Measuring with Google Analytics 4

Google Analytics 4 is for recording which job happened, not only how many people saw the site. Sessions, screens and active users are diagnostic lines. The decision line is the event you define: a quote form, a purchase, an appointment request, a call that was actually clicked. If that split is never written down, the dashboard looks busy and sales says nobody is writing.

The ad account has to speak the same language. The setup order on the Google Ads side is in what Google Ads is and how to set up a campaign. How budget is discussed against that measurement is in how to plan a Google Ads budget. This note is about what the event is named on the site, and what is not counted.

Pick one key event

Making everything a conversion makes nothing a conversion. The primary event is the job the business is chasing this month. A qualified form on a corporate site, a purchase in a store, an appointment request at a clinic, directions plus a call for a local business. Secondary events — a download, a video view, a product view — show where the journey breaks. They do not, alone, enter bidding or the sentence “this month went well”.

Questions while you name the event

• What does the team do when this action happens? If nobody looks, it is noise.
• If the same person submits twice, is that two jobs? Usually no.
• If the thank-you address is public, does opening it count? It should not.
• Is a phone click the same as an answered call? No.
• Does the event parameter hold raw personal data? It should not. The privacy notice comes before the tag.

Keep names plain. generate_lead, purchase and phone_click stay readable for the team and for ad imports. A name like “button 3” tells nobody anything three months later.

Do not open the report before a test

Submit a real form before launch. The event should appear once in GA4 realtime. If it does not, the trigger is usually the problem: a consent banner blocked everything, the thank-you state is not a separate address, checkout is on another domain, or the tag manager container is wrong. If it appears twice, you have double counting. The rate inflates and the campaign that actually worked stays hidden.

A conversion imported into Google Ads and a separate on-site ads tag should not both be primary for the same action. One source decides. The other is observation. Note the day a source changes. Do not stack the old chart on the new one and say “we improved”. When the definition changes, the comparison changes.

SEO, ads and the field point at the same event

A rise in organic traffic is not, by itself, a rise in demand. A form from organic and a form from ads should be the same event, split by channel. Then “SEO did nothing” or “ads are expensive” becomes a table, not a mood. On the technical side, a page that fails to open, a redirect chain or a slow checkout never fires the event. That layer can be discussed in the same week as a technical SEO audit. Speed is one job. The tag is another. Both are checked before anyone says the site is live.

In ecommerce, a transaction id should stop the same order being written twice. The value is the order total, not an invented average. In lead generation, write an approximate value only if you roughly know it. Do not invent revenue. A CRM status of “sold” should flow back to ads only when that status is trustworthy. An early, messy status trains bidding on the wrong example.

Rows that should not decide

• Total users are not a success sentence on their own.
• Do not read GA4 engagement with the old bounce-rate habit. The definition changed.
• Do not put “ten seconds on site” in the same bucket as a lead.
• Do not call one campaign day a trend. A short window is noise.
• Do not judge a channel before classifying the leads sales called empty.

Which number is allowed to decide is framed in which digital marketing metric should decide. Yüz Medya opens Google Ads and site measurement on the same event list. If the tag is suspect, the budget does not grow. If consent or redirects break the tag, it goes on the technical SEO list.

FAQ

If GA4 is installed, is measurement done?

No. Installation means the tag is on the page. Measurement means the right event fires once, at the right moment. An untested tag counts as not set up.

Can I compare it with old Universal Analytics reports?

No. Session and conversion definitions are not the same. The new period is read with its own definition.

What if the consent banner blocks everything?

Without consent, measurement stays limited. Forcing “accept” is not the fix. Test that the event works in a consented session. Do not hide the loss in the report.

Can Meta and Google see the same event?

Yes, as long as they do not count the same job under different names. A channel comparison only means something when the key event is shared. We come back to that split in Google Ads or Meta Ads.