By Feliks Malts, VP of Analytics
If you've been to a 3Q Digital event in the last year or so, you've likely heard us talking up the importance of DMPs (data management platforms). It's our view that today's marketing landscape is much, much more easily and profitably navigated with the right DMP and the right partner to help make good use of the treasure trove of data it provides.
That said, we’ve come across quite a number of misconceptions about DMPs that vary from what their function is, viable alternatives, cost, challenges in setup/onboarding, and if they live up to what they promise. Some of them are legacy issues, and some of them are just lack of information that has been made publically available by DMP vendors for fear of competition getting a leg up on them from a feature, partnership, and offering perspective. But some are just pure misconceptions driven from misunderstanding.
Let’s take them one by one:
Misconception # 1: A DMP just manages data/is a glorified data warehouse.
They are quite the opposite. Data warehouses are places you put data to later be retrieved on an individual record, on a transaction record. They serve the purposes of reporting, or just storing data for later use in the exact same shape, format or structure. A DMP is where you put your user data to understand and segment who your users are and ultimately use them for targeting. Targeting is the biggest piece that you’ll have missing from a data warehouse. When you put data into a DMP, you’re not expecting it to be structured in the same way and you aren’t expecting to retrieve it in the same way. You’re actually putting it in so you can learn about who your users are to better target them and people who look like them.
Misconception # 2: Most DMPs focus on serving ad buyers.
DMPs can benefit pretty much everyone in the organization from the marketer to product manager or owner through to customer service, finance, and executives. Yes, you can use these insights to buy media more efficiently, in a more targeted and personal way. But at the same time, the insights that we’re able to get out of the platform about who the users are, what they look like demographically, why they showing interest in our product, why they are buying our product, why they aren’t buying our product, and how many more people we can go out and buy, is information that will benefit everyone in your organization.
Misconception #3: I can get demographic data from Google Analytics.
Google Analytics’ methodology is very different than you see with a DMP. They are using their own data and methodologies to determine age and gender. A lot of what’s missing, even purely from a demographic perspective, is marital status, household income, home ownership, children, and educational level, amongst many, many other demographic attributes. Even outside of that, Google Analytics won’t give you any kind of financial or professional attributes, so it’s very much siloed to what they can either try to predict or learn based on some of the content networks that they serve ads on. If we had to compare Google Analytics purely by numbers to what’s available on a DMP, you can go with Google Analytics for four to five attributes whereas you can go to a DMP for four to five thousand attributes.
Misconception #4: DMPs cost millions
Truthfully, there are still vendors out there that are providing pretty big barriers to entry with hefty minimums and setup fees. However, we’ve started to see certain DMPs become a bit more lenient towards mid-market brands in terms of pricing. At 3Q, we’ve been able to negotiate an agreement that allows us to resell Lotame’s DMP at a pay-for-what-you-eat cost structure. We effectively have 3 CPMs:
- There’s a CPM for monthly unique users who go into the platform; that CPM covers the analytics component.
- There’s also CPM for targeting. So if we use the platform to deliver impressions against users, we’re paying a CPM for using the platform for targeting, and then there’s the cost of third-party data.
- The cost of third-party data is variable and driven by the data providers and the granularity/specialization of the data.
With that, we’re not imposing minimums or hefty setup fees; our clients pay for the volume they use instead of a flat fee that makes it less justifiable and approachable. If you’re big and have high volume, you pay more; if you’re small, you pay less.
Misconception #5: DMPs are impossible to onboard. They take a long time to set up, which is expensive in resource costs.
I do think this is one of the bigger misconceptions: that a DMP takes a really long time and is really expensive to set up. For us, it depends on the different components and complexity of data our clients have. We’ve had examples where we were able to do a full implementation in four days, and we have some examples that took three weeks. Implementation itself is setting up the pipes for the data and the data collection, then typically we allow two to three weeks of data to collect before we do any type of segmentation or analysis. For most of our clients, we’re up and running with some significant and meaningful insights within a month’s time.
Misconception # 6: The data isn’t actionable.
This is a big misconception we frequently hear. DMPs empower the marketer and the different parts of the business to react to what the data is saying. At 3Q, we generally put together insights around who the users are with recommendations for what we should be doing and let the client make the decision. But I’d very much disagree that the data is not actionable because in many cases you can get insights as to why your product isn’t performing well or not being purchased based on who the audience is. If you are targeting males that are 55+ and make $200K, but your visitors showing interest in the product are males who are much younger, or females, etc., you could be turning someone off from buying your product.
Misconception #7: There is a very low match rate between DMP and DSP cookie matching.
This has been less of the case with what we’ve seen. I’d say it’s the opposite. There are so many data providers that offer similar sets of data and pools; a scenario where our potential to reach 30-40x what our clients can spend is way more likely than that of us not matching a high enough number of users and having low reach. It’s likely never going to be a problem to find an audience that looks big enough and valuable enough to reach.
Misconception # 8: DMPs are blind to a good portion of media data.
This is fair, I’d say it’s 30-50% true. There are still certain gaps - not by designs of DMPs but by designs of how media networks and channels work - where we’re not able to see that someone was exposed to a message that one of our brands delivered. We can see it on the media platform, but can’t see it at the user level because we’re unable to tag the user.
An example would be, an impression in paid search, which we can’t track with a DMP. Another example would be running a lookalike targeted campaign on Facebook, which, for the same reason as previously mentioned, we can’t track. If someone is reached in an offline channel, like TV, we can’t see that with a DMP either.
There are going to be cases where we can’t see everything, but I think there are plenty of actionable insights, data, and segments we can get into, with the signals being driven by the users being engaged in these channels and then showing interest by coming to the site. I’d say these gaps are also going to be a little bit more important from an attribution perspective; a DMP isn’t an attribution platform where it matters way more to be able to collect and capture as much as possible. Here, the sample sizes are typically big enough that we can learn from the people who are showing interest and intent, which is representative of what we might not be able to see.
Misconception #9: Relying on a DMP can lead to wasted impressions, wasted money and poor ROI.
This outcome is possible if you don’t know what you are doing, target an audience that is way too broad, or don’t keep an eye on your spend and campaigns. It’s also possible if you’re not doing your research and not doing analysis of the data that’s available. However, I’d say not relying on a DMP has an even higher likelihood of leading to this result. You can waste a lot of money if you go with the advanced DSP approach, where you guess who your audience is instead of using the data. You always have the opportunity to spend more money; start smarter and more efficiently with the low-hanging fruit instead of just throwing money at it and expecting it to do magic.
If I had to sum up the industry's current view of DMPs vs. the emerging reality, it's that DMPs are more accessible and important than ever, and companies who have waved them off in the past would be wise to reconsider. As always, we're around to answer any questions, so feel free to leave a comment.
