You search Google for your business and see it in position eight. Your search performance report shows an average position of five, while someone in another city checks and finds it in position three. Which number is correct?
All three may be valid. Each one describes what Google displayed under a different set of conditions, or how Google Search Console, Google’s search performance reporting tool, combined many displays into an average. None of them is a permanent position assigned to the page.
A ranking is an observation. It can tell us where a result appeared for a particular search, person, place, device, and time. A report can instead summarize a collection of those appearances. Either form is useful evidence, but it needs context before it can tell us whether search performance is improving or declining.
Why two people can see different rankings
Google explains that search results can differ because of time, location, language, device, and recent search activity. Personalization can affect some results as well, although Google says it does not affect every result.
These factors answer a question that a bare ranking number leaves out: ranking for whom, where, on which device, and when?
Location is an obvious example for a search such as “plumber near me,” but context is not limited to local searches. Google may show different results as its systems change, Google discovers new pages, existing pages change, and different data centres receive updates. Mobile and desktop searches can also produce different result pages.
An incognito search removes some history and personalization, but it does not remove location, device, language, or time. Incognito mode can provide another observation; it cannot reveal one neutral ranking that applies to everyone.
Google makes the same point in its Search Console documentation: a site owner may see a query in a performance report and still be unable to reproduce the result with a manual search. The report and the manual check are measuring different things.
What Search Console’s average position means
An impression is counted when a search result containing a link to the site is shown. For each impression, Search Console records the position of the site’s highest result, then averages those recorded positions. This is why the metric is called average position.
Following Google’s documented average-position calculation, suppose a page appears for the same query in positions two, five, and eight across three impressions. Its average position is five:
`(2 + 5 + 8) / 3 = 5`
That does not mean the page appeared in position five every time. Five is the average of three different observations.
The calculation also changes depending on how the data is grouped. When Search Console groups data by the entire website, it counts the highest position held by any page from that site. When the report is grouped by individual page, each page has its own position data. Google’s explanation of position and impressions gives the full calculation, and its data aggregation guide explains why the website and page views can produce different numbers.
Neither view is wrong. The website view answers, “What was the highest position held by this site?” The page view answers, “What position did this particular URL hold?” A manual search answers another question: “What appeared for this search under these conditions at this moment?”
Once those questions are separated, the apparent contradiction disappears.
What the 2026 local and mobile research found
A 2026 Yext study provides a useful example, but its scope needs to stay attached to its findings. This was not a study of every query, market, business, device, or Google result.
Yext analysed 8,162,307 snapshots of unpaid, or organic, search results from January 1 through August 9, 2026. The data came from 1,272 Places Scout reports and 9,371 report runs. Those runs used location-specific search grids, which check results from multiple nearby points on a map. The researchers compared the ordered top 10 organic results at adjacent points during the same report run. In other words, they measured how much nearby locations disagreed at the same time, not ordinary day-to-day ranking movement.
Within a matched group of 718 reports, 702 showed an increase in mobile organic geographic variation. Geographic variation rose by an average of 48.7%. It rose by 76.3% for queries that named a location and by 30.2% for “near me” queries. The same measure rose by 2.4% in Google’s mobile classic three-pack, the map section with three business listings, and by 2.8% in Google Maps.
Those numbers describe Yext’s location-specific collection of reports and method. They do not establish that every business or query experienced the same change. The study also did not provide enough information to treat that collection as a representative sample of North America.
Joy Hawkins called the observed change the “Dynamic SERP Update”. SERP is short for search engine results page. That phrase is her industry label, not the name of a Google-confirmed update. Google confirmed that a May 2026 core update ran from May 21 to June 2. Yext reported a timing overlap, but it also stated that its data could not establish the cause. Movement in its dataset began before May 21.
The study supports a narrower conclusion. In this local and primarily mobile collection of reports, one observed ranking became less representative of what the wider nearby market could see. The observation still happened; it simply could not carry the weight of a market-wide conclusion by itself.
For a closer look at the local factors that can change what searchers see, the site’s article on what drives local search rankings covers that separate branch of the subject.
What deserves more weight than one manual check
A ranking remains useful when it is treated as one piece of evidence. The mistake is asking one number to answer every performance question.
Read the evidence in this order:
- First, look for a sustained position trend rather than reacting to one search or a small movement.
- Then check impressions and clicks. Impressions show how often the site appeared, while clicks show how often someone selected it.
- Next, break the pattern down by query, page, country, and device. A broad decline calls for a different investigation from a change limited to one page or one type of search.
- Finally, check the business outcomes the site is meant to produce, such as calls, form submissions, bookings, purchases, qualified leads, or revenue.
Google recommends paying more attention to trends in clicks and impressions than to absolute position alone. That is sensible because position measures visibility, while clicks show whether that visibility produced a visit.
The final step moves beyond Search Console. A conversion is a business action completed after someone reaches the site, and the actions that count depend on the business. Website event tracking can connect activities such as form submissions and button clicks to the traffic that produced them.
These measures build on one another:
`ranking trend -> impressions -> clicks -> business outcomes`
Each stage answers a different question. A ranking movement tells us that visibility may have changed. Impressions and clicks show whether search exposure and visits changed. Calls, forms, bookings, purchases, qualified leads, or revenue show whether the change mattered to the business.
When a ranking change deserves investigation
One manual search is a prompt to look at the data, not a verdict on performance. A small position change can also occur without requiring a major response. Google says positions are not static because the open web and its search systems keep changing.
A ranking change deserves a closer look when the evidence forms a pattern. Ask four questions:
- Is the change sustained across a meaningful comparison period?
- Is it broad, or is it limited to particular queries, pages, countries, or devices?
- Did impressions and clicks move with it?
- Did calls, forms, bookings, purchases, qualified leads, or revenue move?
There is no single reporting window that fits every business, so “sustained” has to be judged against normal demand, reporting frequency, and the amount of data available. Compare like with like, and avoid rewriting a page because of one isolated check.
If position declines across many important queries and pages, and clicks or impressions decline as well, the pattern warrants investigation. A sustained change limited to one important page or query can also deserve attention. If one manual result changes while search traffic and business outcomes remain steady, there may be little evidence of a business problem.
The practical rule
Your Google ranking is not one fixed number. It is a series of observations shaped by context, then summarized in different ways by different tools.
Keep the ranking data, but read it in sequence. First, look for a sustained pattern and determine whether it is broad or limited. Then check impressions and clicks. Finally, check the actions that matter to the business. That sequence turns a ranking number into evidence you can use.
