[Mathew]

Sure, I can help you with that. Although the Google Tag Manager (GTM) is one common cause of this issue, it could potentially be caused by several other factors as well. It’s possible that there may be some sort of caching that’s causing the event to fire twice, especially if you’re using a caching plugin on your site. I would try clearing any caching plugins that you’re using and see if the issue persists after that. You also could be having an issue with the theme or another plugin causing a conflict. To isolate this, you could try disabling other plugins briefly to see if the double ‘begin_checkout’ events stop firing.

It’s also important to take into consideration that the way in which Google Analytics events are set up may cause such an issue. For instance, they might be triggered both when the checkout button visibility changes, and when it is clicked. This may require having a developer take a look at the JavaScript related to the event triggering.

I would recommend sharing these possible scenarios with your client for them to investigate further and potentially consult with a developer if needed. If nothing else works, it’s also possible to use the Measurement Protocol API to manually remove duplicates, although this would be a more technical and complex approach.

[Mathew]

In Looker Studio, you should be able to create separate measures for each of your events, then use a table calculation to find the ratio between them. Your issue with the calculated field might be because you’re using a dimension in an aggregate function, which Looker doesn’t allow. So, create two separate measures – one for ‘application’ events and one for ‘view’ events, and then use a table calculation to divide the ‘application’ count by the ‘view’ count.

[Mathew]

So the main issue could be trouble in tracking GCLID. If MP isn’t properly handling the GCLID, Google Ads won’t be able to map the conversions you’re trying to input.

A couple ways to debug this could be to look at the GCLID in the network conversion hits in MP and the front-end. If there’s any discrepancy or truncation in the GCLID, it could cause the issue. Another way would be to take your data from GA4 and compare it to BQ. Looking at two different conversion hits made via the front-end and MP could give you some insight.

If you haven’t already, try checking out these avenues. If the GCLID does happen to be malformed in MP, that should tell you wherein lies the problem. And while it could also be another issue like client ids, session ids, etc., those should have been obvious in GA4 or when exploring user sessions in BQ. Hopefully this helps get you on the right path!