Google Analytics 4

Google Analytics 4 is Google’s free reporting tool for websites and apps. It records what visitors do, groups those actions into reports, and connects them back to where the visitor came from. The number after the name matters because it replaced a system that worked on a genuinely different model, and most of the confusion around it comes from people looking for reports that no longer exist.

In short

• It records everything as an event, including page views.
• It replaced Universal Analytics, which no longer collects or reports data.
• It will not tell you anything useful until somebody marks which events matter.
• It counts differently from ad platforms, and the totals are not meant to match.

The shift from sessions to events

The old system was built around the visit. A person arrived, did some things, and left, and the reports were mostly about visits and the pages inside them. That model came from an internet where people used one device and one browser.

GA4 is built around the action instead. Everything is an event: viewing a page, submitting a form, playing a video, clicking away to another site. A visit still exists as a grouping, but it is no longer the unit the system thinks in. The practical consequence is that GA4 can answer questions about behaviour that the old reports could not, and it answers the simple question of how many people came in a way that looks unfamiliar to anybody used to the previous tool.

What it can tell you

The useful questions it answers are narrower than the interface suggests, and there are about four of them.

Where visitors came from is the first, split into search, paid advertising, social, referrals and people who arrived directly. Which pages actually recruit those visitors is the second, and it is regularly not the pages the business believes are important. The third is where people give up inside a sequence such as basket, checkout and payment. The fourth is how the phone compares to the desktop, which in most Philippine markets is not a close contest and should shape what gets built next.

Everything beyond those four tends to be interesting rather than actionable, which is a distinction worth making before an afternoon disappears into it.

What it cannot tell you

Being clear about the limits prevents most of the arguments that follow a report.

• Who anybody was. Reporting is aggregated and no personal identity is attached, so it will never produce a name to follow up.
• An exact total. Blocked scripts, refused consent and privacy settings mean the figures are a large and consistent sample rather than a census.
• Anything older than the retention ceiling, once you leave the standard reports. Detailed explorations reach back two months by default and fourteen at most, and raising the setting does not apply retrospectively.
• A single true version of a conversion. GA4 spreads credit across the path it can see, which is a defensible method and not the same method your ad platforms use.

The wider limits of measurement, including everything that happens in Messenger or at a counter, belong to tracking rather than to this tool specifically.

Setting it up so it is worth having

An installed property that nobody configured is the most common state to find one in, and it is close to useless.

• Install it once, properly. Two copies of the code on the same site will double count everything and take a while to notice.
• Decide what counts as success and mark those events as key events, which is what GA4 now calls what everybody else still calls conversions. Until that is done, the reports describe traffic rather than results.
• Filter out your own visits, along with the developer, the agency and anybody else who is not a customer.
• Connect the advertising accounts, so paid traffic can be compared against everything else in one place.
• Write down what each event means. A property whose event names only made sense to the person who left is a property that gets rebuilt.

Analytics is one part of tracking rather than all of it, and a property configured without the rest of that setup will answer fewer questions than its interface suggests.

The handful of reports worth your time

The interface offers far more than a small business needs, and four places answer most of the questions anybody actually asks.

• Traffic acquisition, which splits arrivals by where they came from. This is the report to open first, and often the only one needed to settle an argument about which channel is working.
• Landing page, which shows where people entered rather than every page they touched. It tells you which pages are doing the recruiting.
• Events, which lists what people did, and by extension shows immediately whether the setup was ever finished.
• Realtime, which is close to useless for decisions and invaluable for checking that a change you just made is recording correctly.

Everything else can wait until one of those four raises a question it cannot answer on its own.

Reading the reports without fooling yourself

Reports invite conclusions faster than they support them, so a few habits are worth building.

Compare like with like. A quiet week against a holiday week says nothing. Look at the same period last year rather than at last month when the business is seasonal, which most are.

Ignore a metric until you can say what would make it better and whether you would want that. Time on page rises when people cannot find what they wanted. Bounce rate falls when somebody marks one more event as a key event. Numbers that move for two opposite reasons cannot be used on their own.

Treat any single day as noise. The signal in a small business account lives at the level of weeks, and the temptation to act on a Tuesday is the source of a great deal of unnecessary work.

Why it is worth having at all

Every advertising platform reports on the traffic it sold you, and each one credits itself generously by its own consistent definition. Add up the conversions claimed by three of them and you will have more customers than the business actually had.

Analytics is the one place where paid, organic, direct, referral and social are counted by a single set of rules. That is the whole reason to install it. Comparing pay-per-click against search visibility is a question nothing else can answer, and a business that never configured its key events has an installed tool that cannot answer it either. The reason the totals never agree with the ad platforms is explained once, properly, under tracking.