Five Myths about Data Management Platforms (DMPs)

Published on September 8, 2017 - by jerry helou under analytics

I heard about DMPs for the 1st time when Adobe acquired DemDex in 2011 and rebranded it as Adobe Audience Manager (AAM). DemDex’s acquisition had happened within two years of its launch as a “behavioral data bank” in 2009. Back then you could count DMPs on a single hand but in 6 years, we have lost count of the number of newcomers. Adobe is not the only brand that saw the value in DMPs. In 2014, Oracle acquired BlueKai followed by Salesforce acquiring Krux in late 2016. DMPs are still hot and are continuously evolving. In addition to AAM, BlueKai and Krux, there are few other DMPs such as MediaMath, Neustar, Segment and Tealium AudienceStream that are growing fast.

If you are new to the DMP world, I recommend you taking 3 minutes to read about it in my blogpost here. Now that we are on the same page , let me walk you through 5 myths that are often discussed in the DMP domain.

Myth #1: DMPs can replace your analytical tools

The name “data management platform”, is a bit deceiving. Marketers tend to presume that DMPs can replace their analytics platforms. Not exactly. There are few things that analytical platforms do well that DMPs cannot. Let’s highlight the top three things:

1. Access to all hit level data: DMPs will not give you the same level of access to data like an analytics platform simply because DMPs are meant to be a data hub (collect-segment-activate) and not a data mining tool. Generally, you will have to pick and choose the data that needs to be stored in a DMP while an analytics platform will collect and keep everything. Of course, there are few exceptions to this rule where Krux (Salesforce DMP) exposes the data you choose but makes all data available for retrieval and activation any time.
2. Real-time reporting: Not many DMPs offer real-time reporting to audience and segment populations. In most cases, we see a 24-hour lag in reporting.
3. Sequential Segmentation: Sessionization is crucial for analytical platforms which paves the way for sequential segmentation such as a visitor visiting page A, followed by visiting page B, followed by purchasing product C and all within a single visit. DMPs do not have this sessionization capability and therefore their segmentation is limited to the visitor lifetime with very little control over the order of activities.

Myth #2: DMPs create audience segments automatically

Me: Alexa! Ask the DMP to create intelligent segments for the upcoming campaign.
Alexa: Sorry I don’t know that.

I wish we could feed data into the DMP and it automatically creates intelligent segments. Although some DMPs brag about their AI capabilities, we are not quite there yet. Segment definition and creation are still done manually. You need to define the criteria for every audience segment, track their activity and understand how they impact your performance metrics. Based on this analysis, you will adjust and refine the segments until they are performing well. This process does sound like machine learning with human oversight but is still very human dependent. The common predictive and machine learning features that common DMPs offer are around look-a-like modelling and the ability to extend your loyal audience beyond your existing dataset.

Myth #3: DMPs are the centralized source of all your audiences

You have probably heard me say this: “The DMP is the central hub for your audience segments. It should be the source of truth for all your segments in your organization.”

This is true when it comes to segments that contain anonymized audience but not when it comes to segments containing personal identifiable information (PII). You need to be careful with how to mix these two together. Data in a DMP is tied to a device (cookie, mobile device id) or an authenticated user (hashed customer id) but it cannot be tied to a name, email address or SSN. Therefore, PII data needs to live outside of the DMP or needs to be anonymized, stripped off PII and then uploaded into the platform. This makes it very difficult to host all your audiences in a single centralized location.

Myth #4: DMPs operate in real-time

Well what is real-time and what operation are we referencing? For data collection, behavioral data is usually collected instantly via a tag that fires on a page or within an app while on-boarded offline data is dependent on the DMP’s processes to ingest this data and make it available online. In general, this might take 24 – 36 hours which is far from real time. When it comes to activation, it varies from one vendor to another and it depends on the type of integration in place (HTTP, batch file or server-to-server). For example, the server side integration between Audience Manager and Adobe Target allow instantaneous activation within page views, while other server-to-server integrations with DSPs might take 2 – 3 minutes for segments to be transmitted.

Myth #5: DMPs can replace 3rd party marketing tags

DMPs deploy an inline frame within their tags, also known as iframe, to trigger ID Syncs with vendors (data and match providers, DSPs, ad servers, and others). This is crucial for data transfer between the DMP and the vendor. This feature should never replace 3rd party marketing tags on the site. DMPs are not meant to be server side tag managers (with the exception of Signal). Leveraging the DMP server side integrations might be tempting to reduce the number of marketing tags on the website, ultimately to enhance page performance, but be mindful that adding the DMP in the middle introduces another suspect to the finger pointing party when you run into issues.

The good news is that DMPs continue to evolve and one day these Myths might become true, I have no doubts about that. Until then, make sure you understand the main value of the DMP: The hub for your audience segments enabling onsite and offsite experiences.

Jerry Helou, Ph.D.

Linkedin Icon

Jerry Helou, Ph.D.

Jerry Helou leads the Digital Experience Architecture practice at Softcrylic. He helps our clients accomplish advanced digital experiences and strategic business goals by implementing and leveraging multi-solution architecture.

Start typing and press Enter to search