A business owner opens Google Search Console and finds something that does not add up. Impressions are flat or even up. Average position is the same as last year, maybe slightly better. Clicks are down by a third. Nothing on the site broke, no penalty landed, and the competitor who supposedly overtook them is nowhere to be found.
The ranking did not move. The results page did.
If your website traffic is dropping while your positions hold, you are almost certainly looking at the effect of AI answers sitting above your listing. Your page is still there. It is just no longer the thing people read.
Why is my website traffic dropping when my rankings haven’t changed?
Because fewer people click at all. Google now answers many questions directly on the results page with an AI summary, so your listing gets seen and skipped. You keep the ranking and lose the visit. The industry calls this zero-click search.
It is also spreading into query types that used to be dependable, and the scale of it is now measured rather than guessed.
The Pew Research Center tracked 68,879 Google searches from 900 US adults through March 2025. When an AI summary appeared, users clicked a traditional search result 8% of the time. When no summary appeared, they clicked 15% of the time (Pew Research Center). Clicking a source link inside the AI summary itself happened just 1% of the time.
Read those two numbers next to each other. The presence of an AI summary roughly halved the chance that anyone visited a website at all, and the summary’s own citations did almost nothing to make up the difference.
Ahrefs found the same pattern from the other direction, and their two studies show which way it is heading. In April 2025, comparing 300,000 keywords before and after the US rollout, they measured a 34.5% drop in clickthrough rate for the page sitting in position one when an AI Overview was present (Ahrefs). They reran it on December 2025 data and published the follow-up in February 2026. The gap had widened to 58% (Ahrefs).
In eight months, the cost of having an AI Overview above your listing nearly doubled. Position one used to be the finish line. Now it is a seat in the audience.
Not every search lost the same amount
This is the part that changes what you should actually do, and it gets skipped in most of the panic coverage. AI summaries do not appear evenly. They cluster hard on informational questions and stay away from queries where somebody is about to spend money.
| What someone is searching | Does AI answer it now | Where your effort belongs |
|---|---|---|
| “what is a CDN”, definitions | Almost always | Accept the click loss, play for the citation |
| “how to fix a slow site” | Often | Answer in one line, then go deeper than a summary can |
| “best CRM for small teams” | Increasingly | Original data, named tradeoffs, real opinions |
| “web developer near me”, pricing, hire | Less often, but rising | This is your revenue. Protect these pages first |
The businesses hurting most right now are the ones whose entire content strategy was built on the top row. Thin definitional posts, written to catch a question and hand off a visitor, are exactly what a summary replaces best.
The businesses barely feeling it are the ones whose traffic was already commercial.
That distinction is worth an hour of your time this week. Open Search Console, sort your pages by clicks lost since last year, and check whether the bleeding is concentrated in informational content or spread across pages people used to arrive on ready to buy. If it is the first, your problem is a content mix problem, not a ranking problem, and no amount of on-page tweaking to those posts will bring the clicks back.
Getting cited is the new getting clicked
There is a second-order effect worth understanding. AI summaries are assembled from sources, and being one of those sources still carries value: brand exposure, the occasional high-intent click, and increasingly, being the answer when someone asks ChatGPT or Perplexity for a recommendation by name.
Pew found the three most commonly linked sources in AI summaries were Wikipedia, YouTube and Reddit. That tells you something uncomfortable and useful at once. These engines lean on sources that are structured, heavily referenced, and full of specific claims rather than marketing language.
What actually gets a page pulled into an answer:
- Concrete numbers with a source and a year attached, rather than words like “significantly” or “much faster”
- A direct answer in the first two sentences under the heading, short enough to lift out whole
- Claims that stand on their own without needing the paragraph above them for context
- Real citations to primary sources, which is how these systems judge whether you are worth trusting
- Clear identity, meaning a named author, a real business, and structured data that says who you are
Notice that none of that is a trick. It is the same thing an editor would ask of a decent trade publication, which is roughly the standard these systems have quietly imposed on everyone.
How do you get your business cited by AI search?
Publish specific, verifiable claims that an engine can lift as a complete sentence. Use numbers with sources and years, answer each heading’s question directly in under 50 words, cite primary research, and make your business identity explicit with structured data. Vague, unsourced pages do not get cited.
What still brings people to your site
The commercial end of search is holding up, and that is where the strategy resets to something familiar.
When someone searches for pricing, a comparison of two vendors they are choosing between, or a service in their area, they are not looking for a paragraph of summary. They want to evaluate somebody. Those searches still produce clicks, and the pages receiving them matter more now than they did when informational traffic was padding the numbers.
Which raises the obvious follow-up: what happens when they land? A smaller, higher-intent stream of visitors is unforgiving of the things that were survivable before. Confusing navigation, no clear next step, a phone number buried three scrolls down. I have written before about the design mistakes that quietly cost businesses enquiries, and every one of them costs more in 2026 than it did in 2024, simply because there are fewer chances to get it right. Load speed sits in the same category, which is why Core Web Vitals still deserve attention even as the traffic mix shifts.
Fewer visitors means each one has to count for more.
The honest bottom line
Traffic falling while rankings hold is not a failure of your SEO. It is a change in what a search result is for. The volume that used to arrive from people casually looking things up has moved into the answer box, and a good part of it is not coming back.
What remains is worth more per visit than what left. So stop measuring the top of the funnel by raw sessions, start tracking enquiries and conversions from the commercial pages, and rebuild your informational content to be quotable rather than merely rankable. One deeply sourced page with real numbers now outperforms ten thin posts that a summary can replace in a sentence.
If your traffic dropped this year and you want a straight answer on whether it is AI search, a technical problem, or something else entirely, send me a message on WhatsApp at +60 17-427 2807 and I will look at your Search Console data with you.