To analyse a website from the outside you read its domain overview: estimated organic traffic, ranking keywords, position distribution, backlinks and referring domains, and, since this year, its presence in AI answers. In our test on 10 October 2026 on two Italian sites whose real Search Console data we've already published, the organic traffic estimate was 7% out on one and 86% out on the other.
A domain overview opens in ten seconds, for any domain.
The trouble is what happens in the ten minutes after that.
People bring Semrush screenshots to meetings with a competitor's traffic circled in red, and the question is always the same: "Why do they get 4,000 visits and we get 900?"
The mistakes we see most often are these three:
- treating estimated traffic as a measured figure, when it's a multiplication of position, search volume and an assumed click-through rate;
- counting backlinks instead of the domains that link, so two footer links from one Spanish site count for more than twenty articles;
- looking at the keyword total rather than the positions, which is where the real opportunities are.
To find out how wrong an overview is, you need a site whose real numbers you know. We have two, and their data is already public.
Macropix makes LED walls in Milan: over the last twelve months its site collected 16,712 organic clicks and 1,145,717 impressions, with 79 of 135 category keywords on page one.
Sunscape develops agrivoltaic plants in Italy and looks for land to build them on: in August 2026 it closed at 195 organic clicks for the month, up from 46 in March.
So what does the domain overview say about these two sites?
On Macropix it gets the total almost exactly right and is up to 21 times out on a single keyword. On Sunscape it sees a seventh of the real traffic. In this guide you'll find:
- what's in a domain overview and how each number is calculated;
- the estimate-versus-reality test, with the keyword-by-keyword table;
- how to read backlinks without being fooled by the big number;
- the 6 steps to analyse a website, yours or a competitor's;
- the tools, starting with the ones that cost nothing.
Let's start.
What a domain overview is (and why it isn't an audit)
A domain overview is a one-screen summary of how a site performs in search: how much organic traffic it gets according to an estimate, how many keywords it ranks for and in which positions, how many sites link to it. In Semrush it's literally called Domain Overview, in Ahrefs it's the opening page of Site Explorer, and most other SEO suites have an equivalent.
It works on any site, including ones that aren't yours. That's why people use it: it's the first look at a competitor, at a site you're thinking of buying, at a supplier promising you results.
It isn't an SEO audit. An audit goes inside the site, crawls its pages and finds technical errors. The overview stays outside and looks at the site the way Google does: where it appears, for what, with how much authority. For your own site you need both; for a competitor's you only have the second.
Here's what it contains and where each number comes from:
| Metric | How it's calculated | What it tells you | The limit |
|---|---|---|---|
| Estimated organic traffic | For each keyword: search volume × assumed click-through rate for that position, then the sum | Order of magnitude and trend | It's a model, not a count |
| Ranking keywords | Searches in the tool's database where the domain appears in the top 100 results | How broad the topics covered are | Only sees keywords the tool has on file |
| Position distribution | The same keywords split into bands (1-3, 4-10, 11-20...) | Where the nearby opportunities are | A snapshot of one day |
| Backlinks and referring domains | Links found by the tool's crawler | Who talks about the site | Every tool has a different index |
| Authority score | Proprietary algorithm (Authority Score, Domain Rating, Domain Authority) | A quick comparison between sites | Google doesn't use it |
| AI visibility | Brand citations and mentions in the generative answers being monitored | Whether the site makes it into AI answers | Prompt sample chosen by the tool |
The first row is the one behind the most bad decisions. Ahrefs, in its documentation on organic traffic, writes that its estimates "don't, and can't, show you exactly" how much traffic a site gets, and that they're useful for comparison. DataForSEO, the data provider we used in the test, defines estimated traffic as the product of click-through rate and search volume for all the keywords the domain ranks for.
Put like that, it sounds like a technicality.
It isn't: it means that if the volume is wrong, or the click-through rate isn't the real one, the error gets multiplied. You can see it in the numbers below.
