In this article
Use Cloudflare Web Analytics for a straightforward view of website traffic and performance; evaluate GA4 when you need defined events, acquisition analysis and conversion measurement. Choose around your questions, not a requirement that two dashboards show identical totals.
This comparison is based on official documentation. We have not run a controlled benchmark of tracking accuracy or installed either tool on your behalf.
Define what you want to know#
“How much traffic do I have?” is incomplete. A useful measurement plan separates questions:
- Which public articles receive page views?
- Which countries and referral sources appear?
- How is page performance changing?
- How many readers submit a form or finish a sign-up?
- Which campaigns lead to those actions?
Write these questions before choosing a tool. Otherwise you may spend time collecting numbers that never change a decision.
Understand Cloudflare's measurement layer#
Cloudflare Web Analytics provides website analytics through browser instrumentation. Its setup guide supports automatic installation for eligible proxied sites and manual installation where needed.
Do not confuse this with every request seen by Cloudflare's network. Edge requests, page views and people are different units. A crawler requesting resources is not automatically an engaged reader.
Follow the current installation instructions and check whether the script is actually present on the pages you intend to measure.
Use GA4 for a defined event model#
Google's Analytics developer documentation describes website measurement and additional event collection. It requires an account, property and data stream before integration.
Define an event precisely. For example, a click on “Request access” is not the same as a successfully submitted request. Naming both “conversion” without distinguishing them makes the report difficult to interpret.
For a form, decide which confirmed outcome matters, how duplicate submissions are handled and which non-sensitive attributes are useful. Do not send credentials or personal form contents just because an analytics event can contain parameters.
Expect differences between dashboards#
Different tools can use different definitions, filters and collection behavior. Browser blockers, consent choices and incomplete script loading can also affect observations.
When totals disagree, compare:
- The same date range and timezone.
- The same hostnames and public pages.
- Page views versus visits, users or requests.
- Filters for internal traffic and automated requests.
- Installation across navigation paths.
Do not “fix” a discrepancy by multiplying one report until it matches another. First establish what each number represents.
Test a simple journey#
Use a controlled sequence: open an article, navigate to another page and complete a designated test action if one exists. Inspect the supported diagnostic views and allow for reporting delay.
Check for duplicate scripts if an automatic integration and manual snippet might both be active. Test client-side navigation as well as an initial page load.
Record the expected events and what you actually observed. This establishes the tested journey, not completeness for every visitor.
A practical choice for a small site#
Start with the minimum tool that answers the current decision. Add event measurement when you have clear outcomes to track and a suitable privacy setup.
Evaluate consent and data-handling requirements for the actual implementation rather than assuming every analytics product operates the same way.
If the website also needs business email, our custom-domain email guide explains another common service-selection decision. For infrastructure spending, read why AWS budget alerts are not a spending cap.