Let’s Ditch Combo DSPs And Bake-Offs In 2016

jayfriedmannewData-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 Jay Friedman, COO at Goodway Group.

Long ago there were only small specialty stores. Now we have Walmart, where we can buy tires, a pair of shorts, fertilizer and steaks for the grill, all in one place. If you prefer, you could do all of this at Target or Meijer as well.

For some, this product aggregation across all categories is the ultimate convenience and provides significant benefit. This is the core of what a demand-side platform (DSP) does: The best DSPs tie inventory from mobile, video and display together for unified access.

But what if someone built a new one-stop retail shop, where shoppers could get everything from Walmart, Target and Meijer all at one place? I don’t think that sounds all that great – pretty unnecessary, actually.

Yet this is a trend we’re seeing at some agencies and brands, aggregating multiple DSPs under one umbrella as a “combo DSP,” or similar names.

Our industry spent 15 years aggregating online inventory, from publisher networks to ad networks to exchanges and finally DSPs. Now it’s all in one place – the aggregation is done. Why the desire to keep going? I think the answer goes back to Roman times and ties us into our second topic.

Ancient Rome had pits where men would fight each other or animals to the death with a cheering crowd in the background. Well, that’s not too PC today, but a DSP bake-off? That’s the next best thing. I actually spoke to someone from a large brand who told me it pitted eight DSPs against each other “bracket style” and was proud of it.

Whether money is being spent in meta fashion with good intentions or bracket style with really confusing intentions, this is making your spend less efficient. Here’s why.


This one is so obvious I feel silly saying it. The most basic elements of marketing are message, targeting (reach) and frequency. It’s hard enough to get frequency right in a single platform. It’s downright impossible when working across multiple platforms within a single campaign.

Marketers shouldn’t leave to chance something that represents one-third of a campaign’s effectiveness!

Last Ad Seen Or Clicked?

I’m shocked at how many brands and agencies still work off the last ad seen or clicked (LAS/C). The single largest digital priority in 2016 for CMOs of brands operating this way should be to get any attribution model baked into their digital media.

If CMOs are in bake-off mode, it’s highly likely they are using LAS/C to judge success. It’s also likely each DSP is also using LAS/C within its own platforms – double bad. I’d posit CMOs could double their digital marketing efficiency just by using a single platform and any reasonable attribution model within the DSP, not just cross-channel.

Strategic Partner Or Vendor Manager?

Has an agency or marketing department fallen into the “vendor manager” trap rather than being a strategic partner? There are a few easy signs to see if this is happening.

First, look for agency presentations that are littered with vendor logos, essentially using their credibility to supplant their own.

Another indicator: The agency or marketing group derives value from constantly testing and using bake-offs to determine who is best. The agency or marketing group may also treat vendors like 50-cent T-shirts. If there is a stain on the relationship, they toss the vendor rather than work through the problem.

The CMO needs strategic marketing thinking. Of course they need to allow, coach and encourage their marketing team and agency to perform that role.

Digital media is the most measurable form of advertising we’ve ever known. It’s also an easy medium to measure incorrectly. Our industry is at a point now where more media than not is measurable down to exact desired effect. If marketers focus on this with a few intelligent strategic partners, their media will work harder for them than any of their competitors.

Follow Jay Friedman (@jaymfriedman) and AdExchanger (@adexchanger) on Twitter.

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. Obviously this is merely an opinion but not one I share. Some observations.

    Managing multiple dsp’s together is little different in practice to multiple line items in one dsp when it comes to frequency. Add in a common dmp,attribution platform proper programatic planning and the fact that you need to add to a digital plan walled gardens, direct buys search and social and the frequency argument disappears to absurdity.

    No one dsp is perfect or complete. Tie yourself to one and like big supermarkets you are doomed mediocrity.

    A small media plan with limited budgets and your argument holds. However it is not best practice in 2016.

    • I would argue the opposite Rob. I would say that part of the evaluation process for any DSP should be it’s capacity to integrate many inventory sources, ad units and data/measurement partners. The whole point behind programmatic buying is eliminating the exact fragmentation you are describing as ‘ideal’ and ‘necessary’.

      Evaluating DSPs is different then an ongoing and forever bake off that is adding both unnecessary redundancy, collision and waste.

      • When did I say it was ideal and necessary?

        Fragmentation exists whether we like it or not and arguably back on the rise with new walled gardens. I agree this is a shame but the world is how it is, not how we wish it to be.

        Also different DSP’s compete on more than just inventory. Data access, creative tie ins, bidding capabilities, devices (cross device). We live in a never ending industry of innovation that limiting yourself to one DSP closes.

        In some cases one DSP is the correct solution. My point is that this article is wrong to say that this is correct in all circumstances. Indeed I maintain that for large advertisers, for the reasons I mention multiple DSP’s is the right answer for 2016.

