Summary
- The traffic effect is real. In Pew’s analysis of real browsing, users clicked a traditional result on 8% of visits when Google showed an AI summary, against 15% when it didn’t.
- It isn’t the end of search. It moves the contest earlier: being named and cited in the answer now matters as much as ranking beneath it.
- B2B buyers use AI but check it. Gartner found 45% used generative AI in a recent purchase, and 69% turn to sales reps to validate what AI tells them.
- Three decisions for this year: what you want AI to say about you, where the evidence for it lives, and how you’ll measure a visit that never happens.
Written for: CEOs, CMOs and board members of mid-size companies that win a meaningful share of enquiries from search, and want a decision rather than a trend report.
Is AI search actually reducing our traffic?
Probably, for informational searches, and the evidence is solid. Pew Research Center tracked the real browsing of 900 US adults in March 2025. When Google showed an AI summary, users clicked a traditional search result on 8% of visits; without one, on 15%. They clicked a link inside the AI summary itself on just 1% of visits, and ended their browsing session on 26% of those pages, against 16% otherwise.
AI summaries appeared on 18% of the 68,879 searches in Pew’s sample. That was a snapshot of one month in one market, and the feature has kept expanding since. For a mid-size company, the practical reading is simple: some of the research that used to happen on your website now happens in an answer that may or may not mention you.
Source: Pew Research Center
Should we worry about Gartner’s 25% forecast?
Take it seriously without treating it as a countdown. In February 2024, Gartner predicted that traditional search engine volume would drop 25% by 2026 as people turn to AI chatbots and other virtual agents. Forecasts like this are directional, and plenty of analysts dispute the size of the drop. What isn’t in dispute is the direction: more research happens inside an answer, and fewer visits reach your site to do it.
For a board, the useful question isn’t whether traffic will fall. It’s whether the enquiries that matter will still find you, and whether you’ll be able to tell.
How do B2B buyers actually use AI when they buy?
As a research assistant they don’t fully trust. Gartner found 45% of B2B buyers used generative AI during a recent purchase, mainly to gather information on vendors and products, and 69% turn to sales reps to validate AI-generated insights. Among engineers and technical buyers, TREW Marketing and GlobalSpec’s 2026 research found AI answers scored just 4.7 out of 10 for trustworthiness, up slightly from 4.4 a year earlier.
That’s the opening for a mid-size company. The buyer asks an assistant for suppliers, gets a shortlist, then checks it: your website, your proof, a conversation. You need to be on the list, and you need the check to confirm it.
of B2B buyers used generative AI, mainly to research vendors and products
Source: Gartner
What should we want AI to say about us?
Something specific, true and repeated everywhere. AI assistants assemble answers from what’s consistently written about you across your site, directories, press, reviews and other people’s content. A company described twelve different ways gets summarised vaguely, or not at all.
| A question your buyer asks an assistant | What the answer needs to find | Where it should live |
|---|---|---|
| “Who makes precision aluminium castings for EV programmes near Pune?” | Your processes, materials, certifications and location, stated plainly | Capability pages, directory listings, trade press |
| “Which logistics partners in Maharashtra handle pharma cold chain?” | Named sectors, compliance credentials, client types, scale | Sector pages, case evidence, association listings |
| “How does [your company] compare with [competitor]?” | An honest comparison with clear differences | A comparison page on your own site |
| “What does [your service] cost for a company our size?” | Pricing logic or ranges, and what drives them | A pricing or “how we work” page |
What are the three decisions for this year?
1. Decide the sentence
Agree the one-sentence description of the company you want an assistant to give, and the three facts that prove it. This is positioning, and it’s a leadership decision, not an SEO task.
2. Put the evidence where it can be read
Clear pages for each capability, sector and location; specifics over adjectives; pricing logic; named experts; consistent descriptions on every directory and profile. Structured data helps machines read it. None of it works if the page says “world-class solutions”. This is the core of AI search visibility work.
