There seems to be a lot of confusion in use of terms like data warehouse, customer data platform (CDP) and data management platform (DMP). It’s been endlessly confusing for me, so I spent a little time researching what all these terms really mean, and how they differ from one another. I’m probably painting with too broad of a brush here, but here’s what I came up with:
Data Warehouses
The nature of a data warehouse implies structure. It’s built to answer specific business questions. It likely doesn’t contain all your data, but a filtered selection structured and joined with other tables to answer, say, marketing attribution, return on investment, etc.
The structure of a data warehouse is both an an advantage and disadvantage – as the moment you have a question whose scope is beyond that of your current data warehouse, database-level changes are usually required to insert and transform the data required to answer it. In short – a data warehouse is as good as the questions that have already been asked.
Data Management Platforms
A Data Management Platform (DMP) is a bit more specific in function. DMPs tend to be more use-case specific around digital advertising. They’re about unifying your known, first party data with second and third party data sources from across the web to better target advertising audiences, market across devices and platforms, and run more efficient advertising programs.
Customer Data Platforms
A customer data platform CDP shares some similarity with a DMP in that it’s an aggregator of data, but it tends to rely more on first party data and the unique identifiers that join them (name, email, etc). The advantage of a CDP is that you connect the disparate system data from your CRM, marketing automation, and other analytics tools into a stitched together view of the customer. One you can use for analysis, or for feeding back into the end-use systems.
But a CDP seems to differ from both data warehouses and DMPs in terms of scope and structure. The CDP is in theory a complete view of the customer. It’s all the data you know, configured more in a data-lake style configuration. With a CDP you’re trading the up-front structure of a data warehouse for deeper access to data to answer future questions. Most of all, since a CDP is something that is ingesting all of your data, you can start making novel connections back to the end use tools, to, say pass a key customer interaction within your product back to your marketing automation system to send a targeted message to the key user on the account. A big pitch of the CDP is its ease of use – the pitch, anyway, is that you’re not forced to leverage an IT team to make simple changes to it.
The question then is – which do you need? a DMP seems much more of a specific use case – and probably only applicable for advertising-heavy teams. The real distinction seems to be whether a data warehouse or a CDP is necessary. To me, it feels like a CDP (or CDP-like systems) will increasingly replace the needs for a conventional marketing data warehouse. The complexity and scale of marketing data needs will always be at odds with an IT or analytics team’s ability to customize it. The more accessible these systems can be for changes, new data sources, and new insights, the greater value of the system to the organization.