It's a fair question, and I'd rather answer it with numbers than with a feeling. The client is right about the direction. The scale is what matters for planning, and the scale is smaller than the headlines suggest. Businesses that respond well are the ones that adjust their plan based on what they can measure, instead of tearing up the plan because of a trend report.
Here's what the data shows, where it's weak, and how I'd think about the budget.
Google clicks have been falling for a while, and a big part of the reason is that Google is answering more questions on the results page itself. AI Overviews, the summaries that sit above the regular links, are the most visible example.
The clearest measurement of this came from Pew Research Center. In March 2025 they tracked about 69,000 Google searches made by 900 U.S. adults. When an AI summary appeared, people clicked a traditional result in about 8 percent of searches. Without the summary, they clicked in about 15 percent of searches. Only about 1 percent of searches resulted in a click on one of the links inside the AI summary itself. Pew's sample is small and it's a snapshot from one month, so treat the exact figures with care. The direction is hard to argue with, though. A summary at the top of the page means fewer people go further.
Chartbeat, which measures traffic across a large network of publisher websites, tells a similar story from the other side. Its figures show Google Search referrals to publisher sites falling about 40 percent year over year between July 2025 and July 2026, compared with about 22 percent the year before. Google Search's share of pageviews across that network dropped from roughly 9 percent in July 2024 to about 5 percent in July 2026. The most important thing to know about Chartbeat's numbers is that they come from news and media sites. A news site depends heavily on search for traffic in a way many local service businesses do not, so you can't assume the same percentage applies to a plumber's website. It does show how hard the shift can hit when a business relies on informational content.
Similarweb's counts for the wider web paint a less dramatic picture. Their data showed U.S. organic search referrals falling from roughly 2.3 billion visits in July 2024 to about 1.8 billion in June 2025, a drop of around 22 percent in a single year. Those counts cover a different set of sites and a different period than Chartbeat's, which is why the two numbers don't match. Both are real, and neither tells you exactly what will happen to your site.
Google's position deserves a fair hearing too. Liz Reid, Google's Vice President and Head of Search, has argued that click volume from Google has remained relatively stable overall, and that people who get an AI overview still want to go further. In her words, "An AI response might provide the lay of the land, but people click to dive deeper and learn more." She has also pointed to growing interest in forums, videos, podcasts, and personal posts as a sign that people are looking for a range of sources, not only one answer.
Those are fair points. But a company that makes money from search has an obvious interest in the story it tells, and independent researchers are measuring something different. When Pew's data shows fewer clicks with an AI summary present, that's a finding a business owner should plan around, whatever the overall volume looks like.
AI assistants are sending more visitors than they were a year ago. A study of 16,661 websites with Google Search Console data, published by SEO Crawl, found AI assistant referrals grew about 45 percent year over year through August 2026. Over the same stretch, Google organic clicks for those sites fell about 15 percent.
The study also found that AI assistants made up 0.42 percent of all website sessions in August 2026, up from 0.23 percent earlier in the measurement window. That's a near doubling, and it's worth watching. It's also a very small share. Google organic still sent roughly 80 times more visits than all AI assistants combined, and ChatGPT accounted for about 83 percent of the AI referral sessions.
Chartbeat's publisher data shows the same pattern at a much smaller scale. Chatbots accounted for under 1 percent of publisher pageview referrals in early 2026, even though ChatGPT referrals rose more than 200 percent over the year covered in that report. Chartbeat's later playbook puts AI referrals closer to 0.01 percent of traffic. Those two figures don't line up, so I'd quote them as a range, not a single number.
Both studies point to a measurement problem as well. The SEO Crawl researchers note that AI-driven visits are often logged as direct traffic or as regular Google traffic, because many AI tools don't pass referral information cleanly. That means the true share is probably higher than the reported share. Your analytics may be undercounting AI visits, which is a reason to look at your own data more closely. If you want a clean starting point, our analytics and conversion rate optimization service at https://vandesign.ca/services/analytics-conversion-rate-optimization-vancouver sets up tracking so AI referrals show up as their own source.
So the trend is real, the growth is fast, and the current share is small. Those things can all be true at once. A plan that only reacts to one of them will be wrong.
Here's where I'd be careful with anyone who sounds too certain. I haven't found reliable, comparable research showing whether visitors who arrive from an AI tool convert to leads at a higher or lower rate than visitors from Google organic or paid search. Some vendors claim AI visitors are more valuable. Others say the opposite. I'd want to see their methodology before believing either.
There's a logical argument on each side. Someone who asks an AI assistant a detailed question has usually done some research already, so they may arrive with a narrower question and better context. But they may also have gotten what they needed from the chat itself and never needed your site. Until you measure this on your own site, the honest answer is that you don't know yet.
That's why I'd move the conversation from channel labels to outcomes. A visit isn't a lead, and a lead isn't a customer. Track each channel all the way to a call, a form submission, or a booked appointment, and compare on that basis.
