Forrester DSP Wave Finds That ‘Differentiation Is Subtle’

Forrester’s 2017 DSP Wave ranking placed six platforms as market “leaders”: MediaMath, DataXu, Trade Desk, Turn, Adobe and Rocket Fuel.

Three others straddled the line between “leaders” and “strong performers”: Google, AppNexus and AOL. AdForm fell into the “strong performers” category. Time Inc.-owned Viant, including Adelphic, trailed the rest of the group with a placement in the “contenders” category.

To qualify for the Forrester DSP Wave, each platform needed to run $100 million in revenue in their key market, employ more than 300 people and return bids in 100 milliseconds or less 98% of the time.

Finding meaningful differences between DSPs proved a challenge, said Forrester analyst Richard Joyce, because of commoditization.

“All the vendors represent the same kind of clients, and the differentiation is subtle,” he said. “Everyone is building their own version of the same thing that clients want.”

This year, Forrester paid particular attention to the DSPs’ authenticated, people-based data assets. DSPs should be able to recognize people cross-device. If they can create universal IDs with their own user data, like Verizon-owned AOL, Google and Time Inc.-owned Viant, even better.

“Having good, accurate, authentic people-based data makes your ability to do cross-device and omnichannel a little bit better,” Joyce said.

A DSP’s inventory access factored heavily into the rankings. Buying display media as well as video, TV, social and search made the offering more omnichannel for marketers.

Adobe, for example, acquired video-focused TubeMogul, which adds to its search and paid social offerings. MediaMath, which ranked highest overall in the Forrester Wave, supports social platforms like Facebook and Instagram.

When it came to evaluating strategy, machine learning or AI scored points. Adobe Sensei, MediaMath’s Brain, Rocket Fuel’s AI marketing and DataXu’s plans for machine learning all stand to benefit marketers. AppNexus gives marketers flexibility to create their own algorithms, which could be helpful as they try to do machine learning.

Being able to predict how different media decisions would affect outcomes also got attention from Forrester.

“In the past it was about tools for forecasting impressions and spend,” Joyce said. “Now, it’s about the incrementality that might happen when you move $100,000 from one strategy to another strategy.”

The DSP wave comes as the category is changing.

Many of the DSPs in the wave have become part of broader technology offerings in the past couple of years, including Verizon-owned AOL, Adobe-owned TubeMogul, Time Inc.-owned Viant and Adelphic and Singtel-owned Turn. The Trade Desk went public last year, while AppNexus filed IPO paperwork.

Although compelling, these integrated systems take a lot of work to function as seamlessly as promised, Joyce warned. “With some of these acquisitions being fresh, it takes time to incorporate that into your unified product offering.”

But marketers do want more simplicity. “We hear from marketers that consolidation would be nice in a complex ecosystem,” he said.

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!


  1. I think the most salient point is the lack of differentiation between the DSPs. This is doubly true considering the source being Forrester – whose raison d’être is to explain differentiation.

    “Finding meaningful differences between DSPs proved a challenge, said Forrester analyst Richard Joyce, because of commoditization… All the vendors represent the same kind of clients, and the differentiation is subtle,” he said.”

    The thesis at Gimbal (full disclosure, I am the CEO) is that real differentiation can only be unlocked with a vertically integrated tech stack. When evaluating a DSP as a bolt-on technology to a 3rd party data platform, there are inherent limits – based on privacy regulations – regarding what kind of data can be used. This is especially true when using data about people. Having a fully integrated stack (DSP, DMP, Ad Serving, Creative, Data, etc) is the only way to really capture the value of the first party data of an advertiser and make it usable for targeting.

  2. Differentiation in the DSP pitches is subtle. Differences among the platforms when you’re using them are not subtle at all.

  3. Rich Joyce

    Hey Adam – that’s great perspective. Would love to learn from your experience and expertise. Feel free to send me an email or set up an inquiry call (and that goes for everyone) or Looking forward to the discussion!

  4. Agree with Adam : as a third party machine learning provider empowering DSPs & trading desks, I can say there is a huge difference between DSPs when trying to build algorithms layers on top of them ! Maybe an interesting criteria to highlight in the next study is the “openess” of the DSP: some even haven’t an API !