You can check your website's ranking in three ways: search for the keyword on Google yourself, read the average position in Google Search Console, or use a rank tracker, software that queries Google every day for a list of keywords from a city and device you choose. None of the three gives you "the" position: in our test on 10 October 2026, the search "led wall" made in London on a smartphone and on a computer had only 1 site in common among the first 9 organic results.
Two snapshots of the same search, in the same city, a few minutes apart.
Yet when a business asks us "where do we rank?", it usually already has an answer in mind. It got it like this:
- an incognito search from the office computer;
- the Search Console average position read as if it were today's ranking;
- an online checker that only looks at page one, and only on desktop.
Those are three different numbers. And they often end up in the same meeting.
So which one is right?
All three, each for its own question. The trouble starts when you mix them up.
We see it on the projects we run. For Macropix, an LED wall manufacturer based near Milan, we track 135 Italian keywords split into four clusters every month: in September 2026, 79 were on page one and 28 in the top three.
For Sunscape, which develops solar farms and looks for land all over Italy, tracking of 143 keywords started on 9 July 2026: at first only one was on page one, by September 38 were on page one and 112 in the top hundred.
Those numbers only hold up if you know where they were measured from.
Here you'll find:
- the three methods for checking rankings and what each one measures;
- our test: two keywords searched from London and Manchester, on a smartphone and on a computer;
- five online rank checkers you can use without paying, and how much of Google's results page they really read;
- what to set up in a rank tracker before you trust a number.
Let's start.
Three ways to check your ranking (and what each one measures)
Each method answers a different question: a manual search tells you what you see right now, Search Console tells you where you appeared to real users on average, a rank tracker tells you where you appear every day under the same conditions. The number changes because the question changes, not because one of the three is wrong.
| Method | What it measures | Cost | The limit that matters |
|---|---|---|---|
| Manual search (incognito too) | The page you see, from that device and that place, at that moment | Nothing | A single reading: it doesn't tell you what a customer sees in another city or on another device |
| Google Search Console | The average position of your site's topmost result, across the real searches where you appeared | Nothing | Only your own site, and an average: it moves even when rankings don't |
| Rank tracker | Your position on a list of keywords you choose, with a fixed city, device and frequency | From a few pounds to over £100 a month | It only sees the keywords you enter, and counts positions by its own rule |
How a position is counted (ads don't take one, a carousel takes just one) is explained in our guide to keyword rankings. Here we care about something more practical: how much the number changes depending on where and how you look at it.
We measured it.
The test: the same keyword from London, Manchester, smartphone and computer (1 site in 9 in common)
On 10 October 2026 we read page one of Google UK for two keywords from the sector of our Macropix case, "led wall" (1,000 searches a month in the UK) and "advertising screens" (480), in three set-ups: London on a smartphone, London on a computer, Manchester on a smartphone. The scans ran a few minutes apart through DataForSEO's SERP API, with the location set to the city and the language to English.
| Keyword | Where and how | First organic result | Domains in common with London smartphone | Local pack | Who's in the local pack |
|---|---|---|---|---|---|
| led wall | London, smartphone | mediascape.ltd.uk | reference (7 domains) | yes, 2 listings | Led Video Wall Hire, Dynamo LED Displays (London) |
| led wall | London, computer | lg.com | 1 of 9 | yes, 2 listings | the same two London businesses |
| led wall | Manchester, smartphone | lg.com | 1 of 9 | yes, 2 listings | LED Hub UK, Pixite (Manchester) |
| advertising screens | London, smartphone | ukpos.com | reference (8 domains) | yes, 3 listings | Screen Moove, Dynamo LED Displays, Lightmedia Displays (London) |
| advertising screens | London, computer | ukpos.com (a different page) | 3 of 8 | yes, 3 listings | the same three London businesses |
| advertising screens | Manchester, smartphone | ukpos.com (the same page as London computer) | 3 of 9 | yes, 3 listings | DP Sign Print, Spectrum Advertising, The Manchester Screen |
Three things stand out.
Device and city both change the page, and not in a way you can predict. On "led wall", London on a smartphone and London on a computer shared one site out of nine: Amazon. Manchester on a smartphone, on the other hand, showed exactly the same nine organic sites as London on a computer. On "advertising screens" it was the same story: London on two devices had 3 sites in common out of 8, while London on a computer and Manchester on a smartphone had 8 out of 9. In our sample, London on a smartphone was the odd one out.
The city changes who's in the local pack. Google showed Maps listings in all six scans, but with different businesses. In London, on both devices, "led wall" brought up two London companies; in Manchester, two Manchester companies took their place. On "advertising screens" the three London listings became three Manchester ones. The organic results barely moved between the two cities; the map did.