3. Change what you measure
Track mentions and citations in AI answers for your priority questions, alongside branded search and direct enquiries, not just organic traffic. If traffic falls while qualified enquiries hold, the strategy is working. The five numbers in our piece on board-ready marketing reports still apply; add AI visibility beside them.
What does a 30-day AI visibility audit look like?
Short, cheap and revealing. You don’t need a tool to start; you need the questions your buyers actually ask and the discipline to record the answers. We’d run it in four weekly steps, owned by one named person.
| Week | What happens | Output |
|---|---|---|
| 1 | List the 20 questions buyers ask before they shortlist: category, comparison, cost, location and proof questions. Ask each one in three AI assistants and in Google. | A grid of who gets named, how often, and what’s said about them |
| 2 | For every answer that names a competitor, find the pages and sources it drew on. For every answer that misdescribes you, find where the wrong description came from. | An evidence gap map: what’s missing, what’s wrong, what’s inconsistent |
| 3 | Fix the top five gaps: rewrite the relevant pages in specifics, align descriptions across directories and profiles, publish the missing evidence. | Five fixes live, each linked to a question |
| 4 | Re-run the same 20 questions and set up a monthly tracker alongside qualified enquiries by source. | A baseline the board can see every month |
Mid-size companies in Pune and Mumbai have one advantage here: the questions that matter are specific. “Precision casting supplier near Chakan” or “NABL-accredited lab chain in Thane” are far easier to win than generic category terms, and assistants answer them from exactly the kind of concrete, local evidence a focused company can publish in weeks.
The question isn’t whether AI will send you traffic. It’s whether AI will say your name when a buyer asks who to call.
Questions to take into your next leadership meeting
- When we ask ChatGPT, Gemini and Google’s AI Mode who the leading suppliers in our category are, are we named? Who is?
- What one-sentence description of us would we want an assistant to give, and does our website actually say it?
- Which of our pages answer buyer questions with specifics, and which with adjectives?
- Are we measuring qualified enquiries by source, or just traffic?
- Who owns AI visibility here: marketing, IT, or nobody?
Questions leaders ask us about this
Does SEO still matter?
Yes. AI answers draw heavily on pages that are well structured, crawlable and authoritative, which is what good SEO produces. What changes is the goal: from ranking a page to being the source an answer relies on.
Can we pay to appear in AI answers?
Advertising is arriving inside some AI experiences, but organic mentions come from what’s published about you. The durable route is clear, specific, consistent information on your own site and across credible third-party sources.
How quickly can this change?
Unevenly. Clearer, better-structured pages can be picked up within weeks; changing how an assistant describes your whole category takes months of consistent publishing. Measure monthly, judge quarterly.
Expert verdict
Treat AI search as a positioning test, not a technology project.
The companies that do well in AI answers are the ones that are easy to describe accurately: one clear position, specific evidence, consistent everywhere. That’s leadership work first and technical work second. Traffic from informational searches will keep softening. The goal is to be the name that appears when a buyer asks, and the site that confirms it when they check.
Do now
Ask three AI assistants your buyers’ ten most common questions and record who gets named.
Stop
Judging search performance on traffic alone.
Measure
Mentions and citations in AI answers for your priority questions, plus qualified enquiries by source.
Verdict by our Growth lead, 3Anomaly founding team
Sources
- Pew Research Center, “Google users are less likely to click on links when an AI summary appears in the results”, 22 July 2025
- Gartner, “Gartner Predicts Search Engine Volume Will Drop 25% by 2026, Due to AI Chatbots and Other Virtual Agents”, 19 February 2024
- Gartner, “Gartner Survey Finds 69% of B2B Buyers Turn to Sales Reps to Validate AI-Generated Insights”, 20 May 2026
- TREW Marketing and GlobalSpec, 2026 State of Marketing to Engineers (press release, March 2026)
