Response: Dave Morgan's Wrong on Math Men
Dave Morgan’s points in Digiday yesterday struck me as a lamentation of how poor of a job the technology and math folks do in clearly translating what they do and why it matters. He’s right, it’s true. We don’t do it in sufficiently straightforward terms to be broadly accessible by people with real work to do other than unpack acronyms for a living. But Dave’s not right when he says the space isn’t differentiated or is relying on science too much. We’re not going to let Dave forget his pioneering days in digital at Real Media and Tacoda in favor of the relative simplicity of TV: we’re going to dive in and solve the problem together. Here’s a (relatively) quick primer on what’s going on:
The demand-side platform is used by, well, the demand side — agencies, advertisers. It is the "operating system" for a marketer’s efforts. Generally speaking all of the major DSPs (Google/Invite, MediaMath, Turn, etc.) share broad reach across supply, the ability to layer in first-and third-party data (think basic DMP capabilities — more on this later) and the ability to execute media and audience campaigns across sources in real time.
Morgan said he can’t see how they differentiate, but they certainly do. Google’s Invite relies increasingly on integration with DoubleClick products for a vertically integrated stack. MediaMath focuses on integration with other exchanges, supply-side platforms, premium publishers, data companies, and the like, along with a core strength in the ability to decide in real-time, based on sophisticated response and look-alike models. Turn has perhaps the most user friendly interface.
Think of the DSP as the “on-ramp” to the rest of the marketing stack. Or for yet another metaphor: think of building a successful digital marketing program like building a house. The advertiser is the prospective homeowner. The agency is the architect, translating the broad vision into a blueprint. The DSP is the technology and math-powered general contractor that stitches together all of the other elements necessary.
The supply-side platform (leaders being AdMeld, Pubmatic, Rubicon) does the same thing for the supply side — that is, publishers. It helps them manage their yield across multiple different buyers with different objectives, pricing models and sales channels. They have been increasingly bundling third-party data in as well, to allow publisher-sales forces to profit from their data outside of their media.
The data management platform took a slice of the DSP and SSP stack, carved it out and have been delivering it to both advertisers and publishers. A good DMP will allow a marketer to custom define audiences, forecast how many of those audiences they have by channel and provide useful data/audience specific reporting. They can often also help publisher ad sales teams sell audience packages that don't exist typically on their site — it's why you might see "hip urban kids" available on your Forbes.com plan.
Pure-play DMPs are in my opinion a short-term class of entity. They're features, not companies, as they don't do much by themselves in terms of executing, optimizing or monetizing media or data. Most will get bought by DSPs, SSPs, exchanges, or the same functionality will be built by those entities. Look at Adobe scooping up Demdex. I am by no means saying that they are not valuable – they stepped in to fill a market need left vacant by ad servers and site analytics firms – but long-term independent DMPs are unlikely. For reference, one could contrast this with the likes of a BlueKai, for example, which between its DMP, data exchange, and Tracksimple analytics package, is a viable stand-alone for those buyers who strongly prefer two separate platforms.
Lastly, exchanges, like their namesake financial exchanges, are ostensibly neutral, low-fee platforms to help buyers and sellers transact. Right Media's exchange historically represented networks buying and selling from one another, although it is moving decidedly upmarket to support premium publishers with its 5to 1 acquisition, with AppNexus serving many of the original Right Media networks on an upgraded RTB platform and more aggressive technology investment. Google’s AdX is growing quickly, albeit with more long-tail and "semi-transparent" inventory from the AdSense/Google Display Network legacy. Data exchanges, like BlueKai’s and Exelate’s, similarly aggregate data from publishers and other data sources.
Who you talk to first will depend on your primary objective: are you generally trying to acquire customers or sell eyeballs? A DSP will be the right choice if you are a “normal” agency/marketer and an SSP will serve if you are a “normal” publisher/portal. A network will likely be best suited by an exchange (or, obviously, both a DSP and an SSP).
From there, you can start to get fancy. For example, if you think you might have rich data assets and your needs scale beyond the capabilities of your DSP or SSP, then work with them to identify a DMP that would make the most sense to partner with based on your other needs like tag management, analytics, or data monetization which answer would help determine which of Aggregate Knowledge, BlueKai, BrightTag, Exelate, or TagMan made the most sense.
The AMP category is a made-up one that people came up with to annoy you and to see if the VCs were paying attention. I did hear someone claim the AMP was about building custom audiences and then allowing for media execution of those audiences. But it should be obvious that this is just a core capability of the DSP and, for complex use cases, the DSP+DMP integrated. So I propose just returning the AMP moniker back into the soup from which ad technology is sprung before being shaped into products, with the first manifestation on the road to reality always being an acronym and three PowerPoint slides.
I understand Dave Morgan’s point that it’s a reasonably complex marketplace, but hopefully we’ve highlighted that for each job in this crazy world, there’s an obvious first point of contact that has bundled the key features you’ll likely need, while remaining flexible and modular enough to adapt to unique requirements and evolving objectives. TV is purchased based on Nielsen audience GRPs. These ratings points are a proxy based on a proxy. Everyone knows it, and everyone agrees to use it. That doesn’t make it correct. But being correct is far more difficult. It requires some heavy, mathematical lifting. DSPs do more and more of that lifting for buyers every day – and it will transform how digital is bought today, and, yes, how TV is bought in the future.
Joe Zawadzki is CEO of MediaMath, a demand-side platform.
About The Author: MediaMath
More posts by MediaMath