  2. Odd argument. If advertising is more measurable than ever, then the analysis of agency/platform/vendor performance is easier than ever before. Why wouldn’t I aways make sure that my agencies/vendors/partners weren’t performing to my benefit? (Just because one holds a bake-off, doesn’t mean the advertiser needs to change vendors. And I do agree with the rush to toss away agencies too quickly–best to mend relationships then form new ones, when possible.)

    Finally, wild disagreement that deriving value from constant testing is a bad thing. Science and scientific advertising is about constant testing, and testing of everything, including agencies/vendors/partners.

    • Mark – Thanks for allowing me the clarify. Constant testing is important. No question. THE WAY most agencies and advertisers do the testing is the problem. Last ad seen/clicked (90% of bake-offs) is incredibly easy to game and a giant waste of the client’s money.

      Since testing on separate campaigns isn’t apples to apples, most agencies and clients throw two DSPs (or networks, or exchanges or whatever) in a pit to duke it out. Not only do the agencies/advertisers do a bad job of setting up the rules, they do a worse job of policing them. If these competitions are intelligently designed I agree they can be good, but an intelligently designed bake-off is a good theory that rarely pans out in reality.

  3. Justin Gospodarek

    Are these t-shirts that cost 50 cents, or t-shirts of 50 cent?

    On a serious note…

    It’s true, some people make it too complicated just for the sake of impacting the smallest efficiency. To Rob’s point, it’s actually a great thing if you know what you’re doing. You’ll eventually learn the strengths of your partners and be the first to know when something is on a climb.

    It’s not moot, but it’s definitely situational. It’s up to the advertiser or agency to know where they stand.

  4. I couldn’t agree with you more, Jay… why have two DSP’s competing against each other to buy your ads?

  5. Man, so many gems in this post!

    Definitely something I encounter: an empty “testing for testing’s sake” mantra that apparently precludes actually fine-tuning anything to modify the outcome of the test. I chalk it up to a combination of short-term thinking, laziness (or lack of bandwidth), and lack of subject mastery on the tester’s part. An oversimplified pass/fail system is a lot easier to execute than thinking deeply about things like experiment design, troubleshooting, tech stack touchpoints, and campaign strategy. If all you’re trying to do is dazzle your superiors or clients (who in turn are under pressure to dazzle their bosses), then a rapid-fire bake-off will yield plenty of fodder to create the illusion of analytical thinking and added value.

    Obviously testing needs to play a big role in campaign development, but I agree, in many cases I see it regarded as a rote checklist that absolves the tester of any obligation to invest further thought or effort.

  6. Accessing multiple DSP’s has more pro’s than cons highlighted in this article. Frequency – yes can be questioned, but how do you know one DSP is better than the other? Most brands that I work with measure success based on ROI. We access multiple DSP’s because different DSP’s perform differently for different objectives. Not any one DSP consistently performs better than the other. DSP’s are built differently, some focus more on developing best in market video solutions (TubeMogul) whereas some DSP’s are built for display performance, retargeting, social, etc. Google’s DSP solution DBM allows you to integrate data from other Google platforms such as Adwords, DCM (ad-server, etc) and there is a lot of benefit being to able to access your search data across a buying platform. However you cant integrate your search data into the TURN platform. However, TURN offers other targeting capabilities not offered in DBM and therefore can be used on a strategic level based on objectives. Without having had access to multiple DSP’s we wouldn’t have been able to deliver many of our campaigns nor perform. We constantly shifted spends between multiple DSP’s to maximise our return. At the end of the day, thats what’s its about – ROI for brands.

  7. Jay, what you fail to mention, after deriding current measurement methods, is what should be used as a proper signal for media success. I agree there are platforms that provide access to the most relevant inventory and targeting data across devices, but that’s not the end of the story is it? Some of these systems (as you well know being at a “Google Shop” is that when different ad tech pieces are cobbled together as opposed to being built to function together as a single unit, you are still buying disparate systems under a single invoice. For the length of your article I was waiting for you to get to this point, and will make it for you since you stopped leaving me waiting – What you need to look for in a DSP to ensure the tech is aligned with your financial interests as an advertiser are: 1.) Platform doesn’t own any data or media – this is a conflict of interest as they will always sell you their own first and are not aligned with your goals 2.) can see and measure conversion across devices 3.) cross channel fraud mitigation 4.) Measure success in terms of sales revenue or as close as the advertiser can get to that.

    • I agree with your 4 points. I feel like the point I wanted to make was on a different plane than what you’re saying. This article wasn’t about how to choose a DSP but instead about not mindlessly rotating DSPs in and out when it’s impossible to fairly judge who a winner is because the way to value those metrics aren’t right to begin with.