The short answer
Test every event in Tag Manager preview, confirm it reaches GA4 in DebugView, then send a real form and book a real call and check they arrive once, in GA4 and in your inbox or calendar. After you publish, check again the next day. Then, once a month, compare GA4 with your real lead count, read the channel report, and look for anything that suddenly changed. One page is enough.
The tools you test with
- Tag Manager preview. Google's preview help says it runs your current draft as if it were published, in your browser only. Click Preview in the Workspace, enter your URL and connect. Tag Assistant then shows which tags fired, in what order and with what data. Regular visitors see nothing. You can share a preview link with a colleague.
- GA4 DebugView. In Admin, under Data display. Google's DebugView help says preview mode or Tag Assistant puts your browser into debug mode. It also says events will not appear if consent mode is on and you have not consented to analytics, so accept the banner first. A blank DebugView often just means you have not accepted.
- GA4 Realtime. The Realtime report shows activity from the last 30 minutes. Useful after publishing, when preview is off.
- Your browser's Network tab. The final word on what actually left the page.
The test script
Run this on the live site or a staging copy, from a tagged test link, with a test email address you can spot later.
| Action | In preview | In GA4 | In your tools |
|---|---|---|---|
| Load a page and reject the banner | Consent denied, tags held as planned | Nothing, or cookieless pings in advanced mode | No analytics cookies |
| Accept the banner | Google tag fires once | One page view in DebugView | Clarity consent shows granted |
| Click each main button | cta_click with text and link | Same event in DebugView | |
| Click the email and phone links | email_click, phone_click | Same events | |
| Send a real form | lead_sent custom event, GA4 tag fires | One generate_lead with form_name | Lead in inbox or CRM with UTM values |
| Book a real call on a test event type | call_booked | One call_booked | Booking in the calendar |
| Reload the thank-you screen, send an empty form | Nothing new | No new lead | No new lead |
Repeat the form and booking steps on a phone. Then publish the container with a note, and check Realtime the same day and the standard reports the next day.
Common mistakes
| Mistake | What it looks like | Fix |
|---|---|---|
| GA4 loaded twice | Page views roughly double, engagement looks low | Remove every copy outside the container |
| Lead counted on a thank-you page view | More leads in GA4 than in the inbox | Fire on the accepted submit, or guard against reloads |
| Event fires in preview but never arrives | Tag shows as fired, DebugView is empty | Check consent, the measurement ID and the destination. On our own site, custom events only arrived once they named their GA4 destination with send_to |
| Staging traffic in reports | Visits from a staging domain | Load tags only on the live domain |
| Personal data in URLs | Emails in page paths or parameters | Keep names and emails out of URLs and redirects |
| Filter switched to Active too early | Real visits missing for good | Leave filters in Testing until checked, since Active is permanent |
| Too many key events | Key events far above real leads | Keep only real outcomes as key events |
| Tracking broke after a redesign | Leads in the inbox, none in GA4 | Rerun the test script after every form or template change |
Most of these show up the same way: the gap between GA4 and the inbox changes. That is why the monthly check below starts with that gap.
The one-page monthly report
Keep it short enough that someone reads it. This is the page we use.
| Line | Source | What to write |
|---|---|---|
| Leads this month | Inbox or CRM | The real count, by form |
| Calls booked | Calendar | The real count, minus cancellations if you track them |
| GA4 coverage | GA4 generate_lead divided by real leads | A percentage. Flag it if it moved a lot |
| Leads by channel | GA4, your custom channel group | Top channels, including AI assistants |
| Top pages before a lead | GA4 exploration or path report | The three pages most leads read |
| Search | Search Console and Bing | Clicks, top queries, any new indexing errors |
| Behaviour | Clarity | One thing you saw in recordings worth fixing |
| Changes | Tag Manager versions, site changes | What changed this month that could move numbers |
| Next action | You | One fix or test for next month |
The Changes line is the one teams skip and later wish they had. A consent change, a new form or a redesigned page can move every number on the sheet. Writing it down the month it happens saves a long argument the month after.
Monthly checklist
- Compare GA4 leads with the real count. Investigate if the gap changed sharply.
- Send one test form and book one test call, then delete them from the CRM.
- Check that every tag in Tag Manager's consent overview is Consent Configured.
- Look for new scripts on the site that are not in the container.
- Check that your internal traffic filter still matches your office IP addresses.
- Review the AI assistants pattern against your source report.
- Check Search Console and Bing for new indexing or sitemap errors.
- Watch five Clarity recordings of your highest traffic page on a phone.
- Update the measurement plan if an event was added, renamed or removed.
For a deeper check once a quarter, run through our GA4 audit.
When to get help
Everything in this guide can be done in house with a few days of care. If you would rather hand it over, our analytics and tracking service starts from $500 and covers GA4, Tag Manager, conversion events and the consent setup, tested the way chapter 6 describes. Care plan clients get tracking upkeep as part of the plan, see the Webflow care plan or the Astro care plan.
We set up and test the technical side. What your site legally needs is for your adviser to decide. Want us to look at your current setup first? Get in touch.


