How to set up GA4 correctly

By Adrian Diaz

Define what matters

List the decisions the business wants to make and the actions that support them. A lead site may care about completed enquiries, while a store may focus on purchases and product journeys. Write down the event name, when it should fire, and who uses the report. This prevents a long list of events that nobody can interpret.

Create and connect the property

Create or identify the correct Analytics account, property, and web data stream. Confirm that the business owns the account and that administrators have appropriate access. Install the measurement tag through the site’s supported integration or tag manager, avoiding duplicate installations. Record the property and stream details in the handoff documentation.

Review consent and privacy

Decide how measurement should behave according to the site’s consent choices and applicable requirements. Avoid sending names, email addresses, or other personal information in event parameters or URLs. Review forms and query strings for unintended data. The appropriate configuration depends on the site and jurisdiction, so involve the person responsible for privacy when needed.

Configure meaningful events

Use automatically collected or recommended events where they fit, and add custom events only when they answer a real question. Define important actions consistently and mark the appropriate events as key events in the property. Keep names and parameters documented so that developers, marketers, and the site owner can interpret reports in the same way.

Test before relying on reports

Visit the site in a clean session and exercise each event, including the successful completion of a form or test purchase. Use available debugging and real-time tools to check event names and parameters. Watch for duplicate events, missing values, and unexpected firing on page load. Reporting can be delayed, so distinguish a delay from an actual setup problem.

Maintain the measurement plan

After launch, compare the implementation with the documented plan whenever a form, checkout, or integration changes. Give the owner a simple list of reports and access details. Periodically review whether the events remain useful and remove obsolete configuration carefully. Treat analytics as an ongoing measurement system rather than a one-time installation.

Keep a measurement record

A useful measurement plan connects each event to a business question. For every important action, document the event name, the page or interaction that triggers it, the parameters expected, and the way to test it. Note whether the event represents an attempt or a confirmed success. For example, clicking a submit button does not necessarily mean a form was accepted.

Use the same plan during handoff and future site changes. If a form provider, checkout, or consent tool changes, repeat the relevant tests. Record the test date and outcome. Avoid promising that every visit will be counted: consent choices, browser behavior, and configuration can affect what analytics receives.

Example: measuring an enquiry

Start by defining the event in business terms. An enquiry should normally represent a successful submission, not merely a click on the submit button. A click can be followed by validation errors, a network failure, or an abandoned form. Agree with the site owner on the condition that represents success and document where that condition can be observed.

Use a clearly named event and document any parameters that are genuinely needed. Do not include the person’s name, email address, message, or another identifying value. Check that page URLs and query strings do not accidentally contain form data. If the implementation uses a tag manager or a form integration, verify that only one path sends the event so a single enquiry is not counted twice.

Test an invalid submission, a valid test submission, and a page refresh after success. The success event should reflect the agreed behavior rather than firing on every attempt or reload. Inspect the event in the available debugging tools, then allow for normal reporting delays before checking reports. Record the test date, consent state, and result in the measurement plan.

Avoid misleading comparisons

Before comparing periods, check whether the tag, consent configuration, event definitions, or site structure changed. A rise or fall in a report can reflect a measurement change rather than a change in customer behavior. Keep a dated record of important configuration changes so a future reviewer has context.

Give access only to the people who need it and check that the business retains administrative ownership. Document the reports the owner uses and what each key event means. When the site changes, repeat the relevant tests and update the plan rather than relying on old event names.

Put the guidance into practice

Start with the page or workflow that matters most to the people using the site. Write down the current behavior, the change you intend to make, and the person responsible for reviewing it. If the site has a staging environment, test the change there before applying it publicly. Keep a record of the previous configuration and make sure the team knows how to restore it if the change causes a problem.

After the change, repeat the original task with realistic content. Check it on a small screen and with a keyboard, and include any form, payment, or integration that the change could affect. Record the date, result, and any follow-up. Share the outcome with the site owner in terms of what visitors can now do and what remains uncertain. Revisit the check when the underlying content, theme, or connected tools change.

Explore related services.

Need a hand?

Discuss your project