# Check your conversion tracking

> Audit Google Tag Manager, confirm which GA4 events count as conversions, and watch a test visit arrive in realtime, with your AI.

Every marketing report depends on tracking that works. This guide checks it from end to end: the tags in Tag Manager, the key events GA4 counts, the conversions Google Ads records, and a live test. Run it when you take on a new site, after a site change, or when conversions suddenly drop.

Plugsight only reads. Your AI finds problems and tells you what to fix; you change Tag Manager, GA4 or Google Ads yourself.

## What you need

- A **Pro** or **Agency** plan.
- A [Google Analytics 4](https://docs.plugsight.app/connectors/google-analytics-4.md) connection that reaches the property.
- Optional: a [Google Tag Manager](https://docs.plugsight.app/connectors/google-tag-manager.md) connection. Google is still reviewing its permission, so it may not appear in your **Connector catalog** yet. Skip Step 1 if you can't connect it.
- Optional: a [Google Ads](https://docs.plugsight.app/connectors/google-ads.md) connection, for Step 4.

## Step 1: audit the Tag Manager container

1. Ask "List my Tag Manager containers" and note the GTM id (GTM-…) of the site's container.
2. Ask "Audit container GTM-XXXXXXX and tell me what to fix first."

The audit checks the live (published) version for:

- tags with no firing trigger, which never run
- Universal Analytics tags, which collect nothing since Google retired UA
- paused tags
- possible duplicates that can double-count
- triggers and variables nothing uses
- Custom HTML tags, to check they're still needed

It also lists changes waiting unpublished in workspaces. A conversion tag fixed in a workspace but never published still isn't live.

Follow-up questions:

- "Which GA4 measurement id does the GA4 config tag send to?" Check it matches the GA4 property you report on.
- "Are any of our conversion tags firing twice?"

## Step 2: check what GA4 counts as a conversion

- "Which events does this GA4 property count as key events, and how many of each happened in the last 7 days?"

Look for key events with zero counts (tracking broken, or the event was renamed) and for events that shouldn't be key events at all (such as page views).

## Step 3: check the funnel

- "Build a funnel from viewing a product page to adding to cart to purchase for the last 30 days. Where do users drop off? Split it by device."

A funnel has 2 to 10 steps, each an event or a page visit. A step where almost everyone drops off, on one device only, often points to a tracking problem rather than a real one.

## Step 4: compare with Google Ads

- "List the conversion actions in this Google Ads account with conversions recorded in the last 30 days."
- "Compare Google Ads conversions last week with GA4 key events from Google Ads traffic for the same days."

The two won't match exactly (they count differently), but a large or sudden gap is worth checking.

## Step 5: a live test

1. Ask: "How many people are on the site right now, by page?" This reads GA4 activity from the last 30 minutes.
2. Visit the site yourself and complete a test action, for example a form submission.
3. A minute later, ask: "Did a generate_lead event (or your event name) happen in the last 30 minutes?"

## Step 6: write it up

- "Summarise the tracking check as a table: problem, where (GTM, GA4 or Google Ads), evidence, and the fix. Highest impact first."

## Related articles

- [How your AI analyses marketing data](https://docs.plugsight.app/connectors/marketing-data.md)
- [Run a Google Ads account audit](https://docs.plugsight.app/guides/google-ads-audit.md)
- [Get accurate numbers from marketing data](https://docs.plugsight.app/guides/accurate-marketing-numbers.md)

---

Web version: https://docs.plugsight.app/guides/tracking-check
