Which report tells you where visitors came from?
Use Traffic acquisition to ask where a visit came from. Use User acquisition to ask how a person was first recorded as finding you. The same return visit can fall into a different group in each report.
Suppose someone discovers your booking app through a search, then returns the next day from your launch email. Add campaign labels to the email link so you can look for that campaign. Read the session source in Traffic acquisition to compare visits. Read the first user source when you want to know how new people found the app.
Choose the report before comparing numbers. Session source, first user source and credit for a key event answer different questions. A visit can also go unmeasured if the tag does not load or the chosen collection settings block it.
How do you verify one tagged visit and one real success?
Create a GA4 property for the test site, add a web data stream and use the Google tag from that stream. A property is a reporting workspace. The stream connects the test site to it.
- Check the property and the pagesUse a separate property and site that contain only fake records. Leave advertising connections off for the test. Add the Google tag through the builder’s integration or in the site code once. Check the reporting time zone and review the automatic events before publishing the test page.
- Open one tagged landing pageUse a fake campaign name in the link and open the page once. Check Realtime or DebugView to see that the test property received a page view. Look at the page address and campaign labels for values you did not mean to send.
- Check what the app actually savedFor a successful fake signup, send a signup event only after the account exists. Try a failed signup and reload the confirmation page. Verify they do not add successful signups. Once the event is correct, mark it as a key event if you need to report on it.
When does an event become useful for a decision?
Mark the action you actually want to measure as a key event. For the booking app, that could be a saved signup or a completed booking. A click on the form button measures an attempt.
For a campaign report, decide if you want to count every completed booking or count sessions in which someone booked. For a product question, Explore can turn a sequence such as account created → first booking page published into a funnel. Define who can enter it and how much time each step gets.
If you advertise with Google Ads, creating or importing a conversion for ad reporting or bidding is a separate step that uses the linked account and settings. Confirm the definition in both places before comparing the totals.
What does GA4 know about a person?
By default, the web tag can use a browser identifier to group activity. With a configured User-ID, GA4 can associate activity from a signed-in account. Your app has to supply a stable ID and clear it at sign-out.
| Area | What you can change |
|---|---|
| Page addresses and titles | Keep emails, names and access links out before the tag sends them. Built-in redaction can help with some query parameters, and does not replace this check. |
| Signed-in identity | Use a stable value that follows Google’s User-ID rules. Use two test accounts on one browser and verify the second account does not inherit the first one’s ID. |
| Choice about measurement | Connect the consent choice to the tags. Google’s basic mode blocks the tags when there is no consent. Advanced mode can send limited signals without analytics cookies. Check the actual behavior of the mode you deploy. |
When can Google Analytics add costs or limits?
Google offers Standard GA4 without a subscription charge. Analytics 360 is a separate enterprise offering with higher limits. BigQuery is a separate service that stores and queries data. It can charge for storage and processing even if the Analytics property is Standard.
| What you are using | What to check |
|---|---|
| Detailed exploration | Google lists two or fourteen months for relevant user and event detail. Very large Standard properties can be limited to two months. Standard aggregated reports are subject to different rules. |
| Daily raw event export | Standard has a one-million-event daily batch export limit. This limit applies to the export. |
| Streaming raw events | BigQuery can charge for streaming. The streaming export is best effort and omits some new-user and new-session source information. Use the daily export where that information matters. |
What should you keep when you move the reports?
Save the reports you need and, if the underlying events are necessary, set up a BigQuery export early. Keep a separate record of event definitions, campaign labels, report filters and consent settings.
A report downloaded as a spreadsheet contains results for the selected report. A BigQuery export contains individual event rows. Modeled values, traffic attribution and reports can differ from what you see in the Analytics interface. Run one familiar report from your export before counting on it.
Check that the business controls the property and the export destination. If the new tool counts browsers, visits or signed-in accounts differently, record the switch date and carry the definitions forward so people can interpret the change.