recommended is the prize. The visitor an AI sends arrives pre-qualified, and converts accordingly in the evidence we can find.[6] It is the most valuable, least-contested position in search right now.
- It's within your control, and the field is open. Visibility in AI answers is optimisable,[7] and on our own data most firms haven't done the basics yet.[8]
None of this is a promise that any AI engine will recommend your firm. The tools are run by third parties and change without warning, and our research measures foundations, not outcomes. What we can say honestly: the firms that get readable, identify themselves clearly to AI, and hand over the answers prospects ask for are giving the assistants the clearest reason to put their name forward.
Sources
// Every external claim, with name, URL and access date (all accessed 6 June 2026)
- [1] AI Mode the default worldwide; Gemini 3.5 Flash; biggest Search-box upgrade in 25+ years; 1bn+ AI Mode monthly users; background "information agents."
Google, "Google Search's I/O 2026 updates: AI agents and more," blog.google, May 2026.
https://blog.google/products-and-platforms/products/search/search-io-2026/ · accessed 6 Jun 2026. - [2] 83% of AI-Overview searches end with no click (vs 60% without one); 1% click a link inside the AI summary.
Pew Research Center, "Google users are less likely to click on links when an AI summary appears in the results" (analysis of 900 US adults, March 2025), July 2025.
https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/ · accessed 6 Jun 2026. - [3] ~60% of searches end without progressing to another site; ~80% of consumers rely on zero-click results for ≥40% of searches.
Bain & Company, "Goodbye Clicks, Hello AI: Zero-Click Search Redefines Marketing" (Bain–Dynata survey, December 2024, n=1,117), 2025.
https://www.bain.com/insights/goodbye-clicks-hello-ai-zero-click-search-redefines-marketing/ · accessed 6 Jun 2026. - [4] Traditional search engine volume forecast to drop 25% by 2026 (scenario modelling, labelled as a forecast).
Gartner, "Gartner Predicts Search Engine Volume Will Drop 25% by 2026, Due to AI Chatbots and Other Virtual Agents," 19 February 2024.
https://www.gartner.com/en/newsroom/press-releases/2024-02-19-gartner-predicts-search-engine-volume-will-drop-25-percent-by-2026-due-to-ai-chatbots-and-other-virtual-agents · accessed 6 Jun 2026. - [5] AI Overviews associated with a 34.5% lower CTR for the top result (April 2025); 58% lower in a December 2025 update; 300,000-keyword study.
Ahrefs, "Update: AI Overviews Reduce Clicks by 58%."
https://ahrefs.com/blog/ai-overviews-reduce-clicks-update/ · accessed 6 Jun 2026. - [6] ChatGPT traffic converted at 15.9% vs 1.76% for Google organic (single client, ~11k AI sessions, Oct 2024–Apr 2025); Perplexity 10.5%.
Seer Interactive, "Case Study: 6 Learnings, 1 site: How Traffic from ChatGPT Converts." Single-client case study, not an industry average.
https://www.seerinteractive.com/insights/case-study-6-learnings-about-how-traffic-from-chatgpt-converts · accessed 6 Jun 2026. - [7] Targeted "generative engine optimization" methods can boost a source's visibility in AI answers by up to ~40%.
Aggarwal, Murahari, Rajpurohit, Kalyan, Narasimhan, Deshpande, "GEO: Generative Engine Optimization," Princeton University et al., KDD 2024.
https://collaborate.princeton.edu/en/publications/geo-generative-engine-optimization/ · accessed 6 Jun 2026. - [8] Foundations gaps among UK accountancy firms (e.g. only 4 of 21 readable IR35 firms (19%) carry accountant-identifying schema; a quarter served a crawler nothing readable).
Renownly first-party research, n = 93 (34 general + 28 IR35/contractor + 31 e-commerce), 3 June 2026, three convenience samples, aggregate and anonymised, foundations/readiness only. Full method and limitations: the 93-firm report and the contractor cut. · published 3 Jun 2026.
// Notes on figures we deliberately left off
Credibility is the whole point of this page, so a few popular numbers are not here, on purpose:
- "AI Mode is 93% zero-click." We could not trace this to a credible primary; the attribution we found (to a named SEO study) did not hold up when we checked the study itself. Dropped. We use Pew's verifiable 83% AI-Overview figure instead.[2]
- "AI visitors convert ~14% vs ~2.8% organic (~5×)." This widely-repeated headline traces only to vendor reports we could not independently verify as primary sources. Dropped in favour of the named Seer Interactive case study,[6] with its caveats stated plainly.
- Recycled SME-adoption and "thousands-of-percent YoY" growth claims. No credible primary found; not used.