Google now answers many searches at the top of the results page, in an AI Overview or a featured snippet, and a lot of people stop there. For anyone who publishes articles, guides or docs, the visible effect is fewer clicks from search, even on pages that still rank.
That makes each remaining visit count for more. Those readers saw the summary and clicked anyway, so they wanted something more. Traffic reports show they arrived. They can't show whether the page gave them what they came for.
Below is what the research says about clicks, and how to check, page by page, whether the visitors you keep are satisfied.
The short answer
You can't win the lost clicks back by asking, but you can make sure the visits you still get are worth having. Readers who click past an AI summary usually want something it didn't give them. Ask them on the page whether it gave them that, with one Yes / No question and a follow-up about what they came for. Then read those answers next to Search Console, page by page.
How much AI Overviews and zero-click search cut clicks
Three sources measure the drop in different ways:
- Pew Research Center tracked the browsing of 900 US adults in March 2025. When a Google results page showed an AI summary, they clicked a traditional result on 8% of visits, against 15% when there was no summary. They clicked a link inside the summary on 1% of visits, and ended their browsing session after 26% of visits to a page with a summary, against 16% without one.
- SparkToro, using Similarweb panel data, found that 68% of US Google searches from January to April 2026 ended without a click to any website. That figure leaves out the Google mobile app.
- Chartbeat data in the Reuters Institute report "Journalism, media and technology trends and predictions 2026" shows Google organic search traffic to more than 2,500 news sites falling by a third worldwide between November 2024 and November 2025, and by 38% in the US. The publishers surveyed for the same report expect search traffic to fall by more than 40% over the next three years.
The studies use different panels and methods, so the numbers are not directly comparable. They all point the same way, though: fewer searches end on your page.
Who still clicks through, and what they want
A summary handles the quick question: a definition, a date, a yes or no. The searches that still reach you are more likely to be the ones the summary didn't settle. Expect readers who:
- Want the full steps, not three lines of them.
- Have a case the summary didn't cover, such as their own software version, region or recipe.
- Want to check the summary against a source before they act on it.
- Came for something the summary can't give them: a table, a tool, a worked example, an opinion with a name on it.
These are guesses from what summaries leave out, and they will differ from one site to the next. Your readers can tell you which ones hold on your pages.
Why traffic reports can't tell you if they were satisfied
When the quick-answer visits move into the AI summary, your averages change even if nothing on the page changed. Fewer short visits can push time on page up and bounce rate down. A page can look healthier in analytics while it serves fewer people, and still leave the ones who arrive without what they came for.
Search Console shows the clicks and impressions, and analytics shows what visitors did on the page. Neither records whether the reader found what they came for. For that you have to ask, on the page itself. The method for finding out which articles help readers is covered in its own guide; the rest of this post applies it to search visitors.
Ask the question Google asks itself
Google's guidance on helpful content gives site owners questions to judge their own pages by. One of them reads: "After reading your content, will someone leave feeling they've learned enough about a topic to help achieve their goal?" Readers can answer a version of it for you:
-
A Yes / No question at the end of the page
Such as "Did this page answer your question?". One tap, so it gets the most answers, and it gives each page a figure you can compare.
-
A follow-up about what they came for
A multiple choice question such as "What were you looking for?" with options like "Step-by-step instructions", "Detail for my own situation", "To check a fact", "An example" and "Something else". Every reader sees it, so the answers show what the visits you keep are for, whether the page helped or not.
-
An optional comment
Last, for readers who want to say what was missing.
Tip: Google never sees these answers, and they are not a ranking signal. They tell you whether the page works for the people who arrive.
Read the answers next to Search Console
Search Console tells you whether a page lost clicks because fewer people saw it or because fewer of those who saw it clicked. Reader answers tell you whether the remaining visitors were satisfied. Put the two together, page by page, over the same dates:
| Search Console shows | Readers mostly answer | What it suggests |
|---|---|---|
| Impressions steady, clicks down | Yes | The summary took the quick question and the page serves the rest. Don't rewrite it to chase the lost clicks. |
| Impressions steady, clicks down | No | The readers who click want more than the page gives. Read what they were looking for and add it. |
| Impressions and clicks down | Either | The page has dropped in the rankings, so a summary is not the whole story. That is a different job. |
| Clicks steady | No | Not an AI Overviews problem. Fix the page anyway. |
A page with impressions and clicks both falling has usually dropped in the rankings, often after a core update. Deciding which of those pages to update first is its own process.
Example
Example: a gardening site that lost its quick questions
Say you run a gardening site with 250 guides. Over six months, Search Console shows your guide on when to prune roses keeping its impressions while its clicks fall by about half. When you search for it yourself, an AI Overview gives the month and two tips.
Your guide ends with "Did this guide answer your question?" and "What were you looking for?". Over the next eight weeks it collects 40 answers: 15 Yes and 25 No. Most No readers picked "Detail for my own situation", and their comments name climbing roses and cold regions. The guide covers neither.
So you add a short section for each and a table of pruning months by climate. You leave the quick answer at the top as it is: the people who wanted only that have already gone to the summary. Eight weeks later you compare the same question over the same number of weeks.
Where Asklet fits
Asklet puts the short survey above on every page from one snippet and keeps the path each answer came from, so you can line results up with Search Console page by page. It shows counts until a page has 10 responses, then percentages.
It sets no cookies and uses no tracking identifiers, and readers answer without signing in. It doesn't record how a reader reached the page, and it doesn't count widget loads, so it can't give you a response rate. Take clicks and page views from your search and analytics tools for the same dates.
Your next steps
-
List the pages that lost clicks but kept impressions
In Search Console, compare two equal periods and sort pages by the change in clicks.
-
Search for their main queries yourself
Note what the AI Overview or snippet already answers, so you know what readers who click are after.
-
Ask a Yes / No question at the end of those pages
Use the same wording on every page so you can compare them.
-
Add a follow-up about what readers came for
Four or five options plus "Something else", based on what the summary leaves out.
-
Fix the pages where most readers answer No
Add what they say was missing, and leave the quick answer in place.
-
Compare the same number of weeks before and after
Check reader answers and Search Console over matching dates.
Try these steps on your own site. Asklet adds the questions to every page with one snippet. Free plan, no card needed. Compare the plans.
Start freeCommon questions
Can I stop AI Overviews from taking my clicks?
Not in any way that keeps you in Google's results. Google offers controls such as nosnippet that limit what it shows from your page, but they also limit how your page appears in ordinary results. Most sites are better off making the remaining visits count.
Do reader answers help my rankings?
No. Google doesn't see answers given in Asklet. They help you improve the page for people, which is what Google's own guidance asks site owners to judge their pages by.
How many answers does a page need before I act?
Wait for at least 10 before you look at percentages, and more before a big rewrite. A comment that points to a fact you can check, such as an out-of-date figure, needs no statistics.
Should I ask readers how they found the page as well?
You can, as a separate multiple choice question. It gives a rough, self-reported picture of where readers come from, including AI assistants that analytics often files under direct traffic.
Related
Sources
- Pew Research Center: Google users are less likely to click on links when an AI summary appears in the results (2025)
- SparkToro: In 2026, less than one third of Google searches still send a click (2026)
- Reuters Institute: Journalism, media and technology trends and predictions 2026 (2026)
- Google Search Central: Creating helpful, reliable, people-first content
More from the blog
- Privacy and law EU cookie consent changes: where the Digital Omnibus stands in October 2026 9 min read
- Privacy and law Session replay and pixel lawsuits: what website owners should know 9 min read
- Privacy and law How to make a feedback widget accessible, and what the European Accessibility Act asks 9 min read