Is The DMP Finally Dead?

Data-Driven Thinking" is written by members of the media community and contains fresh ideas on the digital revolution in media.

Today’s column is written by Martin Kihn, research vice president at Dentsu Aegis Network.

Pity the poor data management platform (DMP).

After almost a decade since the programmatic power-up, it has been all but left for dead as pundits hang crepe and sing sad songs, wondering where it all went wrong.

Invented almost by accident 10 years ago to manage audiences for ad targeting, the DMP became the center of the programmatic circus. Then it claimed the core of the marketing clouds, as Adobe, Oracle and Salesforce each acquired one.

And, as recently as last year, a survey of tech stacks found more than 50% DMP adoption among enterprise marketers. Only 10% of respondents felt they didn’t need one.

DMP, we hardly knew you

Times change. Beset by forces (mostly) beyond its control, the once-mighty DMP finds itself a victim of second-guessing and the kind of anti-hype that happens to all good tech as it ages.

Now industry commentators say “marketers sour on DMPs.” They claim Duracell abandoned its DMP and others are “using GDPR to exercise break clauses” to “save face” with their CFOs.

Others are saying that anonymous third- and second-party data is “almost dead” – at first, in Europe, due to consumer backlash and regulation. To the extent the DMP’s value relies on this data, it falls into the dreaded spiral. We can hear the dirge from Düsseldorf.

Certainly, DMPs overpromised – or marketers overexpected – and there is a natural recalibration. They were not built to do social advertising or onboarding or to manage customer files. Digital marketing has changed since 2008, but so have DMPs.

It is harder to answer the question “Do I need a DMP?” today than it was last year. But it is still not an easy yes or no.

Three reasons why the DMP is (not) dead

Here are the most common DMP death rattles I’ve heard, along with three signs of life:

1. Death rattle: DMPs rely on cookies, and cookies are dead; therefore, the DMP is dead.

The third-party browser cookie is less useful each year. Adding up time in apps, Safari browsers, ad-blocked browsers and cookie rejecters, most consumers are out of reach most of the time. GDPR and other regulations, requiring opt-in rather than opt-out, restrict their value further.

Sign of life: Cookies are not dead; first-party and session cookies are needed to “maintain state” in browsers.

Third-party trackers may go away, but in-app IDs, such as AAID and IDFA, are better anyway. Audiences can be built offline and onboarded. And audiences can be built using probabilistic signals such as location, which doesn’t require cookies. DMPs have been augmenting cookies with their own identity layers for years now anyway.

2. Death rattle: More and more media budget is going to Facebook, Google, Amazon and other proprietary ecosystems that don’t need outside DMPs to work.

Whether this is a good argument or not depends on your media plan. If you are spending 80% of your media budget on Facebook and YouTube, it’s hard to justify a DMP to optimize the other 20%. It’s back to the future: Set up separate “platform” teams to run each ecosystem and ignore the programmatic open web.

Sign of life: There is too much value in the open web to ignore.

Don’t give up. Rage against the dying of the light. If you’re spending 80% of your budget on Facebook and YouTube, and you’re not an SMB, you have bigger problems. Running platforms in parallel makes it difficult to optimize across those platforms. Media-mix models are slow.

3. Death rattle: First-party data is more accurate and valuable, so you’re better off focusing on your own customers and using a customer data platform (CDP) rather than a DMP.

CDPs are very hot right now, and they offer definite benefits: managing customer data and building coherent omnichannel experiences. It is undoubtedly smart to link pockets of customer and prospect information and analyze it. There is a lot of hype, and where there’s hype there’s hope, right?

Sign of life: CDPs and DMPs are fundamentally different, so it’s wrong to frame the choice as either/or.

CDPs are more like modern CRM, and DMPs are made for advertising. There is some overlap, since DMPs also collect first-party data for things like retargeting, triggering email and even – yes – using customer IDs as their primary key. Likewise, CDPs can be used to build lists for social ad targeting. But your strategy determines whether you need one or the other, or both.

It’s alive, it’s really alive

Every change is a kind of death, and the DMP has changed. In fact, it’s hard to find a pure-play, standalone DMP anymore. They are enmeshed within other solutions and services and come in a half-dozen forms: There are DSP-DMPs like MediaMath, vertical-focused DMPs like Adara, white-labeled DMPs like IPONWEB’s and services-led DMPs like Accordant Media and Amnet.

Even the marketing cloud DMPs are all different and likely to become more so over time. Adobe Audience Manager provides an identity layer for its Marketing and Ad clouds; Salesforce claims it improves on CDPs; Oracle and Neustar stress data and identity services, respectively. Even Google, which doesn’t market a standalone DMP, has mentioned the “DMP-like capabilities” of its Ads Data Hub.

Marketers should know what DMPs do and don’t do by now. They should know about retargeters and personalization platforms, if that’s what they need. Expectations are more realistic, and we know that everything worthwhile is difficult.

Meanwhile, the DMPs themselves need to keep up with the market. They are not all improving at the same rate. They promise control and better measurement but compete with people-based platforms that have their own attractive stories.

So, is the DMP dead? Only if you want it to be.

Follow Martin Kihn (@martykihn), Dentsu Aegis Network (@dentsuaegis) and AdExchanger (@adexchanger) on Twitter.

 

Add a comment

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>