For most of SEO’s history, “position” offered a convenient mental model: position one was the first organic result, position two sat below it, and the first ten results roughly described the first page of Google. That model was never perfectly literal, but modern search results have stretched it far enough that Google itself is questioning how useful the old 1–10 vocabulary remains.
Google’s John Mueller addressed the issue while discussing how Search Console counts generative-search impressions. As Search Engine Roundtable reported on September 10, Mueller said position for these experiences is difficult to express in a useful way and that Google currently tracks it as a block for many search features. He added that today’s result pages offer so many ways to interact that the old positions one through ten are increasingly difficult to map for site owners.
That is not the same as saying rankings are dead. Google still calculates position in its standard Search performance reporting, and moving from weak organic visibility to strong visibility obviously matters. The change is in what a rank number can reliably describe. A page labeled “position 3” inside analytics no longer guarantees that a user experienced it as the third discrete answer on a simple vertical list.
AI Overviews expose the problem with linear position
Google’s official Search Console documentation already treats position as an approximation of where a search-result element appears relative to other elements. Compound features complicate that further: if a carousel occupies one position, multiple links inside that carousel can inherit the same position. The documentation explicitly warns that an identical numeric value can correspond to different visual situations depending on the result type and device.
AI Overviews make the abstraction particularly visible. Google says an AI Overview occupies a single position in the search results and every link inside that overview is assigned that same position. One source may appear prominently in the initial response while another requires scrolling or expansion, yet “position” does not become a separate ordinal ranking of every citation within the generative block. The unit being positioned is the AI Overview as a search feature, not a traditional stack of ten independently ranked blue links.
This distinction matters for competitive analysis. If several publishers are cited inside an AI Overview occupying the first search-result position, a conventional rank-tracking mindset invites the wrong question: which one is “number one”? Search Console’s position model does not answer that. The more useful questions concern whether a source was exposed, how the user could encounter it, whether it earned an impression and ultimately whether that visibility generated a click or another measurable outcome.
The new Generative AI report deliberately does not turn citations into mini-rankings
Google’s dedicated Generative AI performance report, rolled out globally from August 31, 2026, currently covers AI Overviews and AI Mode. It lets site owners examine generative-search impressions over time and break them down by page, country, date and device. What it does not currently provide is a separate position metric for the site’s placement within those generative experiences.
Mueller’s explanation connects that omission to the underlying measurement problem rather than suggesting Google simply forgot to expose a field. The company can count the generative feature as a search-result block, but translating every citation and interactive element into a linear position that site owners can compare meaningfully is much harder. He invited feedback on what position tracking would actually be useful, which suggests the measurement model is still evolving alongside the interface.
There is another important reporting boundary. Google says the generative AI report contains data from the Web search type in the regular Performance report; it is a specialized view, not a second independent pool of impressions to add on top of ordinary Web totals. For marketers building dashboards, that means generative visibility should be analyzed as a segment of Google Search performance rather than double-counted as a separate acquisition channel.
“Average position” was already an abstraction before generative search
The classic metric has always compressed complexity. Search Console defines average position using the highest position occupied by a link from the property for an impression and then averages those values across appearances. Personalization, location, device, query variation and SERP features can all make the result seen by one user differ from the reported average. Google’s documentation consequently recommends monitoring position trends while avoiding overly literal conclusions from the absolute number.
Featured snippets, knowledge panels, image results, video carousels, People Also Ask and other modules had already weakened the idea of a ten-link ladder. A result technically below a large feature can require far more scrolling than its ordinal number implies, while an element on the secondary side of a desktop page can carry a position that sounds comparable to a result in the main column even though the user experiences it differently. AI Overviews amplify a measurement problem that predates generative search rather than creating it from nothing.
AI Mode pushes the interface even farther from the old page metaphor. Google says position there follows its standard methodology, with elements such as carousels and image blocks generally counted according to their normal rules. Follow-up questions are treated as new queries. In practice, a user can move through a conversational research session whose sequence is better described as a series of generated interactions than as page one followed by page two.
SEO reporting needs to separate rank from actual prominence
The practical response is not to delete rank tracking. Organic position remains valuable for diagnosing traditional search visibility, identifying competitive movement and understanding whether pages are becoming more or less discoverable. But it should be treated as one dimension rather than a universal proxy for what users see.
For modern SERPs, impressions and clicks become essential companions to position. Generative-search impressions can reveal whether a site is appearing inside AI features even when no meaningful citation rank is available. CTR can show whether visibility translates into visits. Page-level reporting can identify which URLs Google selects for generative experiences, while external SERP monitoring can add context about the modules surrounding an organic result and the amount of screen space they occupy.
This also changes the language executives should expect from SEO teams. “We rank number three” is precise mathematically but can be incomplete strategically. A better report might say that a page maintains a top organic position while an AI Overview now occupies the dominant area above it, or that the brand is repeatedly cited inside generative answers despite having weaker conventional rankings for the same topic. Those statements describe user-visible search competition more accurately than a single ordinal number.
Mueller’s comment therefore marks an evolution in measurement rather than the end of ranking. Google has not abolished positions one through ten, and Search Console still reports average position for ordinary Search performance. What has changed is the SERP around that number. AI Overviews, citations, carousels, videos, panels and conversational follow-ups create multiple forms of prominence that cannot all be reduced cleanly to a vertical list.
The classic Top 10 will remain useful shorthand, particularly for traditional organic results. It is simply no longer a complete map of Google Search. For SEO teams in 2026, the stronger question is not only “What position are we in?” but “Where, in which search experience, are users actually encountering us—and what happens next?”