In the context of Google Tag Manager, server-side containers are Tag Manager implementations that trigger Tags from a dedicated Server. Unlike the traditional web implementation, which triggers Tags through the user's/visitor's Browser.
The basic concepts of web GTM persist in server-side: Events, Triggers, and Tags. However, there is one additional entity, the Client, which is responsible for receiving HTTP requests and translating them into GTM model events.
The use cases for configuring a Server implementation for Tag management are varied. The main benefits of this approach are:
Prevention of event firing blockage by extensions (Ad Blockers) or privacy-focused browsers;
Redundancy in sending/receiving events via both Web and Server.
Using a server container provides greater coverage in event firing and user data collection. This is possible through the redundancy generated by firing the same event via both web and server. If an ad blocker prevents the Web event, the Server event can still deliver the data to the destination platform.
There are various possibilities for implementing a server container. From simply contracting a service, like Stape, to more technical configurations, like Self-Hosted. Next, we will see some of these options.
Cloud Servers
For cloud service options, Google Cloud itself currently offers two possibilities for hosting containers. Alternatively, Stape.io offers a simplified configuration and pricing option (which can result in lower costs depending on the event volume).
Google Cloud
Currently, the platform options available for this purpose on Google Cloud are App Engine and Cloud Run. Until the end of 2023, App Engine was the default option for automatic provisioning. That is, when creating a new server container in GTM, it was possible to automatically generate a project with the basic App Engine configuration pre-configured. Currently, the automatic provisioning option creates a Cloud Run.
App Engine vs. Cloud Run
The platforms have many similarities, but Cloud Run is newer and proposes higher speed and processing capacity. This difference is only noticed with a high volume of received events. For projects with low volume, the main difference lies in configuration and pricing.
App Engine, unlike Cloud Run, has the option to be used at a "test" level. Although reduced in processing capacity, in this test configuration, App Engine delivers enough for smaller projects. The main advantage is the differentiated pricing, generating much lower costs compared to "production" level configurations.
You can find more about the differences between App Engine and Cloud Run in the official documentation.
It is worth noting that to use App Engine as a server container, it is now necessary to perform manual provisioning. You can check the step-by-step guide in the official documentation.
Stape
Stape is the simplest option, with very low configuration complexity. One of the services currently offered even creates the environment prepared to fire events to Meta's Conversions API.
Stape uses Google Cloud's cloud environment but operates with completely different pricing. It may be more interesting depending on the number of projects that need to be configured and the volume of events in each project.
GTM - Self-Hosted
Alternatively, it is also possible to host Tag Manager Server-side on your own server. This would be the option with the highest technical complexity, as it is also necessary to manage processing scalability in case of sudden event spikes.
Conclusion
In the current state of pricing and available services, I consider three server-side configuration scenarios to be most advantageous:
A small project: An App Engine configuration in test mode will generate the best cost;
A few small/medium-sized projects: Server configuration via Stape.io, with a plan suitable for the data volume;
Several medium or large projects: Projects configured via Cloud Run.
Choosing the right service/platform to host the server container implies variations in costs and complexity. The availability of these services changes constantly, and new options will appear regularly. This article will be updated as new options become available.
Comments