There is no easy button in the demand-side-platform world.
One of the most frequent questions we hear in the demand-side-platform world goes a little something like this: "Where is the easy button?" Or, in other words, "Where do I click to auto-optimize the campaign?"
Some DSPs are actively encouraging this mentality, which we call "Black Box Optimization," or, "Don't Worry, The Algorithms Will Do All The Work For You."
Essentially, some players in the DSP world like to tell customers that there's an easy solution out there that will address the huge amounts of data and save us all a bunch of time. It sounds great in a sales pitch, or a PowerPoint presentation.
Today, I'm going to call "BS" on this concept, and suggest you take a look under the hood before betting your agency's future on complete black box optimization. Here's why:
There is no such thing as auto-optimization.
The analogy we often use has to do with the stock market. Let's say I logged into my E-Trade account this morning and decided I wasn't happy with my mutual fund returns this year. If I could click a button that said, "Make me a 60 percent return on my money in 2011," I would certainly click that button. Then, I'd enter the mailing address of my beach-side cabana, so my trading platform would know where to send my returns.
The problem is that auto-optimization doesn't exist in the stock market world any more than it does in the media-buying world. Making consistent returns from the stock market is hard and takes human judgment. The same goes for buying media.
Now, this is not to say that black box components and auto-optimization do not and should not exist. They should and they do. Just like high-frequency trading couldn't exist without algorithms, they have their place. However, like in all trading environments, people are pushing buttons, making decisions, and active in the "optimization" process.
If there was an easy button, most agencies and media buyers would not have jobs left.
What's really in the black box?
All DSPs worth their salt use algorithms. When you test-drive a DSP, or see a demo, you should be asking about specifics. What variables are being considered? How does the DSP know that a variable - let's say contextual category - is enough of an independent variable to be used for optimization? What variables are not part of the algorithm, and why not? Those are all thoughtful, insightful questions, and questions where true partners welcome a conversation.
But the reality is that some black box-centric DSPs overdo things on the marketing front.
Let's say my company develops a unique approach to let a media buyer bid more or less for media based on the weather in the customer's local region. I can guarantee that within a week or two, there are some DSPs out there who will start telling their agencies, "Oh, that weather thing? Yeah, that's covered by our black box algorithms," as their engineers scramble to hack together a weather-based solution. How will you ever know that they aren't truly accounting for weather?
For that matter, in a black box environment, how do you replicate success and fix failure at all?
Let's say an agency begins using a DSP with an auto-optimization button. And let's imagine that for some campaigns, the algorithms do their job and the campaign performs well. What does the agency tell the advertiser? When the agency wants to go do another ad buy with an ad network, or with a direct publisher, or a video buy, what has the agency learned? Nothing. The DSP owns all of the insights, and the agency is left high and dry.
Or, if the campaign doesn't perform, what can the agency do to fix it? It's a lot like seeing the engine warning light come on in a new car. Just like drivers today are powerless to fix a highly sophisticated and computerized engine system, without a degree in mathematics, how does a media buyer attempt to fix their campaign?
Bottom line: there is no question that people still play a huge role in the media-buying process, and aren't in danger of being replaced by algorithms anytime soon. The best DSPs in the market merge algorithms and automation with a human touch. Optimally there is a dialogue between technology and media buyer, where machines focus on correlation, and humans operate as judge and jury by focusing on causation and the big decisions.
In the future, the best media buyers will be packing more than an easy button.
Jeff started his Internet career working for a small interactive agency, where he led media buying and trafficking and managed all vendor relationships. Afterward, he founded a CPA network, eBound Strategies whose technology was designed by Jeff and later acquired by Nami Media in 2003. Jeff worked for Nami Media as the VP of operations, and subsequently left to found AdECN, the first exchange for online advertising.
Jeff is considered one of the few pioneers of the ad exchange. As COO and founder of AdECN, he led all strategy, product, and business development. Jeff is a thought-leader in real-time bidding technology. At Microsoft, Jeff oversaw AdECN exchange business, ran all reseller and channel partner business, as well as advised the all-up strategy for the online services division. In October 2009, Jeff left Microsoft and AdECN to found The Trade Desk.
2015 Holiday Email Guide
The holidays are just around the corner. Download this whitepaper to find out how to create successful holiday email campaigns that drive engagement and revenue.
Three Ways to Make Your Big Data More Valuable
Big data holds a lot of promise for marketers, but are marketers ready to make the most of it to drive better business decisions and improve ROI? This study looks at the hidden challenges modern marketers face when trying to put big data to use.
December 2, 2015
1pm ET/ 10am PT
Wednesday, December 9, 2015
5pm HKT / 5am ET