Setting Up Google Analytics 4 and the Metrics That Matter

Installing Google Analytics 4 takes about ten minutes. The dozens of reports you face afterwards, though, make many business owners give up in the first week. Our advice is simple: don't try to measure everything. Measure the four or five actions that mean money for your business, and the rest gets collected anyway.

What should you measure with GA4?

A website's visitor count on its own says very little. Imagine a hypothetical service business site with 2,000 visits a month. If 40 of those fill in the appointment form and 60 tap the phone number, the site is doing its job. If visits rise to 4,000 while form submissions stay flat, the extra traffic is probably coming from the wrong people.

So the real question is what visitors do. For service businesses, these usually matter:

• Submitting the contact or quote form
• Tapping the phone number
• Clicking the WhatsApp button
• Clicking the email address
• Site search (on sites that have it)

For an online store, add to cart, begin checkout and purchase join the list. The logic is the same: first answer “what is this site's job,” then measure that job.

Setup steps

Go to analytics.google.com with your Google account and create a new account with a property under it. Two settings cause headaches later, so get them right from the start: time zone set to Turkey and currency set to Turkish lira.

Then add a web data stream and enter your site's address. Google gives you a measurement ID starting with G-. There are three ways to add it to your site: paste it into the relevant field in the site's admin panel, add it through Google Tag Manager, or place the code directly in the page template. If you plan to set up several events, Tag Manager makes life easier in the long run.

A few settings are usually skipped after installation:

Data retention is set short by default. Raise it to the longest period offered in admin settings; you can see the current options on the settings screen. Otherwise some reports may come up empty when you compare with the same period last year.

Enhanced measurement comes switched on and automatically collects actions like scrolling, outbound clicks, site search and file downloads. Its form interactions option can fire on some sites without an actual submission. We recommend setting up form tracking yourself rather than relying on it.

Define visits from your own office as internal traffic and filter them out. A team that opens the site dozens of times a day can seriously distort a small site's data.

Finally, link Search Console to GA4 so you can see what people from specific searches do, all in one place. We covered Search Console setup in a separate post.

When everything's done, browse your site in Tag Manager's preview mode or GA4's DebugView and check that events actually come through. “We set it up, it must be working” is a sentence we hear often, and it sometimes surfaces months later as empty reports.

Key events: forms, search, WhatsApp

In GA4 every action is an event. You mark the valuable ones as key events (they used to be called conversions). Those are the ones highlighted in reports and used for ad optimization.

For form submissions, the most reliable method is firing an event when the form is successfully submitted. There are two ways. If the form redirects to a separate thank-you page, you can turn that page view into an event. If the form says “message received” on the same page, your developer needs to send an event such as generate_lead at that moment. Measuring button clicks is misleading, because forms filled in incorrectly and never sent get counted too.

For phone and WhatsApp, the logic is simpler. Phone links start with tel:, WhatsApp links usually with wa.me. In Tag Manager, set up two triggers that catch clicks on these links and give each its own event name, such as phone_click and whatsapp_click. That also shows which channel visitors prefer.

One caveat: a click isn't a conversation. Not everyone who taps the phone calls, and not everyone who opens WhatsApp writes. Compare these numbers with your real inquiry count once a month; a big gap may mean the button's position or text sets the wrong expectation.

Site search is captured automatically on most sites if enhanced measurement is on. Look at search terms once a month. What people search for on your site and can't find is exactly the next page you should write.

Reading traffic sources

Under Acquisition in the Reports menu, you see where visitors come from by channel: Organic Search, Direct, Organic Social, Referral, Paid Search and others.

The most misread channel here is Direct. Besides people typing your address into the browser, visits whose source can't be detected also land here. A link shared on WhatsApp or a page opened inside the Instagram app often looks like direct traffic.

The fix is UTM parameters. If you add something like ?utm_source=instagram&utm_medium=social to the link in your Instagram bio, those visits get credited correctly. You can do the same for the QR code on your business card, your email signature and the website link on your Google Business Profile.

When reading the report, ask the real question per source: not which channel brings the most visits, but which brings the most key events. A channel with few visits but a high form rate is often the one most worth budgeting for.

Cookie consent and data loss

GA4 uses cookies. The approach in the Turkish Personal Data Protection Authority's guidance on cookies is generally to get explicit consent from users for analytics cookies; confirm the setup on your own site with an expert. If your site has a properly configured cookie banner, visitors who click “reject” either don't appear in GA4 or appear in a very limited way.

That means the numbers in your reports will be lower than your real visitor count. How much lower depends on the banner's design and your audience. In practice, read GA4 as a trend tool rather than an exact counter. Questions like whether form submissions rose compared to last month, which source is growing and which page people abandon still get useful answers.

Google's Consent Mode sends cookieless signals for users who decline and partially models the missing data. It must be set up in line with your banner tool; misconfigured, it can lead to cookies being written before consent, which is both a legal and a technical problem. We covered cookie notices and KVKK in a separate post.

In the first month after setup, you don't need to look at reports much. While data builds up, checking once a week in DebugView or the realtime report that events are arriving is enough. At the end of the month, look for the answer to one question: How many inquiries did our site bring this month, and where did they come from?