Customer Data Platforms: Off-The-Shelf vs Composable
Every business is looking to get closer to their customers and create that ‘one source of the truth’ but this can be immensely challenging. Customer data is generated and stored across various platforms, departments, and systems, leading to a fragmented and inconsistent view of each customer. Integrating data from multiple sources, keeping on top of data quality issues, duplication, and outdated information all pose significant obstacles to achieving a comprehensive and accurate customer profile. So what can you do?
In recent years there has been one key technology aimed at solving this problem. Customer Data Platforms or CDPs. Their job is quite simply to take information from across different channels and collate it in one place. These platforms such as Segment and mParticle have been very popular with marketing teams as a way of getting closer to customers, but what are their pros and cons and could a data pipeline act in the same way?
Off-The-Shelf CDPs
What They Are: Customer Data Platforms (CDPs) like Segment and mParticle are tools designed to help businesses unify and manage customer data from various sources in a centralised platform.
Pros:
Consolidation of data: Customer Data Platforms allow businesses to identify and integrate relevant data sources from various channels and touchpoints. By consolidating data from multiple systems, such as CRM, marketing automation, and e-commerce platforms, organisations gain a comprehensive view of customer interactions and behaviour. This integration enables businesses to make data-driven decisions based on a holistic understanding of their customers.
Event tracking: Customer Data Platforms operate on an event-based tracking model, capturing granular data points for each customer interaction. This approach provides a detailed understanding of customer journeys, allowing for deeper analysis and insights. By tracking events, businesses can unlock valuable information about user behaviour, preferences, and engagement patterns.
Segmentation: Customer Data Platforms enable businesses to segment their audiences based on various attributes and behaviours. This segmentation facilitates targeted messaging and personalised marketing efforts. By creating specific audience segments, organisations can tailor their communications to better resonate with different customer groups, leading to improved engagement and conversion rates.
Cons:
Scalability Costs: While Customer Data Platforms offer numerous advantages, they can become expensive to maintain at scale. As the volume and complexity of data increase, the cost can rise significantly, with many tools charging you per event tracked and this may include non-logged users. This means that as more users interact with your website/app/business your cost will increase dramatically.
Complex Metrics and Data Science Integration: Customer Data Platforms may face challenges when integrating complex metrics and advanced data science capabilities. This could include metrics such as lifetime value or likelihood to churn. While it’s possible to integrate these sorts of metrics within the tools, clients tell us it can be very difficult and relies on complex logic which adds to engineering time and cost.
Mirroring Data Warehouse Logic: Customer Data Platforms may end up mirroring the logic that already exists in the data warehouse. This duplication of logic can lead to redundancy and complexity in data processes, or it may actually lead to a silo in which marketing (who tend to be the prime users of CDPs) have one version of the truth and the rest of the organisation relies on the data warehouse.
The alternative? Your Data Stack
It’s possible to replicate many of the features of a CDP tool within your own data stack without the downsides. What is a data stack? It is essentially all the different tools you’ll use to organise, transform, visualise and analyse data.
Pros:
Not reliant on one technology: Data Stacks leverage scalable and flexible technologies such as cloud-based data warehouses, data lakes, and modern ETL (extract, transform, load) processes. This approach allows organisations to build a robust and adaptable data infrastructure that can handle large volumes of data and support complex data analytics tasks. It also means that your data infrastructure is not solely reliant on a single tool or supplier.
Create Any Metric, Model, or Report Needed: Data Stacks/Composable CDPs empower organisations to create any metric, model, or report they require to drive insights and support decision-making. Unlike off-the-shelf tools, which may have limitations in customisation, these platforms offer the flexibility to define and build metrics and models specific to the business’s unique needs. They are certainly better suited to handle metrics which may rely on longer-term data such as lifetime value or likelihood of churn, which is where Off-The-Shelf CDPs typically struggle.
One place for data: Data stack/composable CDPs streamline the process of reporting, insights generation, and data activation. By centralising these functions within your warehouse you create that singular source of the truth across all departments as opposed to a CDP tool which may end up creating a silo between data and marketing teams for example.
Cons:
Larger Upfront Costs: Implementing a Data Stack/Composable CDP can involve larger upfront costs compared to off-the-shelf solutions. Building a modern data infrastructure, setting up data pipelines, and ensuring proper integration with existing systems require substantial investment in terms of technology, resources, and expertise. While the cost is larger upfront, there is however a longer ROI in that once something is modelled, it doesn’t need to be done again. That means that while there is a larger upfront, there isn’t a cost tied into the amount of users or growth in the way that most Off-The-Shelf tools will, which can work out significantly cheaper.
Requires Engineering Resources: A Data Stack/Composable CDP typically requires engineering resources from the outset. These pipelines rely on the expertise of data engineers and analysts to design and implement the data stack architecture, develop data models, and ensure data pipelines are functioning correctly. Organisations need to have a dedicated team or access to external expertise to successfully deploy and maintain the platform.
Speed Of Reporting: While data stack/composable CDPs offer powerful capabilities, the speed of data activation is not real-time as is achievable with an Off-The-Shelf CDP. The fastest speed most likely achievable is down to the hour. For most businesses this completely fine, but is still a consideration. ‘Real-time’ insights always seem very desirable but you must consider how strong the use-case is on this.
Conclusion
If you haven’t implemented a CDP, then our advice is to use your data stack instead. It’s an approach which will likely be more cost effective in the long run and will offer you all of the functionalities of an Off-The-Shelf tool.
If you’ve already implemented a Segment or an mParticle, we are not suggesting that you swap it out today, use it for event tracking, but do not try and implement data science models within that tool, as this is not its strongpoint. In the long-term you might consider replacing this but do so at a point where the ROI calculation is easy.