Data Management Platforms (DMP) are a set of SaaS offering in the digital advertising ecosystem that have been living for may be around five years now. DMPs have been gaining market noise and Forrester Wave recently published a wave report also. It's easy to get confused with Tag ManagementSystems that we discussed previously. I believe this post will give a basic understanding of what DMPs are and how they differ from TMS. A DMP's function is three fold - first to aggregate various data sources, then to integrate them to provide analytics and segmentation capabilities; and finally to deploy actionable insights for various vendors - be it advertising or user experience.
Let's try to dig a bit deeper into each of these capabilities --
Aggregate data sources
Aggregation of various data sources is the first step that any DMP starts with. Data sources include both online - be it site side analytics, ad serving or optimization and offline - be it transactional data or call center data. These days, all leading DMPs allow collection of data from literally all possible online sources and custom offline integrations.
Integration of data sources
This is the core offering of any DMP. Once the various data sources are integrated, DMPs try to draw conclusions about the profiles of data captured and also provide opportunities to integrate with 3rd party audience segmentation sources like Nielsen. This integration allows Data Management Platforms to provide normalization and segmentation information based on the data accrued. Analysis is done on two fold - based on the first party data that the publisher accrue and then getting input from wider third party data. Let's see a simple example of how this is done.
- You make a purchase from a leading e-commerce website. To make purchase, you have given some basic personal information to the company
- You visit the brick and mortar store of the company to make a further purchase or return something. Now the company can stitch between your online and offline identities
- Based on the information we have, you may be classified into a profile like Male between age group 25 and 35, residing in Kerala state interested in value for money smart phones
- You see a display ad showcasing various offers for the smart phone model's accessories; but you didn't click it. However DMP captures the impression
- You see a 10% discount ad on an accessory and clicks it, but didn't make the purchase. DMP captures the profile as Male between age group 25 and 35, residing in Kerala state interested in value for money smart phones interested in headsets
- Since DMPs create individual profiles, an integration with DSP enables the company to display targeted customized sequential ads for you
- You are shown an irresistible offer on the headset, you click and make a purchase
- Based on these individual engagement tracking and probabilistic modeling, DMPs create further audience segments
- This analysis is fed back into the system for enhancing the segmentation algorithm
Don't believe this will all work? Check Bluekai Registry to see under which all segments you belong and who all are tracking you. Of course Personally Identifiable Information(PII) may not be shared by the publisher.
Deployment of actionable insights
Once audience profiles and other insights are developed, DMPs allow it to be directly passed on to a DSP or other online channels. Independent DMPs allow buying actions to be linked to multiple DSPs. It's not just limited to media buying; one can integrate with other marketing channels, for example like Kenshoo for Paid Search.
DMPs can be pure-play vendors like Bluekai (Oracle) or Lotame; they can part of solution, for example Turn which DMP/DSP integrated or part of a wider marketing cloud like Adobe; or it can be in-house built. DMPs make sense to those organizations which has built first party data over time. Using the first party data collected as the foundation, DMPs allow accurate customer identities across platforms/devices and optimize media.