The test: estimated overview against Search Console (two Italian sites, 10 October 2026)
On 10 October 2026 we opened the domain overview of macropix.it and sunscape.it in DataForSEO Labs (Google Italy database, Italian language) and put it next to the Search Console data we've already published on the two project pages. Both sites work on the Italian market, so the test was run there. On one site the total matches almost to the click. On the other, 86% of the traffic is missing.
| Data | Macropix (estimate) | Macropix (real) | Sunscape (estimate) | Sunscape (real) |
|---|---|---|---|---|
| Organic clicks per month | 1,300 | 1,393 (average Sep 2025 - Aug 2026) | 27 | 195 (August 2026) |
| Estimate error | -7% | -86% | ||
| Keywords on page one | 33 | 79 of 135 tracked | 3 | 38 of 143 tracked |
| Total ranking keywords | 80 (129 counting synonyms) | 1,145,717 impressions in 12 months | 14, 12 of them flagged as new | 5,809 impressions in August |
Two caveats, so the comparison is fair. The Macropix average covers twelve months, while the overview captures positions recorded between August and September: they aren't the same period. And the "tracked keywords" are our own 135 and 143 category searches, chosen by us: the overview doesn't know them, it looks at its own database.
The result fits in two lines.
On a mature site with high-volume keywords, the estimate of the total is good. On the Macropix site the gap is 7%, better than we expected.
On a young site that lives on regional and long-tail searches, the estimate is almost blind. Sunscape gets traffic from searches such as "vendo terreno per fotovoltaico" (selling land for solar) or "agrivoltaico veneto" (agrivoltaics in the Veneto region), which have tiny or no volumes in the tools' databases.
That lines up with the most widely cited study on the subject: SparkToro compared four providers' estimates with Google Analytics across 641 sites and 7,692 monthly data points, and found a correlation of 0.790 for Semrush and 0.504 for Ahrefs (international sample, data from June 2020 to June 2021). Why the estimates break down, with data from two more Italian properties, is covered in our guide on how to check a website's traffic.
Why the total adds up and the individual keywords don't (the same volume counted three times)
The Macropix total is right because errors cancel each other out, not because each keyword is estimated well. We took the five keywords with the most estimated traffic and compared them with the real clicks in Search Console: the estimate for these five is 8.7 times the real traffic.
| Keyword | Position in the overview | Real average position | Estimated visits per month | Real clicks per month | Ratio |
|---|---|---|---|---|---|
| led wall | 2 | 9.1 | 1,312 | 62 | 21.1 times |
| ledwall | 4 | 5.5 | 534 | 125 | 4.3 times |
| ledwall prezzo (LED wall price) | 1 | 7.2 | 219 | 19 | 11.4 times |
| led wall prezzo (LED wall price) | 3 | 4.2 | 47 | 23 | 2.1 times |
| ledwall da esterno (outdoor LED wall) | 4 | 4.1 | 26 | 17 | 1.6 times |
There are two causes, and they apply to any tool.
The first is a position captured on a single day. The overview sees "led wall" in second place; the twelve-month average in Search Console is 9.1. A keyword that swings between page one and page two gets estimated as if it were always in the top three.
The second is volume. "led wall", "ledwall" and "video wall led" all have the same volume on Google Italy: 8,100 searches a month. That's no coincidence: Google Ads groups close variants and returns a single number for the group.
So how much is that group really worth?
Search Console gives "ledwall" 67,560 impressions in a year and "led wall" 27,868: together that's about 7,950 a month, almost exactly the group's 8,100. The real volume splits 71 to 29 between the two spellings. The tool's keyword list, on the other hand, counts it once for each spelling.
The DataForSEO overview merges synonyms (80 keywords instead of 129), which is why its total ends up close to reality. If you go through the list keyword by keyword, that correction disappears.
This has a practical consequence: the overview is fine for saying "this site is ten times bigger than that one", not for deciding which keyword to go after. For that decision you need real positions over time, which is the job of rank tracking, and the share of searches a site really holds, its share of voice.
Choosing keywords by looking at a competitor's estimated traffic?
You've just seen an estimate 21 times higher than the real traffic on a keyword with 8,100 searches. If your content plan or your Google Ads budget is built on those numbers, you risk investing where the clicks aren't. We put your competitors' overviews next to the real data in your Search Console and tell you how much to correct, keyword by keyword.
