TL;DR

1. Rising impressions with falling traffic means AI systems are answering the query before users reach your link.
2. The real question is not whether you rank, but whether you appear inside AI answers, and why competitors do while you do not.
3. Content that is long and detailed but poorly structured is invisible to AI extraction, even when it ranks well.
4. Informational pages are the hardest hit; bottom-funnel pages survive because AI does not intervene on transactional queries.
5. The fix is to restructure existing content for extraction, not to publish more of it.

Why is your traffic falling, but your rankings are stable?

This is the pattern showing up across dashboards in 2026. Rankings are stable. Impressions are growing. Traffic is down. The natural response is to look for a penalty or a crawl error. Neither explains it.

Google is now surfacing AI Overviews for the exact queries where your strongest pages rank. The user sees a complete answer at the top of the page. Your link sits below it. The impression is recorded. The click never happens.

The diagnostic question that most teams ask first is the wrong one. Checking whether your position changed is not useful when your position is held. The useful question is whether an AI Overview is now appearing for your key queries and whether your site is being cited within it or excluded entirely. This is a different problem from a search console CTR drop where impressions rise but clicks fall, and it requires a different first move.

Why are competitors inside AI answers, and you are not?

When AI Overviews appear, and your competitors are cited while you are not, the gap is rarely due to domain authority or backlink volume. It is about how the content is structured.

Google's citation selection inside AI answers favors pages that are easy to extract from. That means pages where the first sentence under each heading is the answer, not the lead-in to the answer. 

Pages where key claims are not buried inside long paragraphs. Pages where headers are phrased as questions, and where tables, numbered lists, and summary blocks make individual facts portable.

The page currently winning the citation is not necessarily the most comprehensive one. It is the most extractable one. Your content may already contain better information. If it requires context to make sense, the AI will pass over it. We have unpacked this exact gap in a separate scenario on why brands stay invisible in AI Overview answers despite strong Google rankings.

What makes content hard to extract?

Extractability is the distance between ranking well and being used by AI systems. A page can hold position 2 on a high-volume query and generate zero citations because its answers are embedded mid-paragraph, its headers describe topics rather than answer questions, and its structure rewards a human reader who scrolls but gives an AI system nothing clean to lift.

Here is what the extractability gap looks like across common content types:

Restructuring for extractability does not mean making content shallower. It means placing the most portable version of each answer at the surface, with depth available beneath it for the reader who wants it. 

The pages that are being cited most consistently in AI answers tend to share three structural traits: they front-load the answer, they break long arguments into self-contained sections, and they use formatting elements that signal hierarchy without relying on the reader to follow the prose.

Why are informational pages hit the hardest?

The split between informational and bottom-funnel performance is consistent across categories in 2026. Pages targeting transactional queries are holding their clicks because AI Overviews do not typically appear when the user intent is to take an action. Pages targeting informational queries are losing traffic because these are the exact queries AI is designed to answer without a visit.

This creates a portfolio decision about what informational content is actually for. If a how-to page is going to generate an AI citation rather than a click, its value shifts from traffic source to brand visibility. 

The metric that matters for that page is no longer sessions. It is whether your brand name appears in the answer and whether users who see that citation later search for you directly. Many teams discover this only after auditing why most of their blog posts are getting no organic traffic and finding the loss is concentrated entirely at the top of the funnel.

Informational content that cannot earn a citation and cannot convert a visitor has a third option: redirect it toward owned conversion paths. Free tools, email capture, and original data all give users a reason to engage that does not depend on a recurring click from Google.

What to do when users land but leave immediately?

Users who click through an AI citation are arriving pre-informed. They already read the overview. They clicked because they wanted depth, a specific detail, or confirmation from a source. If your page opens by restating what the AI summary already told them, they leave within seconds.

The page now has to earn attention in the first scroll. That means opening with what the AI did not provide: an original data point, a specific example, a framework, or a perspective that requires human judgment. The introductory paragraph that restates the topic and previews the article structure is now a liability. It signals to a pre-informed reader that nothing new is coming.

The pages that hold a pre-informed reader tend to do one of three things in the first 200 words: they make a claim the AI summary did not, they present a specific example with named details, or they reframe the question itself in a way that opens up a sharper angle. Anything less and the reader has no reason to keep scrolling.

How to explain the traffic drop to leadership without sounding defensive?

When traffic falls, and leadership asks whether SEO is broken, the framing of the answer shapes the budget decision that follows.

The accurate answer is not that SEO is failing. It is that informational content is being absorbed by AI at the top of the funnel, while branded and transactional search remains intact. These are structurally different situations that require different responses, not a single channel verdict.

The clearest way to present this is a query-level split: which queries still deliver clicks and pipeline, which are now generating citations and brand visibility without clicks, and which have lost value entirely and should be deprioritized. 

Teams that present this breakdown protect and often grow their budget. Teams that present an overall traffic number without context lose it. The reframe leadership actually needs is that the role of SEO has expanded, not declined: the team is no longer just chasing rankings, it is managing presence across an answer layer that did not exist eighteen months ago.

Reclaim visibility where it actually lives now

Rankings still matter in 2026, but they are no longer sufficient. Visibility now lives inside AI answers, inside citation blocks, and inside the first scroll of a page that proves value before the user decides to leave. The brands recovering traffic are not the ones publishing the most content. They are the ones who restructured what they already have, rebuilt their authority signals at the entity level, and shifted their definition of success from clicks to presence.

The work ahead is less about producing and more about diagnosing: which of your ranking pages are being passed over for AI citations, which are losing engagement after the click, and which can be repositioned toward queries where organic still delivers.