How FunnelTrack cut maintenance time and increased measured data by 20% for 30+ clients after migrating from Google Cloud to TAGGRS
The overview
FunnelTrack is a data and web tracking agency based in Utrecht in the Netherlands. It was founded by Brandon Teuben and Martijn Udo in 2021 after they met as freelancers and collaborated on a few projects that focused on building the full tracking stacks and data warehouses. The team currently consists of 7 people, all of which handle measurement.
The agency deliberately leaves advertising out of its services. Brandon spent 12 years in digital marketing before starting the company, including 1.5 years at IPG Media Brands in Australia, so he has plenty of experience in running campaigns. He and Martijn decided not to include it in their offer because they wanted every hour in the business to go into measurement work.
That decision came out of what Brandon saw inside marketing agencies. Around 90% of the revenue came from advertising and tracking was available as an upsell on top of it. The upsell mattered to clients, but it also shaped staffing decisions within the agencies. New people were trained as advertising specialists and a tracking specialist could have been brought in when the measurement required attention.
That focus shows up in hiring and in the kind of work that comes in. Data and tracking specialists whose work was sold as an upsell somewhere else are easier to attract when measurement is the whole job. On the client side, companies rarely approach FunnelTrack to set up a single Meta pixel. They usually come with more complex problems, often ones that another agency has already tried to solve.
The challenge
Brandon has worked with server-side implementations for 8 to 10 years, ever since his time at an agency. For most of it, the builds ran through Google Cloud.
Google Cloud did the job. Every implementation was also a custom build that needed ongoing attention, and the problems were mainly small but varied:
- DNS records on the client side dropped out, took the subdomain offline and stopped the measurement.
- Google Cloud Platform applied limits that the setups ran into.
- Processes stalled when FunnelTrack or the client pushed large backfills into the data warehouse.
- Every live client needed a recurring monthly check.
None of these issues were serious on their own. But, when combined across a growing client base, they amounted to a fixed monthly workload that neither side wanted. FunnelTrack did not want to earn its money maintaining infrastructure and clients saw no reason to pay for it.
- Brandon Teuben, Co-founder and Data & Tracking Strategist at FunnelTrack
At the same time, demand was picking up. Eight years ago, Brandon had to sell Server-side Tracking to clients and clearly explain its benefits. Over the past 2 years, the requests started arriving fully formed, with companies specifically saying they wanted Server-side Tracking and asking FunnelTrack to set it up for them. More implementations meant more hours going into maintenance every month.
The solution
FunnelTrack works independently of any specific technology and can build on whatever stack a client already runs. They were introduced to TAGGRS by one of their clients who had created their own account, could not finish the setup alone and asked the agency to take it over.
The first project went well enough that the team kept using the platform for the following reasons:
- TAGGRS is a Dutch company, so it was easy to communicate and stay in direct contact with the team.
- The support team analyzed each underlying problem whenever it occurred instead of replying with a link to a documentation page.
- Once a client setup was live, FunnelTrack didn’t have to actively spend time on it.
What started as a preference is now standard practice. FunnelTrack sets up every new client on TAGGRS’ infrastructure and currently has more than 30 of them live on the platform.
The parallel setup
Server-side Tracking takes over a large part of a company's existing measurement and, in some cases, all of it. Events that used to fire from the browser now depend on a server container. Brandon calls the implementation phase error-prone, stating that any mistakes can break measurements that were working fine before the project started. FunnelTrack follows the same sequence for every client to keep issues to a minimum:
- The client-side setup gets validated first. Server-side measurement relies on the quality of what already runs in the browser, so the existing container is reviewed and optimized before any server-side work begins.
- Server-side runs next to client-side, never on top of it. The new setup gets its own conversion IDs and labels and forwards data to its own destinations. If someone mistypes a DNS record halfway through the project or the subdomain drops offline, the existing measurement continues running and the client keeps reporting.
- Two weeks of parallel data decide whether the setup is ready. With both measurements live, the team can see how much data the server-side version actually adds. Such comparison often surfaces smaller quality problems that can get fixed before anything changes.
- Only then do the conversion IDs get switched. The swap happens on the back end, so server-side events register under the conversion actions that the campaigns were already optimizing towards. Creating new conversion points would wipe out the history behind them and drop every campaign back into a learning phase.
