Case study: what became visible at drogerieshop24 after the switch

What was the starting point?

The store used classic browser tracking. As with most stores, there was a gap between the orders in the Shopware back end and the purchases that arrived in Google Analytics 4. Visitors using Safari and ad blockers in particular were only partially captured.

What was implemented?

The switchover covered the data layer in Shopware, a server container at TAGGRS under its own subdomain, the server-side connection of GA4 and the ad platforms, and a review of all e-commerce events. The old tracking was switched off in the same step.

What became visible afterwards?

The days before the switch (1 to 19 November 2024) were compared with the days after (20 November to 6 December 2024). An honest caveat up front: the second period includes Black Week and Cyber Monday. Part of every increase is therefore seasonal, not tracking. What is meaningful are the ratios, not the absolute jumps.

• Safari catches up with Chrome: Safari users measured in GA4 rose from 749 to 1,525, Chrome users from 1,399 to 2,001. Both browsers experienced the same season. That Safari grew much more strongly than Chrome is the clearest indication that Safari users in particular were missing from measurement before.

• Purchases are captured more completely: Measured purchase events rose from 91 to 151. How much of that is seasonal only becomes clear when compared with the actual orders in the Shopware back end for the same periods. That comparison is exactly the benchmark Chapter 10 recommends.

• New events become visible in the first place: Form starts rose from 57 to 775 users, form submissions from 39 to 563. Jumps of this size are, in our experience, not caused by more visitors but by events that previously did not arrive at all or only in fragments. That is not growth but new visibility.

What can be concluded?

In this store, server-side tracking did not increase revenue; it increased visibility of it. The ad platforms received more and more complete conversions, especially from Safari users, who had previously been systematically underestimated. Campaigns that reach precisely this audience could only now be properly evaluated. That is the real value: budget decisions based on data that is closer to reality.

What does public data from the platforms show?

A single case study proves little. More robust are the analyses that Google and Meta have published across many ad accounts. They do not measure server-side tracking as a whole, but the building blocks a server container delivers: server interfaces, user data for attribution and loading via your own domain.

• Meta Conversions API: In A/B experiments with advertisers who used the Conversions API in addition to the pixel, cost per result fell by 13 percent on average, according to Meta. In April 2026, Meta reported an average 17.8 percent lower cost per result for advertisers with a Conversions API setup for web events than for those without.

• Google Enhanced Conversions: According to Google, the conversion rate with Enhanced Conversions rises by a median of 5 percent for Search campaigns and 17 percent on YouTube.

• Google Tag Gateway: Advertisers who strengthen their data with the Tag Gateway see 14 percent more conversions on average, according to Google, and more than 20 percent for Demand Gen campaigns.

For context: these are averages from the platforms' own data, not independent studies, and the platforms have an interest in more complete data. But all the figures point in the same direction as the drogerieshop24 example: more and cleaner signals lead to more conversions being counted and budgets being steered better. How much of that reaches your store depends on your starting point, channels and setup.

Can this be transferred to other stores?

The direction, yes; the numbers, no. How large the gap is in a store depends on the Safari share, the consent rate, the product range and the previous setup. The comparison from the checklist in Chapter 1 shows you in a few minutes where your store stands.