For years, the goal of SEO was simple: rank on page one and earn the click. A page can still rank well and lose the click to a summary, though. The same page might also be quoted inside an AI answer, which could send a visit or simply put your business name in front of someone comparing options. Either outcome is different from a traditional click, and you can't measure them the same way.
Pages that do well in both settings tend to share some habits. They answer the question directly near the top, use plain language, state specifics like pricing ranges, timelines, and service areas, and use structured data so search engines can tell what the page covers. Your business details also need to match across the web, including your Google Business Profile and directory listings. We wrote about that in more detail in our post on the top 50 citation sites at https://vandesign.ca/posts/top-50-citation-links, because AI tools often draw on those sources when describing a local company.
For a Vancouver business, local accuracy matters a lot. Someone asking an AI assistant to recommend a plumber in Burnaby or a designer in Kitsilano is making a local decision. If your business is missing from the sources the tool relies on, or described incorrectly, you won't get the visit at all. Our AI optimization work for Vancouver businesses, covering both AI search and generative engine visibility, is at https://vandesign.ca/services/ai-optimization-vancouver.
Speed matters here too. A slow page loses people before they read the answer, whether they arrived from Google, an AI tool, or an ad. We covered that in our piece on how speed quietly costs businesses customers at https://vandesign.ca/posts/heres-why-speed-is-the-invisible-competitor-you-are-are-losing-to.
Paid search hasn't been replaced, and I don't expect it to be soon. It does something organic search and AI answers can't do on their own: it puts your business in front of a specific person searching for a specific service, right now. A homeowner who types "emergency furnace repair Surrey" late at night is ready to act, and a well-run campaign can reach them before anyone else does.
The problem is cost and efficiency. Click prices in many local markets have climbed, and a campaign that worked two years ago may now produce expensive clicks with weak conversion. Many advertisers also find some of their spend goes to searches for their own brand name, which they would often win anyway. And a good ad sending people to a slow, cluttered page will lose money no matter how careful the bidding is.
So the fix usually isn't to drop paid search. It's to narrow it. Focus on terms with clear buying intent, pause broad terms that attract browsers, improve the landing pages, and set a target cost per lead that you actually track against. We manage pay-per-click campaigns with that approach at https://vandesign.ca/services/digital-advertising-vancouver.
I'd resist choosing one channel and abandoning the rest. A better approach is to give each channel a job and fund it accordingly.
Organic Google is still the foundation. It builds long-term visibility and remains the biggest source of search visits for most businesses. Protect it by keeping core service pages accurate, fast, and genuinely useful, and by investing in local search visibility. Our Vancouver SEO services are described at https://vandesign.ca/services/vancouver-seo-services.
AI search visibility is the growing edge, and I'd treat it as a test channel for now. Make sure your business is described accurately, strengthen the pages that answer common buyer questions, add structured data where it fits, and track AI referrals as a separate source in your analytics. Set aside a modest share of the budget to learn from it, then review the results every quarter.
Paid search handles demand that already exists. Keep it tightly focused on commercial terms, measure cost per lead, and cut what doesn't convert.
The mistake I'd warn against is shifting the whole budget into one channel because a trend report says to. The data is still early and changing quickly. A balanced plan lets you adjust based on what your own customers do. If you're building the site that all three channels send people to, our Vancouver web design work at https://vandesign.ca/services/vancouver-web-design is the place to start, and you can see recent builds in our project portfolio.
If a client asks whether Google is finished and AI is taking over, my answer is no, not yet, and not in the dramatic way the headlines suggest. Google still sends most of the search visits that matter to most businesses, even with the drop. AI is growing quickly from a small base, and it's worth preparing for now. The businesses that do well over the next couple of years will be the ones that keep their core site strong, describe themselves clearly, measure what drives real leads, and adjust spending based on evidence.
That's less exciting than a story about disruption. It's also more likely to hold up.
If you're rethinking how your marketing budget is split, or you'd like a second opinion on where your traffic is really coming from, Vandesign helps Vancouver businesses work through exactly these questions.
Pew Research Center, Google users are less likely to click on links when an AI summary appears in the results (July 2025), based on a March 2025 panel of 900 U.S. adults and about 69,000 searches. Reported by The Decoder.
Chartbeat, 2026 Publisher Playbook, and Chartbeat publisher data reported by Axios (March 2026) and Search Engine Journal (2026).
SEO Crawl, AI Traffic vs. Google Organic in 2026: A Study of 16,000+ Websites (seocrawl.ai).
Similarweb data as reported by Digiday, In Graphic Detail: AI platforms are driving more traffic, but not enough to offset zero-click search.
Liz Reid, Google VP and Head of Search, blog post on AI Overviews and click quality, as reported by MediaPost.
%402x.svg){kind=icon}