AI Overviews now appear on 88% of commercial searches: what that does and doesn't mean for your traffic

TechScripts Nepal · Kathmandu · October 26, 2026 · 6 min read

If you work in marketing, you've probably heard two opposite claims in the last few months: that AI-generated answers are about to wipe out organic traffic, and that none of it matters. A round of studies published around October 22 gives more evidence than usual, and the honest summary is that it supports neither extreme. Here's what the data actually shows, with its limits.

What the studies found

PPC Land's October 23 roundup pulls together several pieces of research. The key ones:

AI Overviews are now nearly everywhere on commercial queries. Semrush tracked 400 US commercial keywords across banking, software, and ecommerce from September 2025 to August 2026. The share of keywords showing an AI Overview rose from about 46% to 88%, and to about 97% when weighted by search volume. For queries containing "reviews," the share went from roughly 30% to 89%.

AI assistants send very little traffic by comparison. One analysis of 12 months of Google Analytics data from 100 sites found ChatGPT sessions were about 0.62% of the volume Google organic search delivered. ChatGPT visitors converted at a higher rate, but Google still drove about 96.5% of conversions. A separate analysis found organic search delivered 39 times the traffic of AI assistants.

The link between "AI visibility" and other traffic is unclear. One researcher looking at 12 sites couldn't find a reliable relationship between being visible in AI answers and branded search or direct traffic. As PPC Land notes, that's an absence of evidence, not proof there's no effect.

What the studies don't show

It's worth being as clear about the gaps as the headlines.

• No click data in the Semrush figures. The 88% number measures how often an AI Overview appears, on US desktop only. It says nothing directly about how many people click.
• Small samples. PPC Land points out that the Semrush data covers fewer than 27 keywords per industry-and-modifier combination, and the traffic studies use 100 sites or fewer.
• Vendors with a stake. Semrush sells the tools that its recommendations call for. That doesn't make the data wrong, but it's context.
• Measurement itself is shaky. Ahrefs has said Google alters somewhere between 5% and 20% of the results it serves to suspected SEO scrapers. Other vendors have reported different effects on their ranking data, from under 1% to over 25%. If your rank-tracking tool is partly seeing altered results, the picture it gives you is partly altered too.

The studies fit together uncomfortably, as the article puts it: AI answers are widespread, AI referrals are small, and the tools used to measure all of this are less trustworthy than they appear.

You may have seen specific click-through-rate drops quoted elsewhere, such as top results falling to low single-digit percentages. We couldn't verify those against an underlying study, so we're not repeating them here.

What Google itself says to do

Google's own documentation on AI features is plainer than the marketing around it. Its guidance, as summarized in coverage of the documentation, is that AI Overviews and AI Mode run on the same ranking and retrieval systems as traditional Search, and that there's no separate optimization framework you need specifically for them. To appear, a page has to be indexed and eligible to show in Search with a snippet, and its content has to be publicly crawlable. Structured data still matters for the Search features it supports, but it isn't a special "AI switch."

You can read the primary pages yourself: Google's guide to optimizing for generative AI features and AI features and your website.

What this means in practice

Taking the evidence at face value, with the caveats, a sensible plan looks like this.

1. Don't panic, and don't ignore it. Organic search is still, by these numbers, the larger source of traffic and conversions by a wide margin. But if AI Overviews are on most commercial results, the value of ranking first is probably shifting, as Semrush itself acknowledges.

2. Measure with your own data first. Use Search Console and your analytics to watch clicks, impressions, and conversions on the pages that matter to you, rather than relying only on rank trackers. Look at branded search volume and direct traffic over time, since being known is the thing that holds up when click patterns change.

3. Get the fundamentals right. Pages that are indexed, crawlable, fast, clearly written, and demonstrably authored by people who know the subject are what Google's own guidance treats as the foundation. That's not new advice, which is exactly the point.

4. Answer real questions directly. Content that states clear, specific, verifiable answers, such as definitions, comparisons, step-by-step explanations, and honest limitations, is useful to a reader and easy for any system to quote accurately. A proper FAQ section and accurate structured data (validated for the features they support) are reasonable investments.

5. Be wary of "AI visibility" shortcuts. Anything that promises a special trick for appearing in AI answers deserves skepticism, particularly paid mentions or manufactured brand signals. Google's guidance doesn't describe such a trick, and the cost of being wrong lands on your budget.

6. Don't put all your traffic in one place. If a search engine can answer the question on its own results page, the sites that fare best tend to be the ones people come to directly: a recognizable brand, an email list, a community, or a tool people return to.

How we approach it

We applied the same thinking to our own site: a clear, structured explanation of what we do, a real FAQ with matching structured data, Organization and Service structured data, and a blog that publishes sourced analysis rather than filler. We also publish an llms.txt file. To be candid, there's no good evidence it moves Google rankings, and we treat it as a low-cost extra for AI crawlers whose behavior is less documented, not as a growth lever.

Search is changing quickly, and the right response depends on your market, your content, and what your analytics show. If you'd like help auditing what your site actually gets from search today, restructuring content so that it answers real questions well, and setting up measurement you can trust, that's the work of our growth marketing, SEO, and content teams at TechScripts Nepal. We'd be glad to talk it through, whether or not it leads to a project.