Working with a DTC brand who are going through a transformation which includes opening their first retail store with multiple retail stores in the roadmap. With the business model fundamentally changing while also sitting on a rich dataset it requires a different mindset in how to augment the data.
With a focus on driving better customer insights and being able to activate against those insights to do that in a complex martech world, the solution is a customer data platform which is the long term goal. As a business they had taken strategic steps with a long term view to be better prepared for what would be seismic shift in the business model.
2 of the key areas which was part of taking more control and a move to bring in-house while working with agencies in key areas.
- Customer Insights will be supported by various analytics techniques embedded into the process to answer business and marketing questions
- Building out the right Martech stack that supports the business requirements and then find the right CDP partner
Customer Insights
With the focus going on CDP implementation, it’s easy to get blinded by not having a deep understanding of customer segments + behaviours etc and the trends in how it’s driving revenue for the business. At a high level a CDP is based around customer segmentation, creating personalised + targeted messaging built around key touch points to get the customer to make a purchase. Understanding customer level insights becomes core to the business success and it helps at a strategic and tactical level.
Some of the analysis that will be done to better understand customer insights:
- RFM: segment customers to understand purchase behaviour and bucket them into different stages of the customer lifecycle
- Cohort Analysis: identifying patterns and trends on historical data to isolate groups to understand what is working and not working
- Market Basket Analysis: understand buying patterns of products purchased
- Volume range: understand customers looking at volume range analysis putting customers into four profiles split by high + low volume and high + low range
- Single customer view (SCV) looking at key metrics available will come into play particularly when drilling into RFM segment performance of individuals
Developing and bringing the core of the customer insights team in-house with the change in business model it provides the ability to embed it into the process and drive continuous insights and improvements impacting not only marketing but the wider business. The team will be integral in helping understand the value of the CDP with understanding which segments are best to test as use cases but also when the CDP is deployed help with the optimisation and building the right activation segments.
What is a composable customer data platform
Composable customer data platform provides full control and ownership over the data infrastructure with the data warehouse owning and housing all the data with the CDP activating the segments.
The advantages a composable customer data platform can provide:
- Build approach: take control of building the required ecosystem against the business outcomes that needs to be achieved
- Flexibility: activate against the data to help deliver against the business outcomes. Leverage the required data to help build the use cases
- Agnostic: Building an agnostic data infrastructure provides the option to work with any martech
The composable approach provides the ability to select the right CDP partner depending on business requirements which is likely to be in the form of a composable CDP. With several factors contributing to the decision the maturity of the business, internal or agency resources to manage CDP, costs and changes in martech landscape.
The CDP challenges
Very early in the process the common challenges were made clear with the goal of embedding into the approach and process ensuring the challenges do not become obstacles.
- Not having the right infrastructure and right vendors in place
- Poor data quality
- Not able to deliver against the goals
- Cost becomes a barrier to entry
- Not having the technical team to support
Point 1 to point 4 will be part of the planning process to solve, detailed more below.
- Point 1 to 3: The planning process will help in solve points 1 to point 3
- Point 4: The cost of the CDP will become clearer as the plan is executed but also understand the value the CDP will provide from testing the use cases that will be deployed
- Point 5: This is likely to be a mix of tech resources in-house and agency resources
The challenges can be overcome the most important steps to drive success is:
- Internal alignment across all teams and business units
- Setting the right expectations
Why composable CDP is the right approach
For where the brand is today, with the plan to be opening retail stores there is a clear direction of where the business is going and how they wish to better to provide a better customer experience. The build approach towards a ‘composable’ customer data platform helps in many ways:
- Cost becomes manageable
- The ‘composable’ approach allows getting the right infrastructure in place aligning to expectations and deliverables
- The ‘composable CDP’ provides huge amount of flexibility
Building the right team
To successfully deliver the CDP and wider project it needs the right team set-up to be able to deliver the project but as importantly it needs an executive sponsor who sits on the board and champion the project with providing the right level of support.
