How to Measure Website Performance with Google Analytics 4

Google Analytics 4 is the tool that describes what a site does, in events. The old habit stops at “how many people came”. The more useful business question is: did they send the form, tap the phone, put a product in the cart and finish payment, and which source did that movement come from? GA4 can answer that if it is set up properly. Set up badly, it gives you a swollen session chart.

This article does not tour every menu. It builds the skeleton you need in order to decide. The skeleton has three parts: what counts as an event, which of those become conversions, and how source is read. If the ad account does not mean the same three things, Monday’s meeting splits into two truths.

Define the event first

An event is a specific movement on the site. A page view may arrive on its own. A form submit, a phone tap, a file download, add to cart and purchase have to be connected on purpose. Give each movement its own readable name. “event_1” tells nobody anything three months later. The name should be a word sales already uses: quote_form, booking, purchase.

Double counting starts here. If the same form fires both when the page loads and when it is submitted, the conversion doubles. Counting again when the thank-you page is refreshed is the same mistake. The test is a real submission. One fill, in a private window, shows once in the report. If it does not show, or it shows twice, launch is not finished. Yuz Medya runs that rehearsal without adding another tool to a pile of tags. An old unused tag breaks both speed and the count.

A conversion is chosen from the events

Not every event is a conversion. A page view, a scroll and “stayed 10 seconds” are not success for most businesses. A conversion is the outcome the business accepts. A form or a qualified call for a service. A purchase for a shop. Mid-steps — add to cart, checkout start — do not have to be marked as conversions. If they are, the “conversion count” swells and profit cannot be discussed.

Write the value if you have it. A purchase goes with an amount and a currency. If you roughly know what a form is worth, that can be noted too. If you do not know, do not invent it. An invented value teaches the ad system the wrong goal. If Google Ads will import this conversion, the name and the meaning have to match on both sides. Read the setup next to the measurement section in what Google Ads is. The Meta counterpart is in the pixel and the Conversions API.

Source: where the money went

• Organic search: a visit found while ads are off. Brand search and generic search are not read in the same bucket.
• Paid: Google, Meta or another campaign. It should move in roughly the same direction as the ad panel, not to the penny.
• Direct: swells when the source tag falls off. Treating everything as direct is a hole in measurement.
• Referral: a visit from another site. If your own payment page looks like a referral, the channel is broken.

If campaign links are not tagged consistently, paid traffic shows up as organic or direct. Then the verdict “SEO is great, ads are wasted” is really a bad tag. Write the tagging standard on one page. Every new campaign follows it. If it does not, the report speaks a new language every time the agency changes.

How to read the report

A weekly look needs three numbers: how many qualified conversions, from which source, from which page. A rise in sessions is not a celebration on its own. If conversions fall while sessions rise, you have the wrong audience or a weak page. In the page report, the address where the form sits and the address the ad lands on should be the same. If they are not, design and traffic are arguing. The design side of that argument is in conversion-focused design.

Compare only while the definition stays the same. Note the day you add a new event. Do not say “up by this much” against the period before that day. The rise may be the definition getting wider. An SEO report is also read beside the organic row in GA4 and beside the search engine’s own query report. One shows the visit. The other shows the query. They are not the same sentence. In a monthly note we write that split in the open. One chart hides two different questions.

Privacy is part of the skeleton too. Analytics is not installed to collect personal data you do not need. Pushing health or identity details from a form into an event parameter grows risk, not measurement. Check the current setting of the tool you use, and the privacy notice, on IP and identity. The reflex to “measure everything” leaves a pile you cannot explain in that notice.

Frequently asked questions

If GA4 is installed, is measurement done?

No. Installation is not the code sitting on the page. It is the event landing once, at the right moment, under the right name. Saying “we have analytics” before that test is like hanging a clock on a shop that has no till.

Why don’t the ad panel and GA4 match?

A click and a session are not the same thing. One is the click the ad platform counted. The other is the visit that formed on the site. A gap is normal. A canyon means the tag, the consent banner or a redirect is broken. Do not wave the canyon away as “the model”. Follow one campaign from end to end, with one click.

Does a small site need GA4?

If there is a form or a sale, yes, installed simply. If it is only a card-sized page and there are no ads, a heavy setup is not required. Seeing the phone tap still separates “nobody calls” from the data. The tool can be as plain as the site’s promise. It should not be heavier than the promise.