If you’re like me, and have succumbed to the unpardonably bourgeois luxury of hiring a cleaner, then you may also have found yourself running around your house before the cleaner comes, picking up stray items of laundry and frantically doing the dishes. Much of this is motivated by “cleaner guilt”, but there is a more practical purpose – if our house is a mess when the cleaner comes, all she spends her time doing is tidying up (often in ways that turn out to be infuriating, as she piles stuff up in unlikely places) rather than actually cleaning (exhibit one: my daughter’s bedroom floor).
This analogy occurred to me as I was thinking about the experience of working with a Data Management Platform (DMP) provider. DMPs spend a lot of time coming in and “cleaning house” for their customers, tying together messy datasets and connecting them to digital marketing platforms. But if your data systems and processes are covered with the metaphorical equivalent of three layers of discarded underwear, the DMP will have to spend a lot of time picking that up (or working around it) before they can add any serious value.
So what can you do ahead of time to get the best value out of bringing in a DMP? That’s what this post is about.
What is a DMP, anyway?
That is a excellent question. DMPs have evolved and matured considerably since they emerged onto the scene a few years ago. It’s also become harder to clearly identify the boundaries of a DMP’s services because many of the leading solutions have been integrated into broader “marketing cloud” offerings (such as those from Adobe, Oracle or Salesforce). But most DMPs worth their salt provide the following three core services:
Data ingestion & integration: The starting place for DMPs, this is about bringing a marketer’s disparate audience data together in a coherent data warehouse that can then be used for analytics and audience segment building. Central to this warehouse is a master user profile – a joined set of ID-linked data which provides the backbone of a customer’s profile, together with attributes drawn from first-party sources (such as product telemetry, historical purchase data or website usage data) and third-party sources (such as aggregated behavioral data the DMP has collected or brokered).
Analytics & segment building: DMPs typically offer their own tools for analyzing audience data and building segments, often as part of a broader campaign management workflow. These capabilities can vary in sophistication, and sometimes include lookalike modeling, where the DMP uses the attributes of an existing segment (for example, existing customers) to identify other prospects in the audience pool who have similar attributes, and conversion attribution – identifying which components of a multi-channel campaign actually influenced the desired outcomes (e.g. a sale).
Delivery system integration: The whole point of hiring a DMP to integrate data and enable segment building is to support targeted digital marketing. So DMPs now provide integration points to marketing delivery systems across email, display (via DSP and Exchange integration), in-app and other channels. This integration is typically patchy and influenced by other components of the DMP provider’s portfolio, but is steadily improving.
Making the best of your DMP relationship
The whole reason that DMPs exist in the first place is because achieving the above three things is hard – unless your organization in a position to build out and manage its own data infrastructure and put some serious investment behind data integration and development, you are unlikely to be able to replicate the services of a DMP (especially when it comes to integration with third-party data and delivery systems). But there are a number of things you can do to make sure you get the best value out of your DMP relationship.
1. Clean up your data
This is the area where you can make the most difference ahead of time. Bringing signals about your audience/customers together will benefit your business across the board, not just in a marketing context. You should set your sights on integrating (or at least cataloging and understanding) all data that represents customer/prospect interaction with your organization, such as:
- Website visits
- Product usage (if you have a product that you can track the usage of)
- Mobile app usage
- Social media interaction (e.g. tweets)
- Marketing campaign response (e.g. email clicks)
- Customer support interactions
- Survey/feedback response
You should also integrate any datasets you have that describe what you already know about your customers or users, such as previous purchases or demographic data.
The goal here is, for a given user/customer, to be able to identify all of their interactions with your organization, so that you can cross-reference that data to build interesting and useful segments that you can use to communicate with your audience. So for user XYZ123, for example, you want to know that:
- They visited your website 3 times in the past month, focusing mainly on information about your Widget3000 product
- They have downloaded your free WidgetFinder app, and run it 7 times
- They previously purchased a Widget2000, but haven’t used it for four months
- They are male, and live in Sioux Falls, South Dakota
- Last week they tweeted:
Unless you’re some kind of data saint (or delusional), reading the two preceding paragraphs probably filled you with exhaustion. Because all of the above kinds of data have different schemas (if they have schemas at all), and more importantly (or depressingly), they all use different (or at least independent) ways of identifying who the user/customer actually is. How are you supposed to join all this data if you don’t have a common key?
