Google Analytics for people who do not want to learn it

Four numbers, two screens

Open Reports, then Acquisition, then Traffic acquisition. That single screen shows how many sessions arrived and which channel each came from: Organic Search, Paid Search, Direct, Organic Social, Referral, Email. Then open Reports, Engagement, Pages and screens, which tells you which pages people actually landed on. Those two screens are the whole monthly review for most local sites.

One warning about Direct. It is a bucket for anything that arrived without a referrer: typed addresses, bookmarks, links tapped inside text messages and email apps, links from PDFs. A large Direct number is not proof that your brand is strong. It is usually proof that some of your traffic lost its label on the way in.

The fourth number, the one that actually matters, sits on the right side of that same acquisition screen once you have told Analytics which events count. Read it as a rate rather than a total. Forty sessions from search producing four form submissions tells you something useful about the page. Four hundred sessions producing four tells you something useful about the page too, and it is not the same thing.

Engagement rate is not the old bounce rate

People who used the previous version of Analytics keep reading engagement rate as bounce rate flipped around, and it is not the same measurement. A session counts as engaged if it lasted longer than ten seconds, or triggered a key event, or included at least two page views.

That changes how you read a service page. Someone who lands, reads the answer for fifteen seconds and taps the phone number counts as engaged, and that visit was a complete success. Under the old metric, a single-page visit that produced a phone call looked like a failure. Do not compare the two numbers to each other or to anything a previous agency reported.

Mark your key events, or none of this means anything

GA4 collects events automatically but does not know which ones represent money. Go to Admin, then Events, and switch on Mark as key event for the ones that are worth something: the form submission, the click on the phone link, the booking confirmation.

Two traps here. First, enhanced measurement collects scrolls, outbound clicks and file downloads on its own, and none of those are worth marking. Second, a tap on a phone number is not always captured by default depending on how the link is built, so check by tapping your own number on a phone and looking for the event within a day. A surprising number of the accounts we inherit have been reporting zero calls for a year because that one event was never set up.

Also be careful about counting a thank-you page view as the conversion. If the form can fail silently, the page view still fires and you record a lead that never reached anyone. Test the form monthly by submitting it yourself.

Two things that quietly ruin the data

Your own visits. Go to Admin, Data streams, then the tag settings, and define your office network as internal traffic, then activate the filter. On a site receiving two hundred sessions a month, staff and contractor visits are a large share of the total, and they make your most-visited page the one your bookkeeper opens every Tuesday.

Double tagging. If the measurement code sits in the site theme and also in a tag manager or plugin, everything gets counted twice. The tell is that Analytics reports roughly double what your ad platform or host reports for the same period. Both problems inflate the numbers in the direction owners want to believe, so nobody notices.

How much traffic before a change means anything?

More than most small sites have. With a couple of hundred sessions a month, week-to-week comparisons are noise, and a single busy day from a shared Facebook post can double your week. Compare quarters, or the same month against last year, and treat direction as the finding rather than percentages.

This is also why we do not report weekly numbers for small accounts. A weekly chart on thin data produces panic in bad weeks and false confidence in good ones, and both cost the client more than the report was worth.

What Analytics cannot see

It cannot see a phone call placed after somebody wrote the number down and called from the kitchen an hour later. It cannot see walk-ins. It cannot see the neighbor who was told about you. It cannot see visitors who declined cookies or use blockers, and that shortfall is not random, so your real traffic is always somewhat higher than the number on screen.

Most of all, it cannot tell you which session became a paying job. That link exists in your books and your call log, not in the tool. Use GA4 to answer which channels are growing and which pages get found. Use what you invoiced to answer which channel made money. When those two disagree, the invoices are right.

That gap is worth saying out loud because reporting gets sold as certainty. A monthly summary full of sessions, engagement rate and average time on page can look impressive while telling you nothing about revenue. The version worth reading is short: traffic by channel, the two or three pages that bring people in, key events, and a sentence about what changed and why. If your report does not fit on one page, somebody is padding it.