Google Analytics 4 is capable and unfriendly, and most small businesses install it, look at the visitor count occasionally, and derive nothing useful from it.
That is a reasonable response to the interface. GA4 exposes hundreds of reports designed for organisations with analysts, and the ones a business owner actually needs are buried behind configuration that nobody explained.
This is the short version: configure four things, look at four reports, ignore the rest.
Consent, and what it does to your data
Analytics data is less complete than most people assume, and knowing by roughly how much prevents wrong conclusions.
Browser privacy protections, ad blockers and cookie consent choices all suppress a share of measurement. Depending on your audience that can be ten per cent or considerably more, and it is not evenly distributed: technically literate audiences block more, so a B2B technology site is undercounting more than a consumer retail one.
The practical consequence is that absolute numbers are understated and should not be treated as a census. What remains reliable is the comparison: trends over time and proportions between channels, because the suppression is reasonably consistent month to month.
If you operate a consent banner, how it is configured changes what you collect. A banner defaulting to refusal collects materially less than one defaulting to acceptance, and that is a legal and ethical decision rather than an analytics one. Whichever you choose, do not compare periods across a change in banner configuration and read the difference as a change in the business.
Reading the numbers honestly
The most common analytics failure is not configuration. It is reading a change as a result when it was seasonality, a data artefact or something you did elsewhere.
Compare against the same period last year rather than against last month wherever you have the history, because almost every business has seasonal patterns that swamp the effect of anything you did. A twenty per cent drop in October may be entirely normal for your category.
Be careful with small numbers. A move from four enquiries to six is a fifty per cent increase and it is also two enquiries, which is well within the range of random variation. Below roughly thirty conversions a month, treat month-to-month movement as noise and look at quarters instead.
Watch for data artefacts before concluding anything. A tracking change, a new tag, a bot pattern, a referral exclusion added or removed, or a campaign parameter that was not tagged consistently will all produce movement that means nothing.
And write down what you expect before you look. A prediction that turns out wrong teaches you something about the business; a number interpreted after the fact almost always confirms whatever you already believed.
Set up key events first, or nothing else means anything
Analytics without conversion tracking measures traffic, and traffic is not what you care about. Until GA4 knows what a successful visit looks like, every report is about volume.
For most businesses the events worth tracking are simple: a click on the phone number, a click on the WhatsApp link, a contact form submission, and for eCommerce the purchase. Mark these as key events in the admin, because that is what makes them appear as conversions throughout the reporting.
The phone and WhatsApp clicks matter more than people expect in this market, because a large share of enquiries never touch a form. A business measuring only form submissions is frequently seeing a fraction of its actual enquiries and drawing wrong conclusions about which channels work.
Test each one after setup by triggering it yourself and confirming it appears in the realtime report. Events configured and never verified are extremely common and silently useless.
In practice
The four reports that matter
Everything a small business needs to act on is in four places.
Common setup mistakes
Four errors account for most of the situations where GA4 data is misleading rather than merely unused.
Not filtering internal traffic. Your own team, your developer and your agency visiting the site repeatedly inflate everything and distort which pages appear popular. Exclude your office IP addresses in the admin.
Not excluding payment gateway referrals. When a customer returns from a payment provider, GA4 may attribute the purchase to that provider rather than to the channel that actually brought them, which makes your acquisition reporting wrong precisely where it matters most.
Running two tracking implementations at once, usually a tag added directly and another through a tag manager, which double-counts everything.
And not linking Search Console. The link takes a minute and adds the query data that explains why organic traffic moved, which GA4 alone cannot tell you.
What to ignore
A great deal of GA4 exists for organisations with analysts and a business owner can safely ignore it.
Bounce rate, which in GA4 is defined differently from the old metric and is frequently misread as a result. Engagement rate is the more meaningful number and it is the inverse.
Explorations and custom funnels, until you have a specific question that the standard reports cannot answer. Building one prematurely produces an impressive report nobody uses.
Audience building and remarketing lists, unless you are actually running remarketing campaigns.
And the attribution modelling comparisons, which matter when you run several paid channels simultaneously and are noise when you do not.
The discipline worth adopting instead is looking at the four reports monthly, writing down what changed and why you think it changed, and checking next month whether you were right. That produces more understanding than any additional configuration.
Key takeaways
- Without key events configured, every report is about traffic rather than about outcomes.
- Track phone and WhatsApp clicks. In this market most enquiries never touch a form.
- Four reports suffice: traffic acquisition, landing pages, engagement, devices, all with conversions.
- Exclude internal traffic and payment gateway referrals, or the data misleads where it matters.
- Verify every event by triggering it and watching realtime. Unverified events are silently useless.
- Ignore explorations and attribution modelling until you have a question the basics cannot answer.
Frequently asked
Four key events: phone number clicks, WhatsApp clicks, form submissions and purchases where relevant. Mark them as key events in the admin so they appear as conversions throughout the reports. The phone and WhatsApp clicks matter most in this market, because a large share of enquiries never touch a form and businesses measuring only forms see a fraction of reality.
Usually definitional differences plus configuration gaps. GA4 defines sessions and engagement differently from the old Universal Analytics and from most third-party tools. More often the cause is internal traffic not excluded, payment gateway referrals distorting attribution, or two tracking implementations running simultaneously and double-counting.
Less than people assume, because GA4 defines it differently from the old metric and it is frequently misread as a result. Engagement rate is the more meaningful figure and bounce rate is its inverse. For most business decisions, landing pages with conversions tells you more than either.
Yes, and it takes about a minute. GA4 alone cannot tell you which queries brought organic traffic, so when organic numbers move you have no way to explain it. The link adds that query data into the reporting and it is one of the highest-value pieces of configuration available.