GA4 for Owners: The Five Numbers That Matter

A pattern we see in nearly every audit: the business has Google Analytics installed, someone logged in once, found a menu offering Reports, Explore, Advertising and Admin, landed on a home screen of squiggly lines with no obvious meaning, and never went back. The tag keeps firing. Nobody reads anything. Two years of data accumulates and gets used for nothing.

That isn't the owner's fault. GA4 was built for people who buy media professionally — audience builders, attribution modellers, teams wiring analytics into Google Ads. If you run a plumbing company or a law firm or a six-person agency, most of the interface isn't addressed to you, and the parts that are sit underneath the parts that aren't.

Here's the claim worth arguing about: if your property has no key events configured, everything else in the account is theatre. Sessions, users, engagement rate — none of it tells you whether the website earned a single enquiry. In the audits we run, an Australian small business property with zero key events is the norm rather than the exception, which means most owners who do log in are reading the wrong screen carefully.

Five numbers are worth your attention. Fifteen minutes a month is the right amount of time to spend on them.

The five numbers, and where each one lives

That's the whole dashboard. Never open Explore, never touch Advertising, and you lose nothing you were going to use.

Start with key events, because without them nothing counts

An enquiry is the only number your business actually trades in, and GA4 won't record one unless you tell it what an enquiry looks like on your site. There are two reliable ways to do that and one popular way that fails quietly.

The thank-you page method. If your contact form redirects to a confirmation page — /thank-you, /enquiry-received, anything distinct — this is the sturdiest option. In Admin, under Data display, open Events and choose Create event, then build one that fires when page_location contains thank-you. Name it something plain like enquiry_submitted, then switch on Mark as key event. Four minutes, and it never breaks, because a page load is the least fragile thing on the internet.

The deliberate event method. If your form submits without a page change — most modern custom builds do — the form has to fire an event itself when the submission succeeds. That's a line of code in the build, not a setting in GA4.

The method that fails quietly. GA4's enhanced measurement includes automatic form tracking, and it's tempting to assume it has you covered. Google's own documentation notes those automatic events don't work with every form implementation, and in practice they miss plenty of custom and embedded forms. Check the count against your actual inbox before trusting it.

The same applies to phone taps. On mobile a tel: link tap is often the highest-value action on the whole site, and GA4 doesn't count it by default in any way you can rely on. When we build sites for small businesses, phone taps, form submissions and booking completions get wired as key events during the build, because retro-fitting means the first six months of data are blank.

One warning: mark three to five things as key events, not fifteen. Newsletter signups and scroll depth are measurable too, and flagging everything buries the number that matters.

Where they came from, and why direct is lying to you

Traffic acquisition groups sessions into channels. Organic Search means Google or Bing sent them without you paying. Paid Search means you paid. Referral means another website linked to you. Direct means GA4 has no idea.

Direct is the bucket that misleads owners most, because it reads as "people who know our brand and typed the address in," which flatters everybody. In reality it collects anything arriving without a referrer: links from email clients, links inside PDFs, taps from a messaging thread. A large direct share isn't evidence of brand strength. It's evidence of missing information.

The other trap is comparing Traffic acquisition against User acquisition and finding they disagree. They measure different things on purpose: Traffic acquisition is session-scoped, User acquisition is first-touch — where that person originally found you, possibly months ago. For a monthly read, Traffic acquisition is the more useful of the two.

If you run advertising or send email campaigns, tag those links with UTM parameters. Untagged campaign traffic scatters across Direct and Referral, and you end up judging paid spend on a number that includes people who found you free.

The landing page report, and the mobile split

Most owners look at Pages and screens, which ranks every page by views. Wrong report for a monthly read — it's dominated by whatever people click through to internally.

The Landing page report answers a sharper question: which pages are the front doors? Sort by sessions and you'll usually find something unexpected. A service page written in an afternoon out-pulling the homepage. A blog post from two years ago carrying a third of your organic traffic. Improving the page that gets four visits a month is a hobby; improving the one that gets four hundred is work.

Then add key events to that report as a column, and you can see which front doors produce enquiries and which produce polite browsing. High sessions with no key events is one of the most common findings in any audit, and it usually means the page ranks for a query that isn't a buying query — a diagnosis we've walked through in traffic but no enquiries.

Mobile share sits one report away, and it earns its place because it changes your priorities rather than your scorecard. If seven in ten sessions are mobile, load time on a phone is most of your first impression and belongs at the top of the fix list. If the split runs the other way, as it does for plenty of B2B, the desktop layout deserves the attention instead.