DSPs solve these problems in a couple of ways:
- They provide a unified ID system (usually via a third-party tag/cookie) for all online interaction points (such as web, display ads, some social)
- They will map/aggregate key behavioral signals onto a common schema to create a single user profile (or online user profile, at any rate), typically hosted in the DMP’s cloud
The upside of this approach is that you can achieve some degree of data
However, there are some downsides, also. Key amongst these are:
Yet another ID: If you already have multiple ways of IDing your users, adding another “master ID” to the mix may just increase complexity. And it may be difficult to link key behaviors (such as mobile app purchases) or offline data (such as purchase history) to this ID.
Your data in someone else’s cloud: Most marketing cloud/DMP solutions assume that the master audience profile dataset will be stored in the cloud. That necessarily limits the amount and detail of information you can include in the profile – for example, credit card information.
It doesn’t help your data: Just taking a post-facto approach with a DMP (i.e. fixing all your data issues downstream of the source, in the DMP’s profile store) doesn’t do anything to improve the core quality of the source data.
So what should you do? My recommendation is to catalog, clean up and join your most important datasets before you start working with a DMP, and (if possible) identify an ID that you already own that you can use as a master ID. The more you can achieve here, the less time your DMP will spend picking up your metaphorical underwear, and the more time they’ll spend providing value-added services such as audience extension and building integrations into your online marketing systems.
2. Think about your marketing goals and segments
You should actually think about your marketing goals before you even think about bringing in a DMP or indeed make any other investments in your digital marketing capabilities. But if your DMP is already coming in, make sure you can answer questions about what you want to achieve with your audience (for example, conversions vs engagement) and how you segment them (or would like to segment them).
Once you have an idea of the segments you want to use to target your audience, then you can see whether you have the data already in-house to build these segments. Any work you can do here up-front will save your DMP a lot of digging around to find this data themselves. It will also equip you well for conversations with the DMP about how you can go about acquiring or generating that data, and may save you from accidentally paying the DMP for third-party data that you actually don’t need.
3. Do your own due diligence on delivery systems and DSPs
Your DMP will come with their own set of opinions and partnerships around Demand-side Platforms (DSPs) and delivery systems (e.g. email or display ad platforms). Before you talk with the DMP on this, make sure you understand your own needs well, and ideally, do some due diligence with the solutions in the marketplace (not just the tools you’re already using) as a fit to your needs. Questions to ask here include:
- Do you need realtime (or near-realtime) targeting capabilities, and under what conditions? For example, if someone activates your product, do you want to be able to send them an email with hints and tips within a few hours?
- What kinds of customer journeys do you want to enable? If you have complex customer journeys (with several stages of consideration, multiple channels, etc) then you will need a more capable ‘journey builder’ function in your marketing workflow tools, and your DMP will need to integrate with this.
- Do you have any unusual places you want to serve digital messaging, such as in-product/in-app, via partners, or offline? Places where you can’t serve (or read) a cookie will be harder to reach with your DMP and may require custom integration.
The answers to these questions are important: on the one hand there may be a great third-party system with functionality that you really like, but which will need custom integration with your DMP; on the other hand, the solutions that the DMP can integrate with easily may get you started quickly and painlessly, but may not meet your needs over time.
If you can successfully perform the above housekeeping activities before your DMP arrives and starts gasping at the mountain of dishes piled up in your kitchen sink, you’ll be in pretty good shape.
3 thoughts on “Got a DMP coming in? Pick up your underwear”
Loving it, we are spending so much time in cleaning up and reorganizing data from our client instead of spending most of our time in scoring and bringing machine learning to improve performances. Brillant article.
Merci Laurent! I think this kind of data clean-up is going to become more and more important in the future, as organizations capture ever wider sets of data about their audience/customers.
Comments are closed.