Book a strategy consultationOn young sites the overview lags behind (27 estimated visits, 195 real)
An overview describes a site that has been around for years well, and a site that's growing right now badly. On Sunscape it sees 14 keywords, and 12 of the 14 are flagged as new: the tool has only just noticed the site exists.
Meanwhile, in reality, monthly impressions went from 329 in March to 5,809 in August, and page-one keywords from 1 to 38. The month-by-month story is in our guide on how long SEO takes to show results.
If that site were your competitor, the overview would tell you: "27 visits a month, nothing to worry about."
Wrong.
The right signal in these cases isn't traffic but the column of new and rising keywords. On Sunscape, 12 new and 1 rising out of 14; on Macropix, 38 new and 22 rising out of 80. A competitor with lots of new keywords is a competitor that's publishing or restructuring, and you see it months before the traffic says so.
Backlinks: the big number is almost always a footer (10,234 links, 208 domains)
In a domain overview the backlink count is almost always the biggest number on the page, and it's the one that tells you least. What matters is how many different domains link to the site, and from which parts of their pages.
Take Macropix. The overview shows 10,234 backlinks, but they come from 208 domains. Then you open the detail and find this:
- 5,359 links, 52%, come from .es domains: a single Spanish site repeating the link across thousands of pages outweighs everything else;
- 3,123 links, 30%, sit in a footer or header, meaning they're repeated on every page of the site hosting them;
- 106 links, 1%, sit inside the body of an article, which is the kind of link that usually carries the most value;
- links from .it domains number 88.
Yet in July 2026 Macropix was first in Italy for the "ledwall" keyword group, with a share of voice of 25.55%, ahead of Amazon (Semrush data). The link count doesn't explain the ranking: what explains it is the pages built for each product and the work on category keywords, described in our guide to SEO for manufacturers.
On Sunscape the overview shows something different and more useful: 53 of the 66 backlinks point to pages that no longer respond. Those are links already earned, and a redirect gets them back before you even start looking for new ones.
The method for cleaning up the count, with a test on 11 Italian sites, is in our guide to checking backlinks. And before comparing two sites on their authority score, read what Domain Authority really measures: every tool has its own version, and Google uses none of them.
How to analyse a website in 6 steps (start with the overview, finish with real data)
To analyse a website, start from the domain overview, go down into positions and links, and finish with measured data when the site is yours. It's the order we follow when we take on a project, and it takes about an hour per domain.
1. Set the country and language before reading any number
Many tools open on the United States database. A UK site read on the US database looks half empty, and a comparison made across different databases is worthless. Google UK, English, mobile: always the same settings for every site you compare.
2. Read traffic as an order of magnitude
1,300 or 1,400 is the same number. 1,300 against 13,000 is a different class of site. Look at the twelve-month trend too: a sudden drop usually coincides with an algorithm update or a site rebuild.
3. Separate branded traffic from generic traffic
A competitor with 5,000 estimated visits, 4,000 of them on its own name, isn't taking customers from you on Google: it's well known. Semrush shows the "Branded vs Non-Branded" split right in the domain overview; with other tools, just filter the keywords that contain the brand name.
4. Look at positions 11-20
It's the most overlooked row in the overview. These are the keywords where the site is already relevant but gets no clicks. On Macropix there are 13: that's where the work starts, because moving a page from position 12 to 6 costs much less than creating one from scratch.
5. Count referring domains, not links
Open the detail and look at three things: which countries the links come from, where they sit on the page (body text or footer), and how many point to broken pages. In ten minutes you can tell whether the profile is made of genuine citations or repeated links.
6. If the site is yours, finish with measured data
On your own site you don't need the estimate: you have Google Search Console for real clicks, impressions and positions, and PageSpeed Insights for speed. Use your own site's overview for something else: measuring how far off the tool is in your sector. More on that at the end.
If the point of the analysis is to find out who's taking your customers on Google, the next step is a competitor analysis that starts from the search results, because the competitors Google sees often aren't the ones you know.
