GA4 for non technical people: the five reports you actually need

Admit it, I will not tell anyone. You open Google Analytics once a month, with the air of someone leaning over the dashboard of a car to check whether a warning light has come on, look at the user count, think “good, it went up”, close the tab and carry on with your day feeling in control of something. Am I lying? Have you not done exactly this, more than once? If you recognise yourself even slightly, I have good news: you do not need more reports, you need five specific ones: where people came from, what they read, where they left, what they did before getting in touch, and how this month compares to the same month last year. Everything else in GA4 is for analysts, and analysts get paid precisely for that.

The problem with GA4 is not that it is hard. If only! Hard we would solve with one patient afternoon and a coffee, with time left over for a second coffee. The problem is that it looks like the old Analytics while measuring different things, so most people read the wrong numbers with total confidence. And total confidence resting on a wrong number is the most expensive consultant you will ever hire. He does not even send you an invoice; that is why it takes you so long to notice him.

First, let us agree on what the words mean

Four definitions, and if you skim past them every conclusion underneath changes shape. Here they are as Google states them, so nobody can accuse me of decorating.

Engaged session. A session that lasts longer than 10 seconds, or has a key event, or has at least 2 pageviews. Any one of the three. One, not all three together. That little “or” is doing all the work, and it is also the word everybody sprints past.

Bounce rate. The percentage of sessions that were not engaged. Now notice the consequence, because it is sneakier than it looks: someone who read an article for 40 seconds and left was a bounce in Universal Analytics. In GA4 they are not. So your bounce rate “improved” without a single thing on the site changing. Let us not open the champagne. The definition changed, not your business.

Users. The metric on the main screen is actually Active users, meaning users with at least one engaged session. It is not the same as total users. I know the screen does not shout this at you. It was never required to, they will tell you.

Key events. What we used to call conversions. The rename was announced on 21 March 2024, to separate them from the conversions counted in Google Ads. Nothing changed in how they are measured. The label changed, which in software happens rather more often than any of us would like… and always at the precise moment we had got used to the previous label.

And one technical detail with real practical weight: a session times out after 30 minutes of inactivity by default, with a configurable maximum of 7 hours 55 minutes. If someone leaves a tab open, goes to bed and comes back the next day, that is two sessions and one user. Same human being, two rows in the table. And you, smiling at the doubled number.

The five reports, and nothing else

That last row is the one that matters most for a Greek business, and I will keep insisting on it for as long as it takes. Comparing August with July tells you nothing, because seasonality swallows the signal whole. It is like measuring a child’s height while the child is standing on a step; you are measuring something, certainly, just not the thing you think you are measuring. Comparing August with last August tells you whether your work is working.

In the second row, average engagement time is more useful than views, and by a wide margin. A page with 300 views and a 12 second average is not being read; it is being walked past. A page with 60 views and 3 minutes is being read, and deserves to be expanded. Fewer people, more attention. Come to think of it, that is how human relationships work too, but let us not get philosophical in an article about metrics.

Why Search Console says one thing and GA4 another

The question I get most often, by some distance. Search Console says 1,200 clicks, GA4 says 800 organic sessions, and the client suspects something is broken. Or worse, that somebody is hiding visitors from them. I have sat through that scene, complete with the tone of a prosecutor.

Nothing is broken. They measure different things, at different moments of the journey:

• Search Console counts the click on the result. GA4 counts the page load. If someone hits back before the script loads, the click was counted and the session was not.
• In Greece, anyone who refuses cookies is not counted in GA4. Search Console counts them regardless.
• Two clicks by the same user within 30 minutes are two clicks and one session.
• Redirects, filters and differing time zones each add their own drift.

Google publishes eight specific reasons for the gap. The practical conclusion fits on one line: do not compare the two tools in absolute numbers. Compare each one against itself over time. Two clocks that run slightly differently are never reconciled; you simply watch how far each of them has travelled.

Two more things about Search Console while we are here. Data retention is 16 months, so if you want history, export it before you lose it; nobody will send you a warning on the day it goes. And queries not issued by more than a few dozen users over two or three months do not appear at all, for privacy reasons. That is why the sum of your queries never adds up to your total clicks. No, your arithmetic is not the problem. Nor is your calculator.

