After a new tag, a consent banner or a template change, the way to know analytics still counts correctly is to send traffic you can count yourself. Run 200 real browser sessions to a test page in one hour, then check that GA4, your server log and your CDN report the same visits.
The sessions come from automated browsers. They test how your analytics counts visits; they are not an audience.
Broken tracking rarely looks broken. A tag that fires twice doubles page views, a consent banner that holds the tag back drops sessions, and a single-page app that sends no page_view on route changes hides whole sections of the site. Each of these still produces a plausible number.
The gap shows only when you know what the number should be. A batch whose session count, page count and source you fixed before it ran gives you that figure.
A new GA4 property, a move to Google Tag Manager, a redesign that replaced the page templates, a new consent banner, or renamed referral sources that should land in a different default channel group.
Each check is a small campaign pointed at the pages the change touched, scheduled in a quiet hour so the batch stands out from your regular visitors.
Traffic Generator sends direct sessions, or referral and social sessions with a referrer you name, so you can see that each lands in the channel you expect. To test the referral path through a real page, Link Clicker opens a page that links to you and clicks the link itself.
Organic search is a separate check: GA4 does not show the search term of organic visits at all, and visits that come from a results page are SERP Clicks, not Traffic Generator.
A site has moved to new page templates and wants to confirm that GA4 still counts sessions, page views and channels. The batch runs against a test page that regular visitors do not open, so it can be told apart in every report.
| Target URL | https://example.com/analytics-test/, a page regular visitors do not open |
|---|---|
| Volume and schedule | 200 sessions, all in one hour of a quiet day |
| Traffic source | Direct |
| Session duration | 30 to 90 seconds |
| Page depth | 3 pages, following links from the test page |
| Bounce rate | 0 percent, so every session opens all 3 pages |
| Country | The country most of your audience comes from |
| Referral check | Link Clicker, 50 clicks on one backlink on a site you control |
Cost of the whole test: 200 sessions at $0.0015 come to $0.30, and 50 backlink clicks at $0.003 come to $0.15, so $0.45 in all, well inside the $5 minimum deposit.
Before the batch starts, note the sessions, page views and channel each report should show. A test without an expected figure is only a look at a dashboard.
Look at the data filters on the GA4 property and at how the consent banner loads the tag. If the tag waits for consent, sessions that never accept the banner are not recorded the way accepted ones are, and the test shows you that gap.
Schedule the whole batch inside one hour of a quiet day, so it is easy to isolate in Realtime, in the landing page report and in the log.
Compare GA4, the server log and the CDN for the same hour. When they agree, the tag counts what reaches the site; when GA4 alone falls short, look at the tag, the consent setup or a filter.
Acceptable use, data handling and the limits that apply to every tool are in the Trust Center
The sessions come from Traffic Generator, billed per session at a rate that follows the session duration, and the backlink check uses Link Clicker, billed per click. Both rates are on the pricing page.
In a working setup, yes: they are real browsers that load the page and run its JavaScript, so the GA4 tag fires as it would for any visitor. That is what the test checks. If GA4 shows fewer sessions than your server log for the same hour, something in the tag, the consent setup or a filter is dropping them.
For the hour of the test, yes, because the sessions are counted. Run it against a staging copy with its own GA4 property, or against a test page, and keep a note of the date and hour so the batch is not read as demand later.
Filter the access log by the test path and the hour of the run. On a standard nginx or Apache log, grep '/analytics-test/' access.log | wc -l gives the number of requests to the test pages; narrow it to the hour by adding the timestamp the log uses.
Two ways. Traffic Generator can send referral sessions with the referrer of a page you name, which tests how GA4 groups that source. Link Clicker goes through a real page that links to you and clicks the link, which tests the whole path, including what the linking page's referrer policy lets through.
Write down the figures you expect, send 200 sessions in one hour and compare GA4 with your own log