Most digital marketers are looking at or have already acquired a Data Management Platform (DMP). It somehow can be compared to the shiny new toy that every kid wants for christmas. For those who haven't yet, you would most likely already benefit from "some" advantages of a DMP through your agency's trading desk or whoever managing your media spend. The limitation in such case is data ownership and not being able to leverage that data outside of the paid channels' remit. This is why most brands look at deploying their own DMP internally.
Now for those who have read this first paragraph and wonder what I am talking about, a DMP is a centralised platform, most likely SaaS (Software as a Service), used to aggregate all sort of data from different sources. However, don’t mistake it for a datamart or a datalake, it is not meant for big data but rather for a selected data set that can be leveraged in a “marketing” context.
The key value proposition of a DMP is about turning this selected datasets into audiences/segments that can be made“actionable in real time”. This explains why the SaaS model considering the need for edge servers distributed across the globe, as close as possible from your customers to limit latency when either collecting data or resolving segments. Btw, I’ll use interchangeably the terms segment and audience in the rest of the article to design a group of people with similar defined traits.
As per Gartner’s last magic quadrant, the main DMP software vendors are:
Before drilling into the different use cases, let’s briefly touch on what kind of data can be aggregated within a DMP. I’ll be quick as this is a topic widely covered across many other articles on the net.
"USE CASE #1: Optimise acquisition"
Now getting into the meat of it. I’ll start with the use cases related to paid advertising and first off the rank is how to increase programmatic bidding performance.
In that case, the DMP allows you to bid for customers that are likely to be interested by what you have to offer. Those are either customers that have visited your owned digital properties or customers that you’ve acquired leveraging 3d party data for example. The uplift is significant as you can imagine compared to shooting in the dark and bidding on (potentially) unknown customers without knowing if they’ve ever shown interest in what you have to sell.
"USE CASE #2: Prospecting with modelling." What if you were able to find more of those high value customers of yours? This is the main idea behind this use case. The Look alike modelling from the DMP will simply look for audiences with similar traits to those high value customers of yours. It can do so from your existing customer base or potentially from 2nd/3d party data. Where you’ll have to do a trade off is between the accuracy of the resemblance and the volume of the new audience. The more accurate, meaning the more identical to your high value segment, the less voluminous your new audience is going to be.
"USE CASE #3: Audience suppression" This use case is a cost saving one and pretty straightforward to understand. Typically the last thing you want to do is advertising to a customer for a product he’s just bought. It will be an obvious waste of money and potentially irritating for your customers. Simply adding a customer who just purchased a product to a “no retargeting” audience and configuring your DSP (Demand Side Platform = advertiser side bidding platform) will do the trick and save you money.
"USE CASE #4: Onsite personalisation" Though a DMP’s primary use case is generally in the context of paid channels, brands get great results leveraging it to deliver a better customer experience onsite. So what does a DMP bring on top of a standard targeting solution? A targeting solution can use every signals sent by a browser to create segments and target that same segment with relevant content. The DMP add the possibility to bring 1st party data (non PII) on top of online signals to have a consolidated and actionable view of your customer.. The actionable part should include “real time” resolution of the segments, meaning if I send a signal that qualifies me for a segment, I should immediately be attached to that segment. It may sound like a given but not all DMPs allow it so make sure you ask for that specific use case when selecting your DMP.
"USE CASE #5: Cross-channel segmentation" Being able to know if a customer has interacted with your brand through call centre, app, email or website and build an audience out of that will enable you to target consistently your customers across those same channels. This will result in a much better customer experience as opposed to advertise for product A on your website and for product B on your mobile app, to the same customer. As mentioned in my previous article on "transersal digital capabilities", DMPs provide you with "Profile stitching/merging" capabilities that can link the devices IDs of a same individual together and form some kind of device graph in the back-end. This will ultimately allow you to market people as opposed to market devices..
"USE CASE #6: Remarketing" One of the primary use case for display advertising is remarketing (aka performance advertising). It consists in presenting an offer, on offsite sites (meaning on digital properties that you don’t own) to a customer who abandoned your conversion funnel and went (treacherously) looking for other products. The same way, the DMP can be integrated with RLSA for Adwords (Remarking list for search ads) or facebook custom audience for the same purpose.
The last point I'd like to make is, as usual, not to bite more than you can chew. Pick up the use cases that align best to your KPIs and priorities. Also, because it's a tool that spans across channels and hence across various teams, you'll need to get everyone onboard to build cross-channels use cases and get the data from the right teams.