Yes, for any program planning 2027 budgets. A ranking used to be a fair stand-in for demand captured: rank higher, get more clicks, book more business. With an AI answer above the results, that link has weakened badly. Move average rank to the diagnostics page and put three numbers on the scorecard instead: how often AI answers name each location, qualified search actions, and booked revenue.
That is a position an SEO team with years of rank reporting behind it could reasonably push back on, so this post makes the case from the engines' own documentation and published click studies, works through an example, and then gives the strongest objection its due. We sell AI visibility work, so check each claim against the linked source.
Why did rank stop predicting clicks?
Because the answer now often appears before the results, and buyers stop there. Three independent measurements point the same way.
Ahrefs compared 300,000 keywords with and without AI Overviews using Search Console data. Its February 2026 update, built on December 2025 data, found an AI Overview correlated with a 58% lower click-through rate for the page ranked first. Its April 2025 version of the same study had put the drop at 34.5%. It is a correlation from a before-and-after comparison, not a controlled test, but the direction held across both runs.
Pew Research looked at real browsing rather than keyword samples. Tracking 900 U.S. adults in March 2025, it found users clicked a traditional result in 8% of visits when an AI summary appeared and 15% when none did. Clicks on links inside the summary happened in 1% of those visits. In Pew's sample, 18% of all Google searches produced an AI summary.
Seer Interactive tracked 3,119 informational queries across 42 client organizations from June 2024 to September 2025. In its September 2025 update, organic CTR on queries with an AI Overview fell from 1.76% to 0.61%, a 61% decline. Queries without one fell too, from 2.74% to 1.62%, so not all of the loss is the AI answer.
None of these studies says your rank stopped mattering. They say the same rank now produces a smaller and less predictable share of clicks. A metric that moves independently of the outcome it is meant to track is a poor headline number.
Search Console's position number now hides the change
Even the rank you already report has a new blind spot. Google's Performance report documentation says an AI Overview occupies a single position in search results, and all links in it are assigned that same position. Whether you are the first source in the overview or the eighth, you get one number. Search Console also reports a URL at its topmost position on the page.
For example, suppose a location page is cited inside an AI Overview that sits at the top of the page on half its impressions, and ranks 19th in the ordinary results on the other half when no overview shows. Averaged, the report shows a position near 10, which reads as a page-one asset. In this illustrative case the page is neither: it is one of several sources in an answer most people do not click, and it is buried the rest of the time.
Google's AI features guide adds that clicks from AI Overviews and AI Mode are counted inside the overall Web search type in Search Console. So a click total alone cannot tell you whether traffic came from the answer or the results below it. Google's own advice in the same guide is to track conversions and time on site in other tools, such as Google Analytics.
What should replace rank on the scorecard?
Three numbers per location, each tied to a decision someone will make. Rank stays, one level down.
| Metric | What it answers | Where it comes from | Read it |
|---|---|---|---|
| AI answer share | When buyers ask an AI assistant, is this location named? | Repeated sampling of a fixed prompt set per engine | Brand weekly, each location monthly |
| Qualified search actions | Did search send a call, form, booking request or direction request? | Call tracking, analytics conversions, Business Profile actions | Monthly by location |
| Booked revenue from search | Did those actions turn into paying business? | CRM or booking system, matched to source | Quarterly by location |
| Rank, impressions, citations | Why did one of the three move? | Search Console, Bing Webmaster Tools, rank tracker | Weekly, on the diagnostics page only |
AI answer share has to be measured as a share across many runs, never as a position. SparkToro's study with Gumshoe found less than a 1-in-100 chance that ChatGPT or Google's AI returns the same list of brands in two responses, and less than 1 in 1,000 for the same list in the same order. Rank inside an AI answer is noise. How often you appear is the signal. The sampling method for a large footprint is in our post on measuring AI share of voice across 50+ locations.
Citations sit in diagnostics because they explain answer share. Microsoft's AI Performance report in Bing Webmaster Tools, in public preview since February 2026, shows total citations, cited pages and the grounding queries behind them. Microsoft says the counts do not indicate ranking or placement within an answer. That is the right way to read them: which of your pages and which outside sources the engines lean on, not a new leaderboard.
A worked example: the quarterly review that goes wrong
For example, take an illustrative 60-location service brand going into its Q1 review. The SEO report shows average position on priority terms improved from 8 to 6 and calls the quarter a win. The operations team says inbound calls from search were flat and two regions were down. Both are telling the truth, and the meeting spends an hour arguing about which number counts.
Now run the same illustrative quarter on the scorecard above. Answer share shows the brand named in AI answers for most of its markets, but missing in the two weak regions, where a directory with stale hours is the most-cited source. Search actions confirm those two regions are where calls fell. Revenue per location shows the rest of the footprint holding. The meeting ends with an owner and a task: fix the listings in two regions and recheck answer share next month. The rank improvement is still true, and it explains nothing about what to do.
The point of the example is governance, not the specific figures. A scoreboard is only useful if a change in it tells a regional lead or a budget owner what to do next. Rank no longer does that on its own.
But rankings still drive the clicks we do get
This is the strongest objection, and it is partly right.
Organic results still carry real traffic, especially on short queries: in Pew's data, only 8% of one- or two-word searches produced an AI summary. Google's AI features guide says a page must be indexed and eligible to show with a snippet before it can be a supporting link in an AI Overview, so the SEO health that rank reflects is an input to answer share, not its opposite. Rank data is cheap, daily and well understood, while AI sampling is newer, costs more to run properly and has wider error bars. An SEO lead could fairly say: you are replacing a precise metric with a noisy one.
The answer is that precision about the wrong thing does not help a budget decision. Nobody is proposing to stop tracking rank. The proposal is to stop using it as the measure of whether the program works, because the evidence above shows it can improve while clicks and calls fall. Keep it where it earns its place: on the weekly diagnostics page, explaining why answer share or search actions moved. The noise problem with AI sampling is solved with sample size and a fixed prompt set, not by going back to a number that no longer tracks the outcome.
For 2027 planning, that is the default we would defend: outcomes on the scorecard, rank and citations underneath. If you want a sense of what monthly AI visibility work should produce against that scorecard, see what an AI visibility agency should deliver each month.
How to check where you stand
Pull your last two quarterly search reports. Circle every metric on the first page. If the headline is average position or keyword counts in the top 10, ask one question of each: if this number moved, would anyone change a budget, a listing or a page? Then pick five buyer questions for one location, ask ChatGPT, Perplexity and Google each of them three times, and count how often that location is named.
If that count surprises you, it is worth a structured look. For multi-location and enterprise programs, we can build the baseline across your footprint. Start with the free AI visibility audit.
Frequently asked questions
Should we stop tracking rankings altogether?
How do we measure how often AI answers name our locations?
Does Google report AI Overview traffic separately?
What do we tell leadership when clicks fall but revenue holds?
Sources
- Update: AI Overviews Reduce Clicks by 58% · Ahrefs
- Google users are less likely to click on links when an AI summary appears in the results · Pew Research Center
- AIO Impact on Google CTR: September 2025 Update · Seer Interactive
- Performance report (Search results) · Search Console Help
- AI Features and Your Website · Google Search Central
- Introducing AI Performance in Bing Webmaster Tools Public Preview · Microsoft Bing Webmaster Blog
- NEW Research: AIs are highly inconsistent when recommending brands or products; marketers should take care when tracking AI visibility · SparkToro