SEO numbers come from three different places. Google Search Console reports what Google showed and what people clicked. Your analytics tool reports what visitors did after they arrived. Third-party tools estimate the rest from the outside, without any access to Google’s data. Each source counts a different moment, so their numbers won’t match, and that’s normal rather than a sign that something is broken.
In plain English
Picture your website as a shop on a busy street, with three people counting.
One stands outside and counts everyone who walked past and glanced at your window. One stands at the door and counts who actually came in. One stands at the till and counts who bought something.
All three are counting honestly. All three will hand you a different number at the end of the day, because a glance, a visit, and a purchase aren’t the same event. Nobody is wrong.
That’s SEO measurement. The window count, the door count, and the till count come from different tools, and the whole skill of reading an SEO report is knowing which one you’re looking at.
Why this matters for your site
Most disagreements about SEO reporting aren’t disagreements about the work. They’re two people looking at two different counters and assuming they should agree.
The practical cost is bad decisions. A business sees traffic fall in one report and panics, while the report that actually matters, the one counting sales, is flat. Or the opposite: the window count looks wonderful, the agency reports growth, and nobody notices that nobody is coming through the door.
Deciding which numbers you’ll judge the work by, before the work starts, prevents almost all of this. What SEO is for is a separate question from which numbers prove it happened.
The four kinds of numbers
Visibility. How often your site appeared in search results at all, and roughly where. Impressions and average position live here. This is the window count.
Traffic. How many people actually arrived. Clicks in Search Console, sessions in your analytics tool. This is the door count.
Rankings. Where your site appears for particular searches you’ve chosen to watch. Useful for tracking specific pages, but ranking isn’t one fixed number, since results differ by location, device, and person.
Business outcomes. Inquiries, bookings, sales, revenue. This is the till count, and it’s the only category that answers the question a business owner actually has.
Everything else is a step toward these. If a report is full of numbers that don’t connect to any of the four, ask what decision each one is meant to inform.
Why the numbers never match
Two reports covering the same week won’t agree, and Google documents why. The tools count different moments, run on different clocks, and follow different rules about what gets recorded and what gets withheld for privacy. None of that is a fault in either one.
The mechanics matter when you’re reconciling a specific figure, and they belong with the tools themselves rather than here. What matters at this level is knowing what normal looks like. A gap of a few percent between what Google says it sent you and what your analytics recorded is routine. A gap that suddenly changes shape is worth investigating.
What this looks like in practice
A dental clinic publishes twenty new pages. Three months later the report shows impressions up sharply, clicks up moderately, and average position down from 8.4 to 14.7.
Read as a rank, that looks like a collapse. It isn’t. The clinic now appears for hundreds of additional searches it never appeared for before, most of them at lower positions. Those new appearances drag the average down while adding real traffic. The old pages didn’t move.
This is the single most common misreading of an SEO report, and it happens because average position is an average across everything, not a rank.
What people usually get wrong
“Average position is our ranking.” It’s an average of many appearances across many searches, devices, and countries. It’s a trend line, not a rank.
“More impressions is always better.” Only if the people seeing you want what you sell. Appearing for searches with the wrong intent adds impressions and nothing else.
“One of these tools must be broken.” Usually neither. They’re counting different moments by design.
“The score in this tool is our Google score.” No tool outside Google has Google’s data. Scores from third-party tools are that company’s estimate.
What you can do about it
Set up Google Search Console if you haven’t. It’s free, it comes directly from Google, and it’s the only source that reports what Google actually did.
Agree, in writing and before work starts, on which numbers will be used to judge results. Put at least one business outcome in that list.
Compare periods of equal length, and prefer 28 days or more. Day to day movement in search is mostly noise.
When two reports disagree, check the time zone and the date range before assuming anything is wrong. Those two account for most of it.
Words you will hear
Impression. One occasion where a link to your site was shown in search results.
Click. One occasion where someone clicked from search results through to your site.
Session. One visit to your site, recorded by your own analytics tool.
Conversion. A visit that ended in something you care about, such as a form submission or a purchase.
Organic traffic. Visits that came from unpaid search results, as opposed to ads.
First-party data. Data collected by you or given to you directly by the platform, rather than estimated by an outside company.
For SERP, query and rich result, see What is a SERP? For crawler, index and ranking, see How search works.
