You, or someone working for you, have added a new tag in Google Tag Manager: a WhatsApp click event, a Google Ads conversion, the Meta Pixel. It looks right in the settings. The temptation is to press Submit and move on. A week later the reports show nothing, or double, and nobody can say why.
Google Tag Manager preview mode exists to stop that. It lets you browse your own site as a test visitor while showing exactly which tags fired, which did not, and why. This post walks through the preview and debug screen, a routine for testing each trigger, and how to publish so changes can be undone. If containers, tags and triggers are new to you, start with our plain-language guide to Tag Manager.
Starting Google Tag Manager preview mode
- In your GTM workspace, click Preview near the top right.
- A Tag Assistant window opens and asks for your website address. Enter the page you want to test, usually the home page or the page with the form.
- Your site opens in a new window with a small badge showing that Tag Assistant is connected. Keep both windows open.
- Use the site normally in the new window. Every action is logged in the Tag Assistant window.
Only your browser sees the unpublished changes. Real visitors continue to get the live version of the container. That is what makes preview safe to use on a working site.
If the connection fails, the usual suspects are an ad blocker or privacy extension, a strict browser privacy setting, a cookie banner blocking Tag Manager itself, or a cached page that does not contain the container snippet. Try a clean browser profile with no extensions first.
Reading the Tag Assistant screen
The screen has two parts.
The timeline on the left
Each page you load appears as a group, and inside it is a list of events in the order they happened. On a typical page load you will see stages such as consent initialisation, initialisation, container loaded, DOM ready and window loaded. As you click links or submit forms, new entries appear: clicks, form events, and any custom events your site or plugins push into the data layer.
Clicking an entry shows the state of things at that exact moment. This matters. A tag that should fire on a click will only show as fired under that click, not under the page load.
The detail tabs on the right
- Tags: split into tags fired and tags not fired for the selected event. Click any tag to see its settings and, at the bottom, its triggers with a tick or cross against each condition.
- Variables: the value of every variable at that moment, such as the page URL, the click URL or the click text. This is where you find out why a trigger did not match.
- Data layer: what had been pushed into the data layer, which is where form plugins and ecommerce plugins put their details.
- Consent: the default and updated consent state, if consent mode is set up. Useful for checking your cookie banner.
- Errors: any tags that failed to run properly.
Why a tag did not fire, and how to tell
When a tag sits under “not fired”, click it and scroll to its firing triggers. Each condition shows whether it matched. Common findings:
| What you see | Likely reason | What to check |
|---|---|---|
| Click URL condition has a cross | The link address is different from what the trigger expects | Variables tab: look at Click URL for that click |
| No click event appears at all | Click variables or a click trigger are not set up, or the button is not a real link | Whether built-in click variables are enabled |
| Trigger matched but tag shows as blocked | A consent requirement or blocking trigger stopped it | Consent tab and the tag’s consent settings |
| Tag fired on page load instead of on the action | Trigger set to all pages or the wrong event type | The trigger type in GTM |
| Custom event never appears | The form plugin’s success event is not being pushed | Data layer tab after a real submission |
A routine to test tags before publishing
Run through the same checklist for each change, on desktop and then on a phone.
- Page view tags. Load two or three pages. Your GA4 configuration tag, or Google tag, should fire once per page, not twice.
- Click tags. Tap the phone number, the WhatsApp button and any tracked buttons in the header, footer and floating bar. Each should produce one firing of the right tag, under that click.
- Form tags. Submit a form with a missing field first. Your lead tag must not fire. Then submit it properly and confirm it fires once. Our post on reliable form submission tracking explains which signals to use.
- Conversion tags. Confirm Google Ads and Meta tags fire on the same real actions, and only those.
- Consent. Repeat a few steps after rejecting cookies in the banner. Tags that need consent should be blocked or limited as intended.
- Check the receiving end. While preview is running, GA4’s DebugView shows events from your test session, with their parameters. Confirm names and values arrive as expected.
Testing on a phone matters because floating buttons, menus and forms often behave differently on small screens. Tag Assistant has an option to share a preview, so the same test session can be opened on another device.
Publishing with versions you can roll back
When the tests pass, close preview and click Submit in the workspace.
- Choose to publish and create a version.
- Give the version a name that says what changed, such as “WhatsApp click event added”.
- Write a one-line description: what, why, and who made it.
- Publish.
Every published version stays in the Versions list. If a later change causes trouble, pick the last good version from that list and publish it, which restores the container to that state without touching the website.
A few habits keep this tidy. Publish one change at a time where you can, so a problem is easy to trace. Do not leave half-finished work sitting in the default workspace, because the next person to publish will push it live with their change. And check the published version again with preview after publishing, since caching on the site can delay what visitors see.
Who should be allowed to publish
Publishing in Tag Manager means changing what code runs on your website. Keep publish permission to a small number of people you trust, give others edit or read access, and remove former staff and old agencies. A short note on each version, as above, makes it clear who changed what.
For how testing tags fits into wider measurement, from key events to monthly reviews, see our guide to marketing analytics. When tags are firing but numbers look doubled, our post on duplicate conversions picks up from here.
Frequently asked questions
Does preview mode send data to my real GA4 reports?
Events from a preview session do reach the GA4 property, marked as debug traffic, and appear in DebugView. They can also show in standard reports unless you exclude developer traffic with a data filter. A handful of test events rarely matters, but heavy testing on a low-traffic site is worth filtering.
Can someone without GTM access check my tags?
They can use the free Tag Assistant to see which Google tags load on your public site, but they cannot see unpublished changes or trigger details. For a full review, give them read access to the container, which lets them preview without being able to publish anything.
Why does my site look different while in preview?
Preview loads the site in a fresh window, often with your normal cookies, so a cookie banner, login state or cached page may differ from what you expect. If something looks odd, test again in a private window. The look of the site itself is not changed by preview mode.
How do I test a tag that should fire on the checkout or payment page?
Place a real test order, using a test payment method if your gateway offers one, or a small real order you then refund. Watch the purchase or conversion event in preview and DebugView. If the gateway sends shoppers off-site, check the event fires when they return to your confirmation page.
Preview mode turns tag changes from guesswork into something you can verify, but it takes practice to read. If you would rather someone else build and test your tags, our Tag Manager and tracking setup includes a documented test for every tag. More on how measurement supports other channels is in our small business digital marketing guide.
Next in this series: Duplicate Conversions: Why Leads Get Counted Twice
