GA4 CDP with GTM Server, Metabase, and Omni Analytics

A novel and cost-effective CDP based on GA4 boosted with private cloud extensions

GA4 alone is not a full CDP (a proto-CDP at best) because it does not allow processing or storing complete customer data records (e.g., names, phone numbers, emails). This service enhances GA4 with Omni Analytics, GTM Server, and Metabase, all within your private cloud, turning it into a de facto CDP. This enables enrichment, server-side tagging, and real-time reporting. The service is primarily aimed at budget-conscious businesses that are fine with not having full control over data pipelines inside their private cloud (since GA4 operates in Google's cloud) and accept limited customization options for orchestrating components. GA4 serves as an actual data collector in this service, and it is free since Google does not charge for processing events.

GA4 as a data collector, not just a reporting platform

This service relies on GA4 for both client-side and server-side data collection using Google Tag and GA4 Measurement Protocol. Omni Analytics sits upstream of GA4, capturing, enriching, and ingesting third-party data points into GA4, building a complete user journey. All data is routed to server-side tagging containers for activation. Warehousing uses GA4's streaming inserts to BigQuery, and the reporting layer is built on a data model from intraday event inserts. GA4 provides real-time reporting, which can also be run externally. This approach creates a simplified CDP without storing personal data in GA4.

Google trackers and endpoints

We use Google Tag for first-party client-side tracking and Omni Analytics for sending cleaned and structured server-side events to GA4 via the Measurement Protocol.

Server-side activation and streaming inserts to BigQuery

Tagging operates server-side through environment-specific containers, with all data loaded into the warehouse using the BigQuery loader.

Basic reporting in GA4, custom metrics in Metabase

While GA4's reporting capabilities are limited, we enhance them by developing a custom data model that generates clean datasets for integration into Metabase.

A reliable and cheap entry-level CDP

GA4-based customer data infrastructure is the most cost-effective option for managing customer data. Unlike Snowplow or Omni CDI, GA4 operates without needing a dedicated team and is free regardless of data volume. However, it lacks full infrastructure control, requires rigid tracking implementation with significant custom code, and isn't highly customizable as a data collector.

datomni blog contents data private

Client and server events tracked

We track all data sources using Google Tag and Measurement Protocol. gtag embedded in web pages records user interactions—clicks, page views, and custom events. Omni Analytics formats server-side data from third-party sources for the Measurement Protocol Data Client, aligning it with GA4's data model. Third-party server-side events are enriched with client-side user identifiers and reported with unified user ID, ensuring data stitching in GA4 across all sources. No external identity resolution or tracking tools are needed.

Destinations activated server-side

GA4, serving as the data collector in this service, activates downstream destinations such as marketing or advertising tools and monitoring systems server-side. With a unified user ID across all sources enabled by Omni Analytics, you can optimize your ads for lifetime value, not just ephemeral conversions. No additional SaaS tools are needed, and all extensions operate within your cloud infrastructure.

datomni blog contents analytics

Custom data model preferred

We utilize GA4's streaming insert feature into BigQuery for data processing. Instead of relying on raw data directly inserted into BigQuery, which may contain noise and require sessionization and other transformations for usability, we employ a custom metric model to derive business metrics. These metrics are processed by a data modeling job, producing clean final datasets displayed on your private instance's Metabase dashboard. We have also open-sourced some of these models.

GA4 CDI Custom-Built Components

Your infrastructure will utilize multiple pre-built components that we've open-sourced or made available to our clients. We add new features weekly. Here are the highlighted ones that we use in GA4 CDI. The majority of components that are not already open-source will most likely be open-source in the near future.

Omni Tracking WordPress Pro Plugin for GA4

Our custom WordPress plugin tracks GA4 using both client-side and server-side methods. For server-side tracking, you'll need an active Omni Analytics instance with active authorization tokens. The plugin manages all WooCommerce e-commerce events and other useful actions. It is closed-source and available exclusively to our clients.

Omni Analytics GA4 Measurement Protocol Data Client

GA4 Data Client is integrated into Dockerized instances of Omni Analytics platforms as part of Omni CDI. Omni Analytics structures and dispatches events to server-side Measurement Protocol sinks, regardless of the source. You can define your mapping schema for dispatched events in your Omni Analytics instance.

Omni Activation Meta Conversion API Tag for GA4 Data Client

Activation tag for Meta Conversion API, utilizing deeper logging levels and a variety of configuration options, fully compliant with the standard GA4 Data Client. Closed-source for now.

Omni Activation Mixpanel Tag for GA4 Data Client

Activation tag for Mixpanel, utilizing deeper logging levels and a number of configuration options, fully compliant with the standard GA4 Data Client. It implements all essential Mixpanel API methods, including identify, merge, and profile building.

Omni Modelling CAC & ROAS dbt template for GA4

Datomni's popular dbt package, downloaded frequently, provides essential metrics like CAC and ROAS. It's tailored to calculate these metrics using aggregate spend and attribution models for major advertising platforms such as Meta, Google Ads, TikTok Ads, LinkedIn Ads, and others. Adapted to the Fivetran data model, with the Omni ETL version coming soon.

Let's build this, PoC route

We create a GA4-based customer data platform with a PoC approach. We launch with a quick PoC in 3 weeks, then make improvements in 2-week steps based on what's most important to you. You can choose how many steps you need to make your system better, even none if your team can make it better on their own. Let's schedule a call to get started.

Latest blogs on analytics