The same site can appear twice on the same page. It happened in Manchester on "advertising screens", where Screen Moove had two organic results; in London, on a computer, it had one organic result and one Maps listing. A tool that counts every row and one that counts only the domain's best result will give you two different positions for the site below it.
So which number goes into the report?
None of them on its own. Take Macropix in Italy: in Search Console, over the last twelve months, the Italian keyword "ledwall" brought the site 67,560 impressions and 1,495 clicks, with an average position of 5.5 (figures published in the case study). No single manual search would show a 5.5, and there's no contradiction: the average adds up a year of real searches, from every city and every device, and only counts the ones where the site appeared.
The sample is small, two keywords and six scans, and it doesn't measure how much Google varies in general. Two extra scans of "led wall", from London on a computer and Manchester on a smartphone, asking for 10 results instead of 20, even put a third site, e-screen.co.uk, in first place. What the test does measure is more useful: before you look at a position, you need to know where it was taken from.
Your report says page one, but customers can't find you?
In our test the same search on a smartphone and on a computer had 1 site in 9 in common, and the local pack changed from London to Manchester. Let's check together which cities and devices your customers search from and where you really show up, keyword by keyword.
Book a strategy consultationAn incognito search isn't a check (but it's good for one thing)
A manual search in incognito mode removes your account history, but not your location or your device. Google says so itself in its help article on the Performance report: search results are specific to the time, place, device and recent history of the person searching.
Applied to the test: if your office is in London and you check "led wall" from your computer, in our scan you got LG's video wall page as the first organic result and London companies in the Maps listings. Your customer searching on a smartphone in London sees a page that shares one organic site with yours. The one in Manchester sees two Manchester companies on the map.
So is a manual search useless?
No. It's the quickest way to see what the page looks like: how many ads sit above, whether there's an AI Overview, a local pack, a product carousel. No position number tells you that. How to read a results page, block by block, is in our guide to the SERP.
Use it to look at the layout. Not to measure.
Search Console: the check that measures instead of estimating, at no cost
Google Search Console is Google's own tool that shows the real searches your site appeared for, with impressions, clicks and average position, and it costs nothing. It's the only source that doesn't simulate a search: it counts the ones that actually happened.
To check a page's ranking:
- open Performance, then Search results;
- switch on Average position and pick a range of at least 28 days;
- filter by query or by page, then open the Countries and Devices tabs.
The last step is the one almost everyone skips. The test has just shown you that a smartphone and a computer can see two different pages: in Search Console you separate them with one click, and a "flat" average can hide a page that's rising on one and falling on the other.
And the limits?
Two, and they're specific. You only see your own site, so you don't know where your competitors are. And you only see the searches where you appeared: a keyword where you're on page six, with no impressions, simply isn't there. What the tool shows report by report, and how much data it leaves out, is in our guide to Google Search Console.
Online rank checkers: how much of Google's page they really read (from 10 to 100 results)
Online rank checkers run a search for you and tell you where your domain appears, and the ones below cost nothing for a single check. What sets them apart comes down to three details: how many results they read, whether they check mobile, whether they let you choose a location. We read their official pages on 10 October 2026, picking tools that rank on Google UK for "check google ranking" (1,000 searches a month) or that we already compared in the Italian version of this article.
| Tool | Results checked | Mobile | Stated limits |
|---|---|---|---|
| Portside Google Rank Checker UK | top 100 | yes | Google UK only, optional UK town or postcode, rate-limited |
| Seobility Ranking Checker | top 20 | yes | choice of country and city; 5 checks a day with an account |
| Sitechecker | top 100 | yes | 155 countries, current and historical data |
| SEO Review Tools | first page only (10) | yes on google.co.uk | up to 10 keywords at a time, no local option |
| Ahrefs Keyword Rank Checker | first page (top 10) | not stated | 187 countries; tracking over time is in the paid Rank Tracker |
The detail that matters is depth. A new site, or a page on the way up, is usually beyond page one: a tool that reads only 10 results will tell you "not ranking" even when you're eleventh. JaySearch's guide to free rank checkers (updated June 2026) puts it plainly: new websites need a scanning depth of top 50 or top 100 to track their growth, because that's where their pages sit at the start.
For a one-off check they're fine. For following a keyword over time they aren't: without history you don't see a curve, you see a single frame. Other tools that cost nothing, beyond rank checking, are in our guide to no-cost SEO tools.
Rank trackers: five settings to decide before you look at a number
A rank tracker is software that repeats the same search every day, or every week, for a list of keywords and records where your domain appears. The number it gives you is only as good as the settings you configured it with. These are the five we decide before opening a project.
