Skip to main content

Building the right architecture for your martech stack is a big topic in driving success and that it meets the wider business and marketing objectives. Late summer 2020, I started working with a brand that was primarily retail focused with digital and e-commerce becoming a bigger part of their business model at the start of 2019 which got accelerated in 2020.

To understand the complexity, the brand did not have ownership over the .com domain, so decided to go for a country domain approach. Just my view, as the digital and e-commerce business grows as expected the business will look to buy the .com domain which still has the strongest signal with Google then any country domain, and the acquired 100 country domains will provide brand protection.

For where the brand was in their digital maturity journey and where they expect the brand to be in 3 to 5 years-time. They believe the Google martech stack provides the right integration between ad-tech, analytics, personalisation, and other key tech such as TMS and Cloud. The other consideration was insights through Econometrics that Google had so far delivered a stronger ROI than Facebook. There was more growth in Google combine that with their martech stack and Facebook being impacted by Apple tracking changes was also another consideration in committing to the Google stack.

This project started 20 months into the 5 year plan.

Auditing the martech architecture

Before I had started auditing there was an acceptance to go back to the drawing board. It was evident that the brand was not data and tech savvy at the start of their digital maturity journey. The top headlines from the audit:

  • The analytics data was mainly being used by the different agencies, not internal teams
  • Multiple Google Tag Manager and Google Analytics accounts per country domain and there was no real thought behind the set-up
  • The primary KPI’s that were set-up for GA but also Facebook, Adwords etc was set-up in-correctly
  • There was no trust in the data, it did not tie back to objectives

The GTM and GA set-up mirrors the website structure built around a top level country domain + 3 sub domains.

The domain set-up was as follows:

  • Consumer:
    • abc.countrydomain
  • E-commerce:
  • Content:
  • Logged-in customers:

The 10 countries were split into the following:

  • 7 of the countries were e-commerce led
    • 3 of the countries were the growth drivers
    • 4 of the countries had just launched e-commerce in the last 6 months
  • 3 of the countries had just launched a website and e-commerce would not be launching for minimum 6 months

In total they were:

  • 31 GTM and 31 GA accounts across 10 country domains
    • 4 GTM and 4 GA account’s per country domain x 7 (e-commerce focused)
    • 1 GTM and 1 GA account per country domain x 3 (non-e-commerce focused)

To make it more complex there was no one agency managing everything digital from website, analytics, and measurement related. There were 5 different agencies involved across all 10 country  websites. Frustratingly not one person within the business had full admin + user access to any of the 31 GTM or 31 GA accounts.

The business hired across key areas such as developers, data visualisation to SEO who had started building out a plan to move away from a sub domain structure. My role was to ensure the right architecture for the martech stack is designed and implemented with an optimal set-up of GTM + GA that can drive the right data and analytics aligning to the business and marketing objectives. The focus would be on the 7 countries that were e-commerce led.

Designing the martech architecture

The process had started to move away from sub domains and migrate over to one country domain which dictated there should also be 1 GTM and 1 GA per country domain. The set-up of Google Analytics would be on UA / GA3, GA4 would also be implemented which aligns to the objectives in what it can offer particularly with the free Big Query connector.

My recommendation was there needed to be a proper end to end testing platform with a development, staging and a production (live) environment.

  • This is to ensure the user does not have to deal with any issues on the live website, and any issues are dealt with either on development or staging.
  • It ensures any changes made does not impact or break analytics tracking
  • It allows for robust testing of any changes that need to be tracked, how the data is going to be collected and reported

The end to end testing platform would be supported with a solid analytics set-up where a unique GTM and GA would be implemented for development, staging and production environments. The option was to implement a single GTM across all environments to make the process of managing and rolling out the changes.