To judge whether a setup is ready, the team uses an internal checklist and a range of acceptable variance between the two measurements. Alongside that, the team leans on the reports inside TAGGRS to check the quality of what is coming in.
The consent report is the one they open most often. It keeps surfacing purchases where no consent was recorded, usually because the visitor’s consent came late in the session or they closed the banner without answering. Their decision is respected on the server exactly as it is in the browser, so those purchases stay unmeasured. FunnelTrack uses the report to advise the client on how to configure the cookie banner so more visitors make a choice earlier in their visit.
The results
Running both measurements in parallel lets FunnelTrack show something concrete to clients. Both setups measure the same events at the same time, so the difference between them is the data the client was missing before.
The size of that gap depends on the audience. Companies selling at higher price points tend to reach more Apple users, and, out of all browsers, Safari applies the strictest limits on tracking. Those visitors go partially unmeasured in a client-side setup, which leaves more for the server-side setup to recover. Device distribution works in a similar way, since the balance between mobile and desktop traffic decides how much of an audience is affected by these restrictions in the first place.
For one client, FunnelTrack connected the extra data to the revenue it produced.
- Brandon Teuben, Co-founder and Data & Tracking Strategist at FunnelTrack
However, these numbers have a limit that Brandon points out himself. Measuring 20% more data does not create 20% more revenue, because the revenue behind those unmeasured conversions already existed. The increase comes from advertising algorithms receiving more touchpoints to optimize on and from analyses running on higher data coverage.
There were a few more benefits that are worth highlighting.
Page load improves once the client-side container is stripped back. With server-side as the primary source, only the essential measurements stay in the browser. Purchases and pageviews remain tracked on both sides as a fallback, so a client can return to client-side measurement quickly if the subdomain ever goes down. Soft conversions move to the server, where they carry no weight on the page. Heavy tags such as the TikTok pixel are among the first to go, and containers that collected tags for 5 to 10 years without anyone removing anything show the clearest gain.
Margin logic sits on the server, so no profit data reaches the frontend. FunnelTrack hosts the calculation server-side and uploads the values through TAGGRS, which includes shipping and transaction costs. Nothing about the margins appears in the website code and the calculation adds nothing to page weight. For a client who already has Server-side Tracking running, profit measurement becomes a short follow-up project instead of another tool to buy.
Transactions that never reach the thank-you page get recovered from backend data. A visitor buying inside the Instagram in-app browser often closes the app before the confirmation page loads, so the data layer never fires and the sale is not registered by any tracking method. Feeding the client's backend data through the server-side setup adds those transactions back into the numbers.
Some clients get rid of standalone tools and save thousands of euros a month. Once features like profit measurement and offline conversions run on their own server-side setup in TAGGRS, the separate subscriptions behind them can be removed.
Key takeaways
An uplift of 0% tells you something important. Two measurements running in parallel turn the implementation into a diagnostic. When the server-side numbers match those on the client-side, it usually means that something in the setup is broken. Stacking server-side onto the existing conversion IDs hides that signal completely.
A slower implementation gives you a more reliable setup. Where a stacked build takes a day, FunnelTrack takes 2 to 3 weeks to complete the implementation stage. But, when a stacked build goes wrong, every measurement ends up in the same place and it’s hard to pinpoint the specific cause of the error, making troubleshooting far more expensive than the time saved.
Most sudden data loss traces back to the subdomain. A client who reports no measurement has usually had a developer change or remove the DNS records. TAGGRS supports switching subdomains quickly, but somebody still has to spot the issue, so FunnelTrack asks every client to leave the subdomain alone.
Server-side Tracking has limitations that should be stated upfront. Some data still gets unmeasured, such as visitors who decline consent or a purchase whose confirmation page never loads stays. The infrastructure makes both problems easier to solve through additional work that comes later.
A finished server-side setup is where the more advanced work starts. Brandon treats Server-side Tracking as a strategic choice and compares it to the foundation of a house you are going to build on. Profit measurement, backend feeds and structured datasets that can be queried with AI all run on that same infrastructure, so each one becomes a short follow-up rather than a project of its own.
- Brandon Teuben, Co-founder and Data & Tracking Strategist at FunnelTrack