The core team is built around:
- Project Owner (Champion) – Executive sponsor
- Business Lead – Ownership and mapping out the business goals, KPI’s and use cases
- Technical Lead– Ownership of all technical requirements from architecture, resources from engineering to data scientists
- Marketing Lead – Ownership of the campaigns, delivery and performance of campaigns
Each of the business, technical and marketing leads will have a team of product managers, engineers, data scientists, campaign and marketing manager to deliver the CDP project. This will likely be a mix of internal and agency resources.
Building the Ecosystem
Building the right structured ecosystem underpins the data and tech foundations for the business. The end goal may be a Customer Data Platform but that should not distract from building the right infrastructure that can support the business and marketing requirements.
Building the right ecosystem is split into 4 different categories:
- Execution Owned
- Execution Paid
- Business & Consumer Intelligence
- Data Automation
Building on the Google infrastructure
It made sense to build the infrastructure around the Google ecosystem and build towards a composable customer data platform. Google has been building quietly a composable Customer Data Platform powered by the Google ecosystem with the development of Google Analytics 4 and Big Query. The evolution of the Google ecosystem provides the building blocks to take control of your data in how’s is collected, stored to developing insights and activation.
Google Analytics 4 is a modern analytics platform that allows you to combine website and mobile app data into a single property, its power comes from being a central data collection platform. Google Analytics integration with Big Query is powerful as it becomes one of many data sources that can be integrated to better understand the customer.
The evolution of the Google ecosystem with Google cloud provides the ability to consolidate data across the business building data pipelines not only Google led data but data from the marketing ecosystem, including CRM + ERP etc. It can provide insights to the business not only the marketing team but with predictive analytics capabilities. With the flexible infrastructure to building a composable CDP it can easily adapt to changes within the business and marketing plans.
Building the Composable Customer Data Platform
When building out a composable CDP the core martech is built around.
- Tag Manager
- Web + Mobile Analytics
- Data Warehouse
- Activation platform
This is where the Google infrastructure can power a composable customer data platform.
- Google Tag Manager
- Google Analytics + Firebase
- Google Cloud (Big Query)
- Google Marketing Platform
The bonus of the Google ecosystem it’s seamless how each of the tech integrates with each other.
The missing tech to add to the ecosystem is a Data Connector which pipes the data into the Data Warehouse which is connected to the Customer Data Platform.
The engine running the composable customer data platform is Big Query which is where the data sources will be housed. The key data sources in question are:
- Google Analytics (GA4) + Firebase
- AppsFlyer or another MMP
- Shopify (E-Commerce + In-store POS platform)
- Endear (CRM)
Planning for the Customer Data Platform
A long term view was taken when planning for the CDP which was driven by opening a retail store. The opening of a retail store provided an ‘end goal’ to work towards, it still requires all the Martech + data to be in place for a CDP to run smoothly.
The key milestones that had to be factored into the planning process:
- New E-commerce platform with Shopify
- New CRM with Endear
- New Marketing Automation with Klaviyo
- Retail store launch with Shopify POS
- New Marketing Data Warehouse with Big Query
- Launch of a new Mobile App (potentially) with Firebase / GA4 + MMP partner
What helped the planning building the Martech stack and data flows was starting from a blank piece of paper. It was a case of out with the old and in with the new. The Google stack will be heavily used with a new CRM + E-commerce platform.
- With a new Shopify led website the decision was made to start afresh with a new Google Analytics 4
- There was no marketing data warehouse in place that stores customer data + multi data sources and in likelihood it will become the central hub eventually for all business led data housed in Big Query
- First mobile app with Firebase / GA4 + MMP partner
- New CRM partner with Endear has a direct integration with Shopify providing a single view of customers
- New marketing automation with Klaviyo has many integrations. Shopify and Endear being the core ones
Knowing the key milestones it made the planning process ‘easier’ and to be more measured in the approach with a medium to long term plan. It helps to navigate not to fall into the trap most brands get into when deploying a CDP. With a half-baked implementation, minimal or no insights, not having the right use cases which leads to poor execution and unsatisfactory outcomes.