The one setting that quietly deletes your history

This is where I have watched two years of history disappear. No alarm, no warning, nobody personally at fault. One day you simply ask for a year on year comparison, and the screen looks back at you like the clerk who says “we do not have that” and returns to his coffee.

GA4 has a data retention setting for user and event level data. On standard properties the options are 2 or 14 months. The longer 26, 38 and 50 month options exist only on Analytics 360 properties, the paid tier. Age, gender and interest data is always kept for 2 months regardless.

What it actually affects, so we panic no more than necessary: only explorations and funnels, the reports you build yourself. Standard aggregated reports are not affected. Breathe out.

Go to Admin, Data settings, Data retention right now, not at some point, and see what it says. “At some point” is the favourite hiding place of everything that never gets done. If it says 2 months and you want year on year comparisons, change it to 14. And hold on to this part, because it hurts: the change is not retroactive.

And a reminder for anyone with old history they assume is safe: Universal Analytics stopped processing data on 1 July 2023 for standard properties and 1 July 2024 for 360. Access was removed and the data deleted during the week of 1 July 2024. Whatever was not exported then is gone. For good. And “for good” is one of the very few phrases in software that means exactly what it says.

Why the numbers refuse to add up

Two mechanisms that make numbers refuse to add up, which nobody explains. Let us get them out of the way here.

Sampling. Applies in explorations when a query exceeds 10 million events on a standard property. Above that, GA4 answers from a sample. You will see an indicator at the top of the report, assuming you notice it, which nobody does when they are in a hurry. For standard reports, Google states that sampling is possible, not that it is ruled out.

Thresholding. When a data set is small enough that it could reveal an identity, GA4 simply hides the row. You will see a warning icon. That is why individual figures sometimes do not sum to the total, and why you add them again, blaming your arithmetic. Your arithmetic is fine; privacy is just doing its job.

High cardinality. Dimensions with more than 500 unique values per day get grouped into a single (other) row. If you see it often, unfiltered URL parameters are usually the cause. Think of (other) as the drawer where everything we never got round to tidying ends up.

And now consent, measured in Greek

Two separate issues that get constantly conflated, even by people who know better. Let us put each one back where it belongs.

The first is legal and Greek: Recommendations 1/2020 of the Hellenic Data Protection Authority name Google Analytics explicitly and require consent before it loads. There is no “statistics exemption” in Greece today, however hard you look for one. And I know you looked. I covered this in the article on cookies and the GDPR.

The second is Google’s own requirement of its advertisers: since March 2024 it requires consent signals to be passed for EEA users, through what it calls Consent Mode. In basic mode the scripts do not load at all before consent. In advanced mode they load but send anonymous signals.

In practice, a correctly configured Greek site will show fewer users in GA4 than it actually has. I know, it stings. It is a bit like counting the guests at a wedding using only the people who signed the guest book. It is also normal, and it cannot be fixed. What can be fixed is the comparison: look at trends, not absolute magnitudes.

The ten minute check, once a month

1. Compare with the same period last year, not with last month.
2. Traffic acquisition: which channel moved, and why.
3. Landing pages: which entry page has the worst engagement.
4. Key events per channel: where the enquiries actually come from.
5. One look at Search Console: which queries gained position without gaining clicks. That is usually where a bad title is hiding.

Ten minutes. About one supermarket queue, with a better return.

So what do we keep from all this

GA4 is not harder than what came before. It is more honest, in that it measures engagement rather than visits, and more limited, in that consent and thresholds hide part of the picture. Neither your enemy, then, nor your oracle. It is an instrument, and instruments need somebody who knows what they are reading on them.

If you keep two things out of everything we have said: always compare with last year, and check your data retention setting today, before you need history that no longer exists. Remember that user count we were looking at once a month at the start, in the manner of a warning light on a dashboard? Looking at it was never the mistake. The mistake was that it was the only thing you looked at.

If you want to go through what your GA4 is and is not measuring correctly, get in touch.