Google AI Overviews are eating clicks. What still works for small websites in 2026
When an AI summary appears, people click results about half as often. Here is what the data actually says, which pages lose the most, and the seven moves we are making on our own site.
You did everything right. The page ranks on page one, maybe even first. Traffic still slid.
You are not imagining it. Google now answers many questions directly at the top of the results, and fewer people scroll down to click. Here is what the numbers say, who gets hurt most, and what a small website can still do about it.
What changed
Google's AI Overviews put a generated summary above the normal results for a growing share of searches. For quick questions, that summary is often the whole answer, and the search ends there.
The pain point, in numbers
Two studies are worth knowing, because they measure different things:
- Pew Research Center tracked real browsing. When an AI summary appeared, users clicked a traditional result in 8% of visits, against 15% when there was no summary. Only 1% of visits included a click on a link inside the summary itself. People also ended their browsing session more often after a page with a summary: 26% against 16%.
- Ahrefs compared 300,000 keywords and found a 58% lower click-through rate for the top-ranking page when an AI Overview was present. That was up from 34.5% in its earlier version of the study. Ahrefs is clear that this is a correlation, not a controlled experiment.
Different methods, same direction: when the summary shows up, clicks drop by roughly half.
Who loses the most
The pages that suffer most answer a question that fits in a paragraph: definitions, quick how-tos, "what is X" explainers. If an AI can summarise your page in three sentences without losing anything, it will.
The pages that hold up better are the ones a summary cannot replace:
- First-hand experience. "Here is the exact mistake we made and the command that proved it" is hard to compress.
- Decision-stage content. Comparisons, pricing, "X vs Y for this specific job." People still click when money or risk is involved.
- Original data. Your own tests, measurements or survey results. Summaries need a source to cite, and you can be it.
- Branded searches. Someone looking for you by name still clicks through to you.
Fixed or not?
Not fixed, and it will not be "fixed" in the old sense. Google has argued that overall traffic to the web is not falling as sharply as these studies suggest, and it keeps adding links inside AI answers. But the incentive is clear: the easiest answers will stay on Google's page. The opportunity now is to be the page the answer is built from, and to own an audience that does not come only from search.
What we are doing about it on our own site
These are the moves we are applying to gigasonet.com, starting with this blog. They work for any business website, not only ours:
- Lead with experience. Every post includes something only we can say, like our App Links fix with the real error messages and commands.
- Answer first, then go deep. Each post opens with a short "In 30 seconds" summary. It helps readers, and it gives AI systems a clean, quotable answer to cite with a link.
- Structure everything. Clear H2 sections, tables for comparisons, and an FAQ marked up with structured data, so machines understand the page as well as people do.
- Name a real author. Posts carry a named author with a public profile. Anonymous content is easier to ignore and harder to trust.
- Target decisions, not definitions. "Which model should you use for this job" beats "what is an LLM".
- Build a second channel. We share each post on LinkedIn with the key insight inside the post itself, so the post is useful even if nobody clicks.
- Watch the right metric. In Search Console, compare impressions with clicks per query. A query with rising impressions and falling clicks is probably being answered by an AI Overview, and that page needs more depth or a sharper angle.
The result you should aim for
Rankings alone are no longer the scoreboard. A healthy small site in 2026 looks like this: fewer but better visits from search, mentions and citations inside AI answers, and a growing group of people who come back directly or through social posts. That traffic converts better, because the people who click past a summary are the ones who actually want more.