A keyword ranking is the point at which a page appears on the search results page for a specific query, counted from the top down. Position 1 means the first organic result, position 10 means the last one on the first page.
That number, though, does not exist as a single copy. Google shows different results depending on who is searching, from where, on which device and at what moment: the position you see and the one a customer two hundred miles away sees may not match. So the useful question is not "what position am I in", but "which tool measured what, and when".
What takes up a position and what does not
Google explains in the Search Console documentation how it assigns the numbers. Position is counted from top to bottom on the primary side of the page, then on the secondary side. Each results element takes a single position, even when it contains several links: a carousel with eight results inside does not take eight positions, it takes one, and every link in it inherits that number.
Three details from the same official Google guide change the way you read any report:
- Ads do not occupy a search position. If there are four ads above your result, Search Console still has you first even if on screen you are the fifth block from the top.
- A featured snippet contains a link to its source, so it usually takes position 1.
- Elements made only of query refinement links, the ones that start a new Google search instead of taking you off Google, do not take a position and do not move the numbers below them.
That is why position 1 can be a result the user reaches after two screens of scrolling. The number describes the order of the elements, not how far down the page yours sits.
Average position and tracked position are two different metrics
Search Console's average position is the topmost position held by a link to your property in the search results, averaged across all the queries in which your property appeared. Google gives an example: if for one query the property appears at positions 2, 4 and 6, the value recorded is 2; if for another it appears at 3, 5 and 9, the value is 3; the average of the two queries is 2.5.
Two things follow that are often misread in reports. First: no impression, no position. A result on page three that nobody ever viewed does not enter the calculation, and the average stays better than it would be if everything were counted. Second: the average moves even when rankings are flat, because all it takes is a change in the mix of queries, countries or devices the impressions come from.
A rank tracker does something else. It queries Google at regular intervals for a list of keywords you have chosen, from a location and a device you have set, and records where the domain appears in that snapshot.
| Aspect | Search Console | Rank tracker (Semrush, Ahrefs, SE Ranking and similar) |
|---|---|---|
| What it measures | Where you appeared to the real users who saw you | Where you appear in a simulated search with fixed settings |
| Data source | Impressions actually served by Google | An automated query repeated by the tool |
| Keywords covered | All those for which you had at least one impression | Only those you added to the project |
| Personalisation effect | Included: it is the average of real sessions that differ from each other | Neutralised by design, so comparable over time |
| Update frequency | Daily, with the most recent hours still preliminary | Depends on the plan, usually daily or weekly |
| What it is for | Understanding what happened to traffic | Seeing whether a single page is rising or falling for a query |
Why two tools give different numbers
It is not a flaw in the tools. Google puts it in writing in the Performance report guide: "Even if a query appears in your list, you might not see your site in results if you run the same query in Google Search. This is because Search results are specific to the time, place, device, and recent history of the person searching."
On its How Search Works page Google uses the word "football" as an example: someone searching from Chicago sees results about American football and the Chicago Bears, someone searching from London sees soccer and the Premier League. Same word, two SERPs. Within the UK the same mechanism shows up on queries with local intent, where the result changes from town to town without anyone touching the site.
Checking your own position by hand in an incognito window is therefore not a check. It is a sample of one measurement, taken from one town, on one device, at one moment.
What a position is worth today
A position matters for the clicks it brings, and that relationship has moved a lot in recent quarters. Advanced Web Ranking publishes the trend in organic CTR every three months, on an international sample of thousands of sites and millions of keywords covering the United States, the United Kingdom and other markets. In the Q1 2026 study the two device types moved in opposite directions: on desktop searches the top five positions gained 10.54 percentage points of CTR compared with the previous quarter, while on mobile position 1 lost 2.20.
On the same sample, the Q2 2025 study had recorded the share of results showing AI Overviews rising from 34.32% to 46.22% in three months. Position 3 under an AI Overview and position 3 on a SERP of plain blue links are not worth the same traffic, and the number alone does not tell you that. This is why rankings should be read alongside impressions and clicks for the same query, not in place of them.
How to read a drop without misdiagnosing it
Before rewriting a page because it "has dropped", rule out the cheapest explanations first, in this order:
- The time window. Google itself suggests weekly or monthly granularity in the Performance report to smooth out daily fluctuations such as weekends and holidays: a day-by-day comparison sets a Monday against a Saturday and produces swings that are not changes.
- The query mix. A page ranking for a hundred queries can have a worse average simply because it has started getting impressions for new, harder queries.
- Country and device. Filter them separately, because a loss on mobile and stability on desktop cancel each other out in the aggregate figure.
- The SERP. A block that has appeared above you pushes your result down without you having lost a position in the strict sense.
Only when these four checks fail to explain the movement does it make sense to move on to the page and the ranking factors. In the projects we handle as SEO consultants, half of the reported "drops" are closed at the first or second item on this list.
The detail the industry rarely mentions
Google's spam policies for Search have a paragraph on machine-generated traffic, and it says something almost no article on rank tracking quotes: automated traffic "includes scraping results for rank-checking purposes or other types of automated access to Google Search conducted without express permission", and "such activities violate our spam policies and the Google Terms of Service".
Nobody has ever seen a site penalised for buying a rank tracker, and the market for those tools carries on undisturbed. It is still worth knowing which side Google is on: the only source of ranking data it recognises as legitimate is the one it provides itself, in Search Console and through its API. Everything else is an estimate built on queries that, by its rules, should not be sent.
Keyword rankings FAQs
It depends on what the SERP for that query looks like. Position 3 under an AI Overview and four ads brings far fewer clicks than position 5 on a page of organic links only. Rather than setting a threshold, look at clicks and impressions for the individual query in Search Console: if impressions grow and clicks do not, the problem is the snippet or what sits above you, not the position.
For editorial decisions a weekly or monthly check is enough. Google itself suggests weekly or monthly granularity in the Performance report, because a day-by-day comparison sets a Monday against a Saturday and creates swings that are not real changes. Daily monitoring makes sense when you are following an algorithm update or a migration.
Because the Search Console value is an average across all recorded searches, while yours is a single measurement. Google states that results are specific to the time, place, device and recent history of the person searching, so your own search is not a check on the data: it is one of the many cases that make up that average.
No. For each query Google takes the topmost position held by a link to your site and uses that, then averages it across the queries in which the site appeared. If for one query you are at positions 2, 4 and 6, the value recorded is 2. Positions that received no impressions are not included.