1. Your customers' cities, not your office's
On keywords with a local pack, the city changes everything, as "led wall" showed between London and Manchester. For Macropix, which has passed five thousand installations worldwide, a single location in Milan would have painted a rosier picture than the real one. When Google shows Maps listings it's a different game, and we cover it in our guide to the local pack.
2. Smartphone and computer tracked separately
In the test, the sites in common between the two devices in London were 1 in 9 on "led wall" and 3 in 8 on "advertising screens". If the tool charges for the second device, it's worth paying for it on the keywords that bring in enquiries.
3. Depth of at least 100
To see a page rising, you have to read it before it reaches page one. Sunscape started with 46 keywords in the top hundred and just one on page one: with a tracker that stopped at the first 10 results, the first two months of work would have been invisible.
4. The right frequency for the decision
Entry plans don't all update at the same pace. According to the price lists we read on 10 October 2026 for our comparison of low-cost SEO tools, Ahrefs Lite ($129 a month, 750 keywords) updates once a week, SE Ranking Core ($129, 2,000 keywords) every day, and SERPRobot checks 75 keywords every 24 hours for $4.99 (prices in US dollars as published, excluding VAT). For a monthly report, weekly is enough; for following a site migration, it isn't.
5. An overall metric, as well as position
Nobody reads 135 rows for 135 keywords. In its guide to rank tracking (June 2026), Similarweb suggests two summary metrics: share of voice, the share of potential clicks the domain takes across the tracked keywords, and weighted average position, weighted by search volume. For Macropix we use the distribution by bands (top 3, page one, first two pages) and share of voice, which we explain in our guide to share of voice.
And which software?
For our clients' daily tracking we use Keyword.com. The full comparison of the suites, with checked price lists, is in our guide to the best SEO tools. If you also want to follow citations in Google's AI answers, trackers work by different rules and we've covered them in our piece on rank trackers for AI Overviews.
One note tracker vendors don't put on their homepage: Google's spam policies list, under machine-generated traffic, "scraping results for rank-checking purposes" without express permission. We don't know of any site penalised for buying a rank tracker, but the only ranking data Google recognises as its own is still Search Console's.
How often to check your ranking (and what to do if it drops)
For content decisions a weekly or monthly check is enough; a daily check is for site migrations and algorithm updates. On Macropix, tracking and reporting are monthly, split by cluster: the daily figure helps you understand an anomaly, not make decisions every morning.
When a keyword drops, before you touch the page, check three things in this order:
- whether the drop is on one device only or in one location only;
- whether a new block has appeared above you on Google's page (a local pack, a carousel, an AI Overview);
- whether Search Console confirms the drop in clicks, not just in position.
If the drop is still there after these three checks, the diagnosis path is in our guide on why your website isn't ranking. And to present rankings to the people who set the budget, without 135 rows, there's our guide to the SEO report.
Frequently asked questions
Yes. Google Search Console shows impressions, clicks and average position for every query your site appeared for, at no cost. For a one-off check there are online tools such as Portside's UK checker or Sitechecker, which read the top 100 results, but without history: to follow a keyword over time you need a rank tracker.
No. Incognito removes your history, but not your location or your device. In our test on 10 October 2026 the same search for "led wall" made in London on a smartphone and on a computer had only 1 site in common out of 9 on page one.
No. It's the average of your site's topmost result across all the real searches in the period, from every city and device, and it only counts the searches where your site appeared. That's why it may not match any search you make by hand.
Yes. On "led wall" in London the first organic result was mediascape.ltd.uk on a smartphone and LG on a computer, and on "advertising screens" the two devices shared only 3 sites out of 8. In Search Console you separate the two devices in the Devices tab.
It depends: for content decisions a weekly or monthly check is enough, while during a site migration or an algorithm update you need daily data. Watch the tracker's plan: Ahrefs Lite, for example, updates positions once a week.
Yes, but not with Search Console, which only shows sites you've verified as yours. You need a rank tracker, which checks any domain on the keywords you enter, or one of the online rank checkers, which also accept a competitor's domain.
A ranking is a map (not a number)
In short:
three methods, three questions: what I see, what users saw, where I am every day;
in London, a smartphone and a computer had 1 site in 9 in common on "led wall";
the city decides who gets into the local pack;
Search Console is the only source that counts real searches.
There's an idea you don't hear much in the industry. A ranking isn't a number, it's a map: on "led wall" the London Maps listings belong to two London firms, in Manchester to two Manchester ones, and the organic page you see depends on the device in your hand. A useful check doesn't ask "where do I rank?", it asks "where are my customers, and what do they see from there?". If the locations in your tracker match your office address rather than your customers', the report is describing the wrong market.
If you'd like us to set up the tracking, on the cities and devices your customers really search from, it's the work we do in our SEO consultancy.
See you next time!