My preference was to have 1 GTM and 1 GA per environment:

  • Splitting out the 3 environments into its own GTM helps the process of not exceeding the 70% size limit. Which would have a domino impact on performance from site speed to tags not always firing
    • All 3 GTM containers would be aligned particularly staging and production
  • Having a separate development and staging GA, will allow the data to be easily validated
  • It ensures new tracking features which are still being tested are not accidently pushed into the production GA
  • Working with multiple agencies and internal stakeholders providing and managing access will be much easier depending on requirements

Implementing the martech architecture

We had a total of 100 days to implement the agreed martech architecture across 7 country domains that were e-commerce focused to be completed by Christmas 2020. With the goal of going into January 2021 having a solid set-up of the martech architecture and data that can be trusted. 100 days was an aggressive push, but the business could not afford to be going into their 3rd year with any further delays could massively impact future milestones.

Prior to the 100 days the process had started to start afresh with implementing GTM + GA across all environments and domains. Across the 7 e-commerce domains (development + staging and production environments) and 3 non e-commerce domains (production environments).

  • 20 GTM containers
  • 20 GA3 properties
  • 20 GA4 properties

GA3 was the core focus of the set-up, enhanced e-commerce was set-up + across GA3 and GA4. With GA4 eventually going to be become the core analytics tool start collecting data allowing for comparison between both tools but also analysis using data from GA4.

With e-commerce becoming a bigger part of the business model there were 4 core analytics requirements:

  1. Enhanced e-commerce
  2. Content grouping
  3. Product stock status
  4. Error page tracking

  (download Production Live GTM Container)

We looked to create a templated approach to create efficiencies with the dataLayer and GTM containers that can be easily replicated across all country domains and environments. We picked one of the countries that just launched e-commerce to understand processes, building and testing of the dataLayer with GTM + GA was the most critical component in the 100 days.

The focus was on getting this one country perfected from testing to rolling out live. The time invested was 40% (from the 100 days) which was done with limited setbacks (helps having shit-hot developers) came down to having a vision, right expertise, a clear plan, and a goal of what needed to be done.

  • 15% designing the dataLayer (followed Google enhanced e-commerce template with adding extra data variables such as stock status etc)
  • 40% building and testing the dataLayer (within staging environment)
  • 35% GTM and GA set-up (3 GTM + 3 GA development, staging and production environments)
  • 10% collecting and validating data (pushed changes into production environment)

The implementation was split into 5 phases: (100 days started from Phase 1)

  • Phase 0: Implement a single GTM and GA across all 10 country domains and environments
  • Phase 1: Created a templated approach to create efficiencies focusing on one of the country domains that just launched e-commerce
  • Phase 2: DataLayer deployment, GTM and GA set-up, rolled out the changes to the 3 main e-commerce domains with the goal to roll out the changes few weeks before black Friday
  • Phase 3: DataLayer deployment, GTM and GA set-up, rolled out the changes to the 3 smaller e-commerce country domains
  • Phase 4: Testing, validating the data and building dashboards

This was the agreed plan, the only change that occurred was phase 3 got extended out by an additional week with it due to be completed over black Friday weekend.

Evolving the martech architecture

Having built the core foundations evolving the martech stack it’s important when adding any new tech that it aligns to the business objectives and the 3 to 5 year plan. The tech that I see providing the most value and when it fits into the plan:

  1. Google Cloud: where I see the value is building a data warehouse of all data sources available from GA4, CRM etc to help power measurement, insights, and visualisation (planned to launch in 12 months)
  2. Server Side GTM: having only launched a robust client-side GTM there is no immediate requirement to launch a server-side GTM. The growth of e-commerce will dictate the importance of migrating over to server-side GTM. Weighing up the cost benefit will be important (look at in the next 12 to 18 months)
  3. CDP: would have to be a non-Google tech, due to the nature of the business model being a mix of retail and e-commerce it would require a CDP that can segment and activate in an omni-channel world (look at in 18 months)
  4. Clean Room: would have to be a non-Google tech, looking to do richer segmentation with 1st party data (i.e., publishers) and build audiences to activate against (look at in 18 to 24 months)