For decades, one of SEO’s most intuitive questions was also one of its simplest: does a page rank first, third or tenth? Google’s John Mueller now says that familiar model is increasingly difficult to translate into something useful because a modern search results page no longer behaves like a clean vertical list of ten links.
Responding to a discussion about how generative AI impressions are measured in Google Search Console, Mueller said that search pages now give users many different ways to interact, making the old “position 1–10” concept hard to map. His comments were reported by Barry Schwartz at Search Engine Roundtable on September 10 and add important context to how Google is approaching measurement for AI Overviews and AI Mode.
The key point is not that Google has eliminated ranking positions. Search Console still reports average position, including defined position behavior for generative search features. The shift is more subtle: a single number is becoming less capable of describing how visible a result actually is when users encounter AI answers, carousels, expandable modules, images and other interactive elements before or alongside conventional links.
Google says the old position model is becoming harder to make useful
Mueller’s comment came in response to questions about Search Console’s treatment of generative AI features. He said Google has tried to document impression and click measurement clearly, but position is difficult to represent in a way that remains useful for site owners.
He explained that Google is currently tracking these generative features in the same general way it handles many other search features: as blocks. Position is not broken out separately inside the Gen-AI Performance report. Mueller then invited feedback from site owners about what kind of position reporting would actually help them.
That invitation is revealing because it exposes the analytical problem rather than pretending the old metric translates perfectly into a new interface. If several links sit inside one AI-generated block, the question “What position did my URL rank?” no longer has the same meaning it had when ten blue links occupied ten obvious slots.
Search Console still has a position metric
It is important not to overstate Mueller’s remarks. Google has not announced the end of average position, and its Search Console documentation on impressions, clicks and position still defines the metric across Google Search.
Google describes position as an attempt to show approximately where a particular link appeared relative to other search results. The company itself warns that the value is complex and can be misleading if its subtleties are not understood. Search result pages contain many kinds of elements, and different elements follow different rules when Google assigns a position.
For AI Overviews, Google says the entire overview occupies a single position in the search results and every link inside that overview receives the same position. That means several cited publishers can all be recorded at the position of the containing AI Overview even though their visual prominence, order and likelihood of attracting a click may differ inside the block.
For AI Mode, Google says position follows the same general methodology used for a Google Search results page, with carousels and image blocks handled according to the standard rules for those elements. Follow-up questions in AI Mode are treated as new queries for impression, click and position reporting.
One block can contain many very different visibility experiences
This block-level logic explains why position becomes less intuitive. Imagine an AI Overview occupying the first recorded search position. A publisher whose link is immediately visible in the initial answer and another whose link appears only after expansion can inherit the same block position under Search Console’s methodology.
From a reporting perspective, both URLs may therefore share a positional value while experiencing very different real-world exposure. The same challenge appears in carousels, image modules and other rich results where multiple destinations live inside one search element.
Traditional rank tracking encourages a mental model in which position three is necessarily below position two and above position four in a stable stack. Modern search can instead place several interactive surfaces around one another, some of which expand, scroll horizontally or open additional layers. The numerical position remains a reporting convention, but it no longer maps cleanly to the exact attention available to every URL inside the interface.
The Gen-AI Performance report avoids a separate position breakdown
Google’s newer Gen-AI Performance report in Search Console reinforces that shift in emphasis. The report covers supported generative AI features in Google Search, including AI Overviews and AI Mode, and is designed to show how a site’s organic impressions from those experiences change over time.
Google made the generative AI performance insights globally available beginning August 31, 2026. Site owners can use the report to examine impression trends, identify pages receiving more or fewer generative-search impressions and break visibility down by dimensions such as device and country.
What the report does not provide is a dedicated position dimension that tries to translate those AI appearances into a separate generative ranking. Mueller’s explanation makes the rationale clearer: Google does not currently believe a simple positional representation would necessarily produce a useful measure for site owners.
Average position was already an approximation before AI
The weakness of position as a visibility proxy did not begin with AI Overviews. Google’s own documentation has long emphasized that Search pages are complex. Featured snippets, image results, carousels, knowledge features and other modules already complicated the idea that every result could be understood as one item in a ten-link list.
Search Console reports average position across impressions, not a permanent rank. The value can vary with location, device, personalization, query context and the particular search features shown. Some elements count as positional placeholders while others do not, depending on whether they contain qualifying links and how users interact with them.
Generative AI intensifies that existing complexity. Instead of merely adding another rich-result box, an AI Overview can synthesize information from several sources into one answer, while AI Mode can support a continuing interaction in which users ask follow-up questions and encounter new sets of links.
Position and visibility are becoming different questions
For SEO teams, the practical implication is that “Where do we rank?” increasingly needs to be separated from “How visible are we?” A URL can have an apparently strong recorded position but receive limited attention inside a crowded or expandable search feature. Another result can appear numerically lower while attracting more clicks because its presentation is clearer or its context better matches user intent.
AI search makes that distinction particularly important because being cited and being clicked are separate events. A source may contribute information to an AI-generated answer and receive an impression without producing the kind of navigational behavior associated with a classic organic result.
That does not make position useless. Average position can still reveal meaningful directional changes, especially for conventional web results and when compared consistently over time. It does mean the metric should be interpreted alongside impressions, clicks, click-through rate, page-level trends and the search feature in which the visibility occurred.
SEO reporting needs to move beyond a single ranking number
The industry has already been moving in this direction. Search marketers increasingly track visibility across AI Overviews, AI Mode, featured snippets, product modules, local results, video surfaces and other search experiences rather than treating a list of ten organic ranks as a complete picture.
Mueller’s comments effectively acknowledge that Google faces the same measurement challenge. A mathematically precise position number is not automatically an analytically useful one if the interface surrounding it has changed so much that users no longer experience results as a simple ordered list.
For publishers, that suggests reporting should focus more heavily on outcomes and exposure patterns. Are impressions from generative features rising? Which pages are appearing? Are those impressions producing clicks? Does CTR change when AI features are present? Are important landing pages losing traffic even when their reported average position appears stable?
Those questions can reveal changes that a ranking chart alone may hide.
The ten-blue-links model is fading as an analytics model too
Google Search has been moving beyond ten blue links for years, but generative search makes the old abstraction harder to preserve. AI Overviews can place many publishers inside a single recorded position. AI Mode can turn one search into an interactive sequence of follow-up queries. Other rich features compete for attention in ways a linear ranking number cannot fully express.
Google’s current solution is pragmatic rather than revolutionary: keep position methodology for Search Console where it can be defined, treat complex features as blocks and avoid presenting a separate position metric in the Gen-AI Performance report when that number may imply more precision than it delivers.
That leaves SEO teams with a less comfortable but more realistic measurement model. Positions one through ten have not disappeared, and Google still records position. They simply no longer describe the entire search experience cleanly. In modern Google Search, knowing that a page was “position three” may be the beginning of the analysis rather than the answer.