Nearly every business owner we work with has Google Analytics installed and almost none of them have ever got anything useful out of it. They log in, land on a screen of cards nobody asked for, click three things, find a number that seems to have gone down, and close the tab.
That is not a failure of the owner. GA4 was built for organisations with analysts, and it assumes you want to define your own measurement model. Most small businesses want to know four things, and the tool does not lead with any of them.
Here is the short version: there is some setup you genuinely have to do or the data is not worth reading, and then there are four reports worth your time. Everything else you can ignore for years without missing anything.
First, the setup that makes the data trustworthy
Skip this part and everything downstream is noise.
Define your key events. GA4 does not know what matters to your business, and it will not guess. You have to tell it which actions count (a form submission, a call button tap, a WhatsApp click, a completed booking) and mark those as key events. This is the single most important step, and it is the one most commonly skipped. Google renamed "conversions" to "key events" a while back, which is why older guides and the current interface disagree; they are the same idea.
If nothing is marked, the tool can only tell you how many people visited. That is the least interesting fact available.
Filter out your own traffic. You, your staff, and whoever built the site visit constantly. On a site with modest traffic that is enough to distort everything. Set up internal traffic filtering so your own visits are excluded, and check that it is switched to active rather than left in testing mode, a common half-finished state.
Fix your referral exclusions, especially for payments. This one bites hard locally. If a customer goes off to a payment provider to pay and comes back to your thank-you page, GA4 counts the return as a brand new visit from that provider. The sale gets credited to the payment gateway rather than to the ad, search, or post that actually earned it. Add every payment provider you use to the unwanted referrals list. Businesses have concluded their ads do not work on the strength of this misattribution alone.
Change the data retention setting. The default keeps user-level data for a short window, which means when you go looking at last year you find nothing. Set it to the longest option available. It is one dropdown and it is only useful if you change it before you need the history, not after.
Link Google Ads, if you run ads. Without the link the two systems tell you different stories and you cannot see which campaigns produced which behaviour on site.
Set that up once and the numbers become worth reading.
The four reports worth opening
1. Traffic acquisition: where people came from.
This tells you which channels are actually bringing people: organic search, paid, direct, social, referral. Compare it month over month rather than staring at one month, and look at it alongside your key events rather than just the visitor count. That last part is what makes the report worth opening. A channel that sends a third of your traffic and none of your enquiries is a very different thing from one that sends a tenth and half your enquiries.
2. Landing pages: where they arrived.
The page somebody entered on tells you what is actually pulling people in, which is almost never what you expected when you built the site. This is where you discover that one service page is doing most of the work, or that a page you were about to delete is the busiest entry point on the site.
It also tells you where to spend effort. The page that brings the most people and converts the worst is your highest-value thing to fix.
3. Key events over time.
Your enquiry actions, plotted over months. This is the closest thing GA4 has to a scoreboard. You are looking for direction, not precision: is this going up, is it flat, did it move after we changed something.
4. Pages and screens: what people actually read.
Useful mainly for deciding what to write more of and what nobody wants. If your articles or service pages are being read, this is where you see it. Lower priority than the other three, but it stops content decisions being pure opinion.
That is the lot. Four reports, once a month.
What to ignore
Real time. Genuinely fun to watch on launch day and useless for every decision you will ever make.
Demographics and interests. On low traffic volumes the samples are small and the inferences are shaky. Do not build a strategy on it.
Bounce rate. GA4 defines it as the inverse of engagement rate, and for a local business it is actively misleading. Somebody landing on your contact page, reading your phone number, and calling you is recorded as a bounce and is one of the best outcomes available.
Most of the default home screen. The cards are automated suggestions, not a considered summary of your business.
Explorations, for now. Powerful, and a rabbit hole. There is nothing in there you need before you have the four reports above as a habit.
The honest caveat about accuracy
GA4 undercounts. Ad blockers, privacy settings, browsers that limit tracking, and people declining cookie banners all mean a real share of your visitors are never recorded. The gap is not small and it is not consistent between channels.
This does not make the tool useless. It makes it a relative instrument rather than an absolute one. Use it to compare periods, channels, and pages against each other, where the undercount roughly cancels out. Do not treat any single figure as the truth.
Which is why the number you actually manage the business on is not in GA4 at all. It is the count of real enquiries you keep yourself, from every channel including the phone and WhatsApp. GA4 explains movements in that number. It does not replace it, and if the two ever disagree badly, believe your own enquiry log and go find out what is broken in the tracking. There is more on which numbers to keep here.
A monthly routine that takes fifteen minutes
- Open traffic acquisition, set the comparison to the previous month, and look at key events by channel. Anything moved sharply?
- Open landing pages. What is bringing people in, and did that change?
- Look at key events over three to six months. Direction?
- Write down one thing you will change, and check it next month.
That is it. If a report does not change a decision, stop opening it.
If you would rather this was configured properly once (key events defined, calls and WhatsApp tracked, ads linked, payment referrals excluded) and then reported to you in plain language every month, that is part of what we run. Most of the value is in the setup, and it is a one-time job.