The first half hour in Google Analytics 4 feels like a cockpit with no manual. Cards, tiles, dropdowns, a menu full of items that all seem to show roughly the same thing. The problem is not that GA4 shows too little. It shows too much, and the useful part sits buried under three layers of clicks.
You need four reports. Everything else is decoration for day-to-day work.
Report 1: acquisition by session, not by user
Under Reports, Acquisition, Sessions. This is where visitors come from, broken into "Organic Search", "Direct", "Referral" and so on.
The mistake almost everyone makes is looking at user acquisition instead of session acquisition. The difference is not a detail. User acquisition credits each person to the source that first brought them in. Somebody who found you through Google three months ago and types the URL directly today counts as a Google visitor forever. Session acquisition credits each individual visit to its actual source. For the question "what is bringing me visitors right now", only the second one is any use.
Report 2: pages and screens
Reports, Engagement, Pages and screens. A list of your most-visited pages with average engagement time beside each.
That engagement time is the number where the old world and the new one part ways. Universal Analytics had "average session duration", calculated from timestamps, and for single-page visits it was structurally zero. GA4 instead measures the time the tab is actually visible in the foreground. Closer to the truth, but a different number. Comparisons from an old report or somebody else's industry benchmark no longer fit. How to read these per-page numbers at all is covered in more depth in analyzing user behavior.
Report 3: audiences, but built yourself
The prebuilt audience reports are pretty and mostly useless. Value only appears when you set up a comparison. Click "Add comparison" at the top, build a segment of every session with a purchase or a submitted enquiry, and place it next to the rest.
What you almost always see: the people who convert differ from the average in exactly one thing. They came through a particular landing page, they used desktop rather than a phone, or they saw one specific page along the way. That one thing is your lever. Without the segment you average it away with everyone else.
Report 4: realtime, but only for checking
The realtime report tempts you to stare at it for an hour and tells you close to nothing about the state of your site. It has exactly one good use: checking whether a measurement you just added works. You click the button you want to track on your own page and watch the event appear within seconds, or not. For everything else, close it again.
The custom report you build yourself
GA4 lets you customize reports, and that is the moment the tool flips from annoying to useful. The one report I set up for everyone: landing page as the row, with sessions, engagement rate and conversions beside it.
That answers the question that matters at a glance: which entry page brings plenty of visitors who then do not convert? That is the page where work makes the biggest difference. A page with a thousand sessions and two percent engagement is a different problem from one with fifty sessions at the same rate, and the default page report blends the two together.
Explorations are a time trap
GA4 has a second area called "Explore" that lets you build almost any analysis you like: funnels, paths, free-form tables. Tempting, and for ninety percent of sites a waste of time.
The reason is uncomfortable. A four-step funnel looks analytical, but when half the people drop between step two and step three, the funnel only tells you that it happens, not why. The answer is rarely in GA4 and nearly always on the page itself. Build the elaborate analysis only once the four standard reports have raised a concrete question they cannot answer. Before that you are drawing handsome charts about a problem you do not understand yet.
What to set up before you analyze
Analysis assumes the right thing is being measured in the first place. On most freshly set-up accounts, page views come in but no conversions, because nobody defined a key event. Then every report shows an empty conversion column and half the analysis falls away. How to fix that in a few steps is in setting up conversion tracking.
And one source GA4 does not know belongs beside it: Google Search Console shows you which search queries brought people to your site in the first place. GA4 tells you what they do afterwards. Only together do they form a picture. How to connect Search Console is in setting up Search Console.
The filter almost every account forgets
One thing distorts every analysis until it is fixed: your own visits. Anyone working on their own site daily produces sessions that land in the same reports as real customers. On a small site with thirty visitors a day whose owner drops by ten times, a third of the data is noise.
The fix sits under Admin, Data streams, the web stream, then "Define internal traffic". There you enter your own IP address (find it by googling "my IP"). After that these sessions can be excluded from the report. It is fiddly and it is a one-time job, but without this filter you are measuring yourself for a good part of it.
Where to start tomorrow
Open the landing page report you built, sort by sessions, and find the first row with plenty of visitors and a low engagement rate. Open that page on your phone and ask whether it answers the question somebody arrives with. Usually you find the answer in the first ten seconds, and it is not in Analytics.