Direction of travel, and the setting that deletes your history

A single month means almost nothing in a small business. Forty enquiries is good or catastrophic depending entirely on what last August looked like.

Every GA4 report has a date selector with a comparison option. Turn it on, set it to the same period a year earlier, read the percentage change. Year-on-year survives seasonality; month-on-month mostly tells you how many public holidays there were.

Before that works, there's a setting to fix. Standard GA4 properties default to retaining event-level data for two months. You can change it to fourteen, free, in Admin under Data collection and modification, then Data retention. The nuance: standard reports keep their aggregated totals regardless, but anything you want to slice yourself — a custom exploration, a year-on-year breakdown by landing page — depends on the underlying event data still existing. Do it before you finish this article. It applies from the moment you change it, and not at all to anything already deleted.

What to ignore, with reasons

Bounce rate. GA4 counts a session as engaged if it lasts longer than ten seconds, or fires a key event, or includes at least two page views. Bounce rate is the inverse. So a visitor who reads your services page for nine seconds and then rings you counts as a bounce, while a visitor who leaves a tab open counts as engaged. The threshold is adjustable up to sixty seconds, so the metric moves when you change a setting rather than when your website improves.

Real-time. Genuinely entertaining, completely useless. It exists to verify a tag fires after installation. Nobody has ever made a business decision from it.

Average engagement time. Averaged across people who found what they wanted immediately and people who found nothing at all. Those two groups pull in opposite directions.

Knowing the enquiry count is only half the read — you also need a sense of what share of visits should become enquiries. We've set out realistic benchmarks in what a good conversion rate looks like for an Australian service business, and for longer decision cycles, the case for measuring a handful of stage-level metrics instead of everything at once applies well beyond schools.

GA4 has never counted everyone, and it's counting fewer

Safari caps the lifetime of cookies set by JavaScript at seven days, so a customer who visits early in the month and returns at the end is very likely counted as two people. Ad and tracker blockers stop the tag loading entirely for a meaningful slice of visitors. Consent banners, implemented properly, mean some visitors are never measured at all.

Privacy expectations in Australia are tightening rather than loosening, and the Office of the Australian Information Commissioner has been steadily clearer about what businesses must disclose regarding data collected through their websites. The consequence isn't that GA4 becomes useless — it's that the totals were never a census and are drifting further from one. Ratios and directions hold up even when the underlying count is incomplete: organic against paid, this year against last, which landing page converts and which doesn't.

FAQ

Do I need Google Analytics 4 at all for a small business?

If you spend money on marketing, or you're deciding whether the website deserves further investment, yes. It's free, and the alternative is judging your website on how busy the phone feels. If you spend nothing and take every enquiry by word of mouth, an account you never open isn't worth setting up.

What replaced Universal Analytics, and does my old data still exist?

GA4 replaced it. Universal Analytics stopped processing new data on 1 July 2023 and Google has since shut down access to the old properties. If nobody exported those reports beforehand, that history is gone. Sites that had a tag installed but were never migrated have a gap starting mid-2023 — worth checking, because plenty do.

How do I know if my GA4 is set up correctly?

Three checks. Open Reports and confirm the last 24 hours shows sessions; a flat zero means the tag isn't firing. In Admin under Data display, open Key events and confirm at least one is listed. Then submit your own contact form and watch that count rise within the day. Our free website audit also flags whether an analytics tag is present and firing, which is the quickest way to check a site you didn't build.

Is fifteen minutes a month really enough?

For reading, yes. Five numbers, a year-on-year comparison, ten minutes of thought about what changed. The time-consuming part is setup, and that's a one-off. The failure mode we see isn't owners spending too little time in GA4 — it's an hour spent inside an account that was never configured to answer their question.

What to do this week

Open analytics.google.com and, in Admin, change data retention from two months to fourteen. Ten-second job, longest tail.

Then check whether a single key event exists. If not, set up the thank-you page trigger using the steps above, or send those steps to whoever maintains your site. Until that's done, every other number in the account describes traffic rather than business.

Once it's running, put fifteen minutes in the calendar for the first Monday of each month. Traffic acquisition, Landing page, Key events, Tech details, year-on-year comparison. Write the five numbers in a note. After three months you'll have your own baseline, which no benchmark article can give you. Google's Analytics Help centre covers the setup detail if you'd rather work through it yourself.

If the numbers turn out to be discouraging — real traffic, almost no enquiries — that's a website problem rather than an analytics problem, and it's the one we're most often called in to fix. Custom builds start at $8,000 AUD; to work out whether the maths stacks up before spending anything, book twenty minutes and bring the five numbers with you.