If your organic traffic has softened over the past year and nothing on your site changed, you have probably heard the explanation already: AI search is answering questions directly, and fewer people are clicking through. That explanation is partly true. It is also incomplete in a way that leads many service firms to worry about the wrong pages and measure the wrong outcome.

The useful starting point is that the two most credible sources on this question disagree, and understanding why is more valuable than picking a side.

What the independent research found

The Pew Research Center analyzed the actual browsing behavior of 900 US adults across nearly 69,000 Google searches in March 2025. When an AI summary appeared at the top of the results, users clicked a traditional search result in 8% of visits. When no summary appeared, that figure was 15%. Links cited inside the AI summary itself were clicked in roughly 1% of visits, and users were more likely to end their browsing session entirely after seeing a page with a summary.

Two further findings matter more than the headline. About 18% of searches in the study produced an AI summary at all, and questions and longer, more conversational queries were considerably more likely to trigger one. Short, direct queries mostly did not.

What Google says

Google disputes the broader conclusion. In an August 2025 post, Google’s head of Search, Liz Reid, wrote that total organic click volume has remained relatively stable year over year, that average click quality has increased, and that Google is sending slightly more quality clicks to websites than a year earlier. Google defines a quality click as one where the user does not quickly return to the results. The same post argues that AI features are prompting people to search more often and with longer questions, which creates more opportunities to be clicked overall.

Treat that claim with the same care you would apply to any company describing its own product. The post does not publish the underlying data, and Google has an obvious interest in the conclusion. But it should not be dismissed either, because it is describing something the Pew data is consistent with.

How both can be true

Put the two together and a coherent picture emerges. AI features are absorbing a particular kind of click: the quick informational lookup, where someone wants a definition, a fact, or a short answer and leaves as soon as they have it. Those were always low value clicks for most businesses. They inflated traffic reports without producing much else.

Meanwhile, people who are actually evaluating a purchase, comparing providers, or ready to act still click through, because a summary cannot hire a contractor, sign an engagement letter, or book a consultation. Google’s own framing describes exactly this: an AI response gives the lay of the land, and people click when they want to go deeper.

For a service firm, that distinction maps almost perfectly onto your site. Your blog, especially the “what is” and “how does” articles written to attract search traffic, sits squarely in the zone AI summaries absorb. Your service pages, location pages, and contact paths sit mostly outside it. If your traffic has fallen and your inquiries have not, this is very likely why, and it is a far less alarming situation than the traffic chart suggests.

How to tell what is actually happening to you

Do not diagnose this from a single traffic graph. The answer is in data you already have.

Start in Search Console’s Performance report. Google’s documentation on AI features and your website confirms that clicks and impressions from AI Overviews and AI Mode are counted inside the standard Web search totals rather than reported separately. You cannot isolate AI traffic directly, but you can see its fingerprint.

Group your pages by intent. Put informational articles in one group and commercial pages (services, locations, pricing, contact) in another. Then compare the two over the same period.

The pattern that points to AI absorption looks like this: impressions on informational pages holding steady or rising while clicks and click-through rate fall. The page is still being shown; it is simply being summarized instead of visited. If your commercial pages show the same decline, AI summaries are a less likely explanation, and the cause is probably something else entirely. A drop that hits your money pages deserves a proper diagnosis before anyone blames AI for it.

Then check the only number that ultimately matters: are qualified inquiries down, or just sessions? If inquiries are stable, you have lost traffic you were never converting. That is worth knowing, and it is not a crisis.

What does not need to change

Google’s documentation is unusually direct on one point. There is no special markup, file, or structured data you need to add for AI Overviews or AI Mode, and no separate optimization is required. A page needs to be indexed and eligible to appear in Search with a snippet, and the existing fundamentals of being crawlable, well structured, and genuinely useful still apply.

That is worth knowing because a small industry has grown up selling special “AI markup” and similar shortcuts. Be skeptical of anyone promising that a technical trick will get you cited. The work that legitimately improves your odds of appearing in AI answers is real, but it is not a trick: clear structure, unambiguous information about who you are and what you do, content that answers the question precisely, and authority that other sites recognize. Most of that is ordinary good practice applied more deliberately.

What does need to change

Stop judging your blog by sessions. If informational articles were never producing inquiries, losing half their traffic costs you very little. Judge content by what it contributes: inquiries it starts, buyers it reaches who later convert, pages it sends people to next. If you are not measuring that yet, these are the numbers worth tracking.

Write what a summary cannot replace. An AI response can paraphrase a definition. It struggles to replace first-hand experience, a specific point of view, original data, or a detailed worked example from someone who has actually done the work. Generic explainers are the content most exposed to summarization. Content built on your own expertise, your own process, and your own judgment is the least exposed, and it is also the content most likely to be cited as a source rather than silently absorbed.

Invest in the pages buyers actually land on. If your informational traffic is shrinking, the share of your visitors arriving on service and location pages is growing. Those pages carry more weight than they used to, and most firms underinvest in them. Your service pages, not your homepage, are where buyers decide.

Build demand that does not depend on a results page. Branded searches, direct visits, referrals, and an email list are all traffic that an AI summary cannot intercept. A firm people search for by name is far less exposed than one that relies on ranking for generic questions.

Make sure you are eligible to be cited. Check that important pages are not blocked from snippets, that they are indexed, and that each one answers its central question clearly and early. This is not optimization for a special system. It is making sure the ordinary system can use your content.

A note on where this is heading

Everything above describes the situation as of the research available in 2025 and 2026, and this is the least stable topic in search. AI features are appearing on more queries over time, AI Mode is expanding, and click behavior will keep shifting. Revisit your own data quarterly rather than relying on any article, including this one, as a fixed picture.

What will not change is the underlying logic. AI search is very good at answering questions and very poor at replacing the people who do the work. If your marketing depends on being the answer to a question, it is exposed. If it depends on being the firm buyers choose once they understand the question, it is in a far stronger position than the traffic chart suggests.