Building the plan
The plan starts in January which would be mapped out for 30 months:
- The Shopify E-commerce website would go live within 3 months
- Klaviyo marketing automation will go live once Shopify is deployed latest by month 4
- The Endear CRM was planned in for month 13 to 15
- The plan to integrate the CDP is around month 25 in an ideal scenario it would be around month 18
- The mobile app if it does launch will be before the retail store open so around month 25 onwards
- The goal to open the retail store within 2.5 years ideally in good time for the Christmas period
The CDP is the most flexible in the plan but its reliant on other variables:
- Deployment of CRM
- Data Warehouse becomes the single source of truth
- Understand the value of the CDP testing out use cases by activating audience segments
The CDP is planned to go live around 12 months after the CRM is launched. The 12 months will allow to go from unknown to known customer, with the data collected impacting seasonality, behaviour changes which become critical when it comes to building segments to activate against. The other consideration is how to deal with customer data from the old CRM. The common view was the old CRM should be stored in Big Query and joined with the new CRM data to provide a more robust dataset which can be used for analysis and activation.
The deployment of the marketing data warehouse is the cornerstone of the CDP and as important is data quality and governance are the foundations to the success of the plan.
Customer Data Table
Getting the right customer data table is going to be built from the new + old CRM that automatically updates collecting the following information:
Core table:
- Customer ID
- Order ID
- Revenue
- Order Date
- Product + Categories purchased
- First / Last purchase date
- Profit per order / product
It would be enriched with the following insights:
- LTV
- RFM
- Single customer view (SCV) metrics
This customer data table will provide all the relevant customer information to build segments to activate against.
Testing use cases to prove the value of a CDP
Working with the customer insights team helping find the best segments to test out at different levels of opportunity to provide the insights to understand what is possible. For each test need to identify and put into a plan:
- Map out all the tests that will take place including if multiple tests are required to validate
- The period required to run the test
- The reporting period to analyse the test
- The core KPI’s to monitor pre, during and after the test to understand the impact
For each use case it needs to be structured around:
- Outcome: What’s the desired outcome for the use case
- Stakeholders: Who will be most interested in the outcome of the use case
- KPI’s: What is the KPI to measure the outcome
Some of the potential tests use cases:
- Suppression of ads to customers who have purchased i.e., last 2 weeks
- Users who have added to basket but not bought at a certain value the last 3 days
- High value segment to improve retention rate
- Low value segment to improve purchase rate
The segments can be activated in multiple ways:
- Build segments in GA4 and activate in GMP
- Build segments in GA4 and activate in Facebook
- Export list of customers and import into i.e. Facebook, TikTok etc
At first run each of the tests separately to get a clean dataset for analysis, there may be an opportunity to run multiple tests at the same time once there is more confidence in the process and ability to split the segments and tests results for analysis.
How this helps:
- Testing of the use cases becomes a manual process; it’s likely to help understand the requirements and in turn it help grow the data maturity of the team
- Testing multiple use cases with different opportunities will provide insights on the business impact
- Understand the ROI impact for a CDP
- Understand what kind of CDP is required and the goals for the CDP
The goal is to get 3 solid use cases before proceeding to find the right CDP partner. In the plan there is an 18 month period for testing but till 3 solid use cases are not found the testing period will likely be extended.
Finding the right CDP Partner
Finding the right CDP partner is very much dependent on delivering against the plan and having 3 solid use cases. It will also be very much dependent on the maturity of the business at the stage, the business and marketing needs will determine the right CDP partner. During the period there will be a lot of time and investment put into people, training and process which is a requirement in getting success out of the CDP.
Other considerations of the plan
Both very important, server side tag manager and identity resolution was not baked into the main plan as it was seen as more flexible but dependant on other implementations.
Server Side Tag Manager
With the current plan the goal should be to deploy server side tag manager around month 15 which is 1 year after the Shopify deployment, and any issues should be ironed out. In all likely this would be outsourced to an agency or vendor to set-up and manage for the short to medium term.
Identity Resolution
Is the identity resolution going to be managed by the CDP or the data warehouse? With the approach taken to the planning and implementation the data warehouse is likely the best solution. With Big Query the data warehouse the identity resolution is likely to be done with Neo4j. If it’s decided that the CDP can manage the identity resolution, then it would become one of the requirements when finding the right CDP to partner with. Assuming the plan is delivered in the timeframe the identity resolution should be looked at around month 18 post the deployment of the new CRM and by then the old CRM should be stored within Big Query.
