The primary job of a CDP is to make data ready for business use through unifying customer data and creating an accurate, up-to-date unified profile. High quality data is essential for being able to trust the profile, also known as a Golden Record, for being able to trust the accuracy and validity of audience segments and, ultimately, to trust that the brand is powering a personalized customer experience (CX) across all touchpoints in the cadence of the customer – up to and including real time.
You need to be confident that your CDP powers the best possible CX. This is true for any CDP – composable or not, and this is a key point that some self-proclaimed composable CDP vendors tend to gloss over when they tout composability. In their telling, it is enough to call yourself a composable CDP if all you’re doing is assembling CDP components on top of data that exists in a data cloud warehouse. Bring the application to the data, and voila – you’re a composable CDP. They make an assumption that your customer records have already been created, without addressing the shape of that data.
Prioritize Data Quality
If you take a closer look at what a composable CDP is meant to achieve – less complexity, greater agility, greater ROI at a low cost – much of that is lost when you entrust what should be core functions of the CDP out to a patchwork of multiple vendors. If your view of a composable CDP is that it extracts data from a data warehouse or data lake and syncs it back into operational tools (CRMs, marketing platforms, etc.), where are data quality processes being performed? You’re either assuming data that enters the data warehouse is already 100% ready for business use – cleansed, enriched, de-duplicated – or close enough that you think a basic deterministic match somewhere downstream will suffice for your CX use cases.
But how many organizations can say that their CRMs or other databases are in excellent shape and don’t have any outdated information?
These vendors are playing a data quality shell game where there is no central ownership for ensuring not only that data is ready for business use, but that every activation touchpoint is using the same consistent, unified view of the customer. Without that ownership there are always going to be shifting priorities and different interpretations for what’s considered “good enough” in terms of data preparedness.
For a basic, real-world example consider a marketer executing an email campaign. A composable CDP seamlessly integrates with an ESP, and in moving customer records for a John Smith from the data cloud it returns two different emails. One is [email protected] and the other is [email protected]. They’ve both been linked to the same record. But which one is the marketer supposed to use?
Some CDP vendors that claim that composability means that all you have to do is hand over your data (likely in the data cloud or data lake) and they will stitch it together. The email example is just one of many reasons why that concept isn’t ideal. A basic stitch is a far cry from an accurate Golden Record that is continuously updated in real time as data enters the system.
Composability + Data Quality = A Better Workflow
It goes back to trust, and providing that trust is the difference between a complete, composable CDP and one that is a loose connection of components that may be a superstar at one specific function, but in the big picture really just creates more work for both marketing and IT. Any composable CDP worth its salt should make it easier – not harder – to optimize CX by providing core CDP services, starting with the prioritization of data quality from the moment data enters the system. That includes data cleansing, enrichment, advanced identity resolution using deterministic and probabilistic matching, and real-time data processing capabilities for use cases that require real-time decisioning.
The right composable CDP will also make marketers’ lives easier by offering these services in a no-code environment – which applies to pulling data in, cleaning it and activating it to your end channels. When assembling various “best-of-breed” components for your composable CDP, ask if any will require your marketers to learn SQL.
Getting your data right is foundational for an exemplary CX, and that should be the overarching question when considering which composable CDP is right for your business. Will the composable environment help you take the data you have – in a data cloud, a private cloud or on-premise – and make it better? If the composable CDP you choose can dramatically improve your CX, you’ve made the right decision.
Editor’s Note: A previous blog entry on the composable CDP environment focused on the reverse ETL function.