Domain overview tools (starting with the ones that cost nothing)
To analyse a website from the outside you need one tool with a database of UK keywords and one with a backlink index; for your own site, Google's own tools are enough. The choice depends on how many domains you look at and how often.
| Tool | Which sites | What the overview gives you | When to use it |
|---|---|---|---|
| Google Search Console | Only yours, verified | Real clicks, impressions, positions and queries | Always, for your own site |
| Semrush (no-cost account) | Any site | Full overview with traffic, keywords, backlinks, branded split and AI visibility; 10 Domain Overview queries a day | Occasional checks |
| Ahrefs Free (formerly Ahrefs Webmaster Tools) | Only yours, verified | Backlinks and keywords from the Ahrefs index, up to 1,000 at a time | To compare your site with an external index |
| Bing Webmaster Tools | Only yours, verified | Backlinks, keyword research and search performance on Bing | For a second link index and Bing data at no cost |
| Semrush, Ahrefs, SE Ranking (paid) | Any site | Unlimited overviews, multi-domain comparison, history | If you analyse competitors every month |
| Similarweb | Any site | Traffic from every channel, not just organic | If you need the picture beyond Google |
Before paying for a subscription, check that the tool's UK database covers your sector: look up your own site and count how many of your real keywords appear. If it finds fewer than half, you're in the Sunscape situation. We've compared prices and plans in our piece on low-cost SEO tools.
In 2026 the overview has one more row: AI answers
Semrush's domain overview now opens with an AI Visibility block: how often the brand is mentioned in generative answers and which pages get cited. It's there because some searches end without a click.
The most solid figure is American. Pew Research Center tracked the browsing of 900 US adults in March 2025: with an AI summary on the page, users clicked a traditional result on 8% of visits, against 15% without one. The summary appeared on 18% of searches. We haven't found a click study of the same kind for the UK; the figures we have, starting with the share of searches that show an AI Overview, are in our guide to zero-click searches.
For website analysis this means one specific thing: a site can lose clicks without losing positions. If the overview shows stable positions and falling traffic, check whether an AI Overview has appeared on those searches before you go looking for a technical problem. How to measure presence in AI answers is explained in our guide to AI visibility.
Frequently asked questions about analysing a website
Yes. For your own site, Google Search Console and PageSpeed Insights give you real data on clicks, positions and speed at no cost. For a site that isn't yours, a Semrush account without a subscription opens the full domain overview, with a limit of 10 Domain Overview queries a day.
It depends on the type of site and the level of detail. In our test on 10 October 2026 the estimate of the total was 7% out on a mature site and 86% out on a young site that lives on long-tail searches. On a single keyword the gap reached 21 times the real traffic.
No. It shows an estimate: for each keyword it multiplies search volume by an assumed click-through rate for that position and adds up the results. Only someone with access to the site's Search Console or Analytics sees the real traffic.
No. What counts is the number of different domains linking and where on the page the link sits. In the Macropix case, 10,234 backlinks came from 208 domains, 52% of them from Spanish domains and only 1% from the body of an article.
It depends on how fast the sector moves. Once a month is enough for most B2B markets; if a competitor shows lots of new or rising keywords, look at it every two weeks, because it's publishing or restructuring its site.
No. No external tool has access to another site's Search Console or Analytics data. You can get a reliable order of magnitude and a trend, and correct them with the ratio between estimate and reality measured on your own site.
The short version (and the number you actually need)
A domain overview is a model, not a count.
On the total for a mature site it's only slightly off: -7% on Macropix.
On a single keyword it's a long way off: up to 21 times.
On young sites it lags behind: -86% on Sunscape.
With backlinks, the domains and the position on the page matter, not the number.
Something the industry rarely says: the most useful overview isn't your competitor's, it's your own. Open it next to Search Console and work out the ratio between the estimate and real clicks: that's the tool's correction factor for your sector and your type of keyword. If on your site the estimate is half the real figure, your competitor's is probably half too. Without that number, every comparison with another domain is a comparison between two errors whose size you don't know.
If you'd like us to run this comparison for you, on your data and your competitors', it's part of our SEO service. And if you want to see how the analysis looks once it's turned into a document for the board, there's an example with real numbers in our SEO report guide.
See you next time!
