Search Console’s position and impression numbers for AI Mode and AI Overviews are not a fixed standard, according to Google’s John Mueller, who says the metrics will keep moving as the surfaces they measure keep moving. Mueller made the comment on LinkedIn after a site owner asked how Google actually counts a ranking inside an AI Mode response. The exchange leaves the underlying counting method undocumented for now, which keeps analysts guessing at exactly what a “position” means on a generative results page.
The question came from Kenichi Suzuki, who asked whether individual citations inside an AI Mode answer are numbered sequentially, like standard web results, or whether the entire AI Mode response counts as a single block occupying position one, the way AI Overviews are currently tracked. Suzuki noted that Search Console’s own documentation does not spell this out.
Mueller did not give a numbered answer. He wrote that he expects tracking for AI Mode, AI Overviews and standard Google Search to “evolve over time” as those surfaces “evolve too.” He confirmed “there are edge-cases” in how the metrics get counted across the three formats, and he framed Google’s goal in plain terms: “The goal is not a written-in-stone absolute truth for position counting (that’s impossible), but rather to make something that’s useful for site owners in terms of understanding how their site is shown.” He added that Google will update its documentation “when / if there are significant changes.”
That is not a commitment to a specific fix or a new reporting feature. It is an acknowledgment that the current system is approximate, and that publishers comparing month-over-month position data across AI Mode, AI Overviews and classic search results are comparing figures built on inconsistent counting rules. Mueller’s own framing puts the burden on site owners: “there will be some work required on your end when it comes to interpreting, understanding, comparing these metrics.”
This is the second time in a week that Mueller has told publishers their existing measurement tools were not built for the current SERP. Search Engine Roundtable reported on September 11 that Mueller said the old position-1-to-10 framework is hard to map onto a results page built from AI Overviews and People Also Ask blocks. Read together, the two statements describe the same gap from opposite ends: one says the ranking concept itself does not translate cleanly, the other says the counting mechanism behind it is still unsettled and undocumented.
Google has not said whether a revised counting method would apply retroactively or only to future reporting periods, so teams cannot yet tell whether a jump or drop in AI Mode position reflects a real ranking change or a definition change on Google’s side. Search Console’s official help documentation, which Suzuki linked to and which Mueller did not dispute, has not been updated to address the sequential-versus-single-block question directly.
Teams tracking AI Mode or AI Overviews performance should treat current position figures as directional rather than precise, and log the date of any Search Console documentation update alongside their own historical data so a future methodology change does not get misread as a ranking swing.
Search Engine Roundtable, in a report by Barry Schwartz published September 16, 2026, first reported Mueller’s LinkedIn comments.
