Your Rankings Haven't Moved. So Why Is Traffic Falling?
The short answer: your page is still ranking, but fewer people need to click it. An AI-generated summary now sits above your result and answers the question completely, so the visit never happens. Your content is being read more than ever — inside someone else's interface.
This is the defining symptom of AI search, and it is the one that breaks every dashboard built for the previous era. Here is how to confirm it is happening to you, which pages it hits hardest, what the actual numbers look like, and what is worth doing about it.
What the data actually shows
This is not a hunch. Pew Research Center analyzed 68,879 real Google searches and found users clicked a traditional result 8% of the time when an AI Overview appeared, versus 15% of the time when it did not — and only 1% of visits to a page with an AI summary resulted in a click on a link inside that summary. Ahrefs ran the same test twice on 300,000 keywords: in April 2025 the top-ranking result lost 34.5% of its clicks on AI-Overview queries; by December 2025, that had grown to a 58% reduction.
Those two studies alone explain most support tickets that start with "our rankings haven't moved." The ranking is not the thing that changed. The page above it did.
The evidence, in one place
How to confirm it is AI Overviews
Traffic can fall for a dozen reasons. Before you rewrite anything, rule out the boring ones: seasonality, a tracking change, a redesign, a competitor outranking you.
The signature of an AI Overview is specific and easy to spot in Search Console. Impressions hold steady or rise. Clicks fall. Average position barely moves. That combination means the same number of people are seeing your listing, in roughly the same place, and choosing not to click.
Then check the query itself. Search it in an incognito window and look at what sits above the organic results. If there is a generated summary answering the question in four sentences, you have found your answer — and quite possibly two of your own sentences inside it.
Which pages get hit hardest
The damage is not evenly distributed, and it is no longer a matter of intuition — Seer Interactive's 2026 study, tracking 53 brands across 5.47 million queries and 2.43 billion organic impressions, measured exactly which query shapes trigger an AI Overview and how often.
Definitional and factual pages lose the most. "What is X", "how long does Y take", "what does Z cost" — questions with one correct answer that fits in a paragraph. Comparison queries ("X vs Y") trigger an AI Overview 95.4% of the time in Seer's data. Question-format queries trigger one 85.9% of the time. Once the summary states the answer, there is nothing left to click for.
How-to content loses selectively. A three-step process gets absorbed. A twelve-step process with screenshots does not, because the summary cannot carry it.
Commercial and comparison pages hold up best — with one exception. Comparison queries have the highest AI Overview rate of any category measured. Someone deciding between two named products still wants pricing and specifics no four-sentence summary can supply, but the summary shows up anyway and takes the first click before they get that far.
Which query types are most exposed
If your traffic loss is concentrated in the query types below, that is not a content quality problem. It is a content type that AI answers absorb by default.
The citation math
Here is what actually happens to the clicks, in real numbers. Seer's data translates cleanly to a per-impression basis: for every 1,000,000 organic impressions on an informational query, a page with no AI Overview at all earns roughly 33,500 clicks. The same query, with an AI Overview present and your page cited as a source inside it, earns roughly 20,743 clicks. If the Overview appears and you are not cited, that number drops to roughly 9,445 clicks — a 72% loss compared to the no-Overview baseline, even though your page may still rank exactly where it always did.
Put a real scenario on it. A SaaS blog post ranking #1 for "what is a CRM" that used to pull 500,000 monthly impressions and convert at a 3.35% click rate — roughly 16,750 clicks — will, once an AI Overview appears on that query and cites the post, fall to something closer to 2.07% CTR, or about 10,350 clicks. If the Overview appears but does not cite the post, the same impressions might return closer to 4,700 clicks. Same ranking. Same content. A 38%-to-72% swing in traffic depending entirely on one variable: whether the Overview names you.
The 2026 twist: it is not a straight line down
The honest update, and the one most 2025-era advice misses: the decline did not continue indefinitely. Seer's most recent data shows organic CTR on AI-Overview queries hit a floor of 1.3% in December 2025, then climbed back to 2.4% by February 2026 — an 85% rebound in two months, though still well below pre-Overview levels. Being cited in an Overview now delivers roughly 120% more organic clicks per impression than not being cited on the same page — but even a citation still underperforms a no-Overview result by about 38%.
The practical read: this is not a one-way door closing forever. It is a new, still-shifting equilibrium where citation status is now the single biggest lever you have — bigger than rank position itself, on any query where an Overview shows up.
Three honest options
There is no way to reverse this, so the useful question is what to do instead.
1. Be the source that gets summarised
If the summary is going to exist regardless, the outcome worth having is being the page it draws from and credits. That means answer-first structure, self-contained paragraphs, and definitions that survive being lifted out of context. Given the citation math above, this is no longer a soft brand-awareness argument — it is the difference between roughly 20,700 and roughly 9,400 clicks per million impressions on the same query.
2. Move investment toward content summaries cannot absorb
Original data. Comparisons with specifics. Interactive tools. Anything where the answer is not a paragraph but an experience. This is not a trick; it is a straightforward reading of what still earns a visit — and it lines up with Seer's finding that no-Overview queries are themselves becoming more valuable, since the remaining click-through pool skews toward higher-intent users who were never going to be satisfied by a four-sentence summary.
3. Change what you measure
If sessions no longer describe your visibility, tracking sessions alone tells you less each quarter. Presence inside answers becomes the metric that reflects reality — how often you are named, on which platforms, for which questions.
Two things not to do
Do not block AI crawlers in a panic. It is tempting, and it does stop your content being used in summaries. It also removes you from the answers entirely, including the ones that would have named your brand favourably. That is a serious decision with a real cost, not a quick defensive measure.
Do not rewrite everything longer. The instinct when traffic falls is to add depth. But a summary that absorbed your answer did so because the answer was clear, not because the page was short. Padding a page does not make it harder to summarise — it makes it worse to read.
The uncomfortable part
Some of this traffic is not coming back, and no amount of optimisation returns it. A question with a single factual answer will increasingly be answered where it is asked.
What is still available is the association. When someone reads a summary that names you, or cites your page, you have been present at the moment of research even though your analytics recorded nothing. That is genuinely harder to measure and genuinely still valuable — which is exactly why measurement has to change before strategy can.
What to check this week
- In Search Console, sort your top pages by click decline over the last six months.
- For each, compare impressions and average position over the same period. Look for the flat-position, falling-clicks pattern.
- Search the main query for those pages in an incognito window. Note which return an AI Overview.
- For each Overview, check whether your page is among the cited sources.
If you are cited, you are still in the conversation and the loss is traffic, not visibility — and per the numbers above, you are likely still keeping roughly two-thirds of what a citation-free page in the same spot would earn. If you are not cited, the loss is both traffic and visibility, and that is a different problem with a different fix.
