“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 Jay Friedman, COO at Goodway Group.
All too often in today’s digital advertising industry, new agency hires are quickly handed the keys to technologies that can unleash millions of dollars in ad spending.
One of those technologies is the demand-side platform (DSP). It’s a powerful tool, in terms of how much it can spend, how much return it can drive for an organization or client and by the depth of insights that can be gleaned if marketers know where to look.
But are DSPs sometimes too powerful for the users behind them?
I’m not saying the government should license DSP users, but potential users should know what they’re getting into.
Great Performance Is Typically Hidden
Over-the-counter aspirin will address aches and pains but won’t cure anything serious. You’ll likely need a prescription medication for that. The same can be said for basic campaign optimizations within a DSP.
You can pull out bad sites, geos, browsers and devices, and also focus on variables that perform well. But any basic trading algorithm can do the same thing. A great trader within a great trading organization looks deeper at things such as recent cookie activity, frequency curves and conversion lag. Most DSPs don’t put the more complex metrics right alongside the main ones, and many don’t offer the ability to easily optimize these data points at all.
But you’ll need to optimize the complex metrics, too, if you’re to deliver the best ROIs possible. This leads me to my next point.
Every Business Has Unique Requirements No DSP Can Address
Every business runs uniquely. Different businesses value different KPIs, reporting and target audiences, even within the exact same vertical. No DSP – or software of any kind, for that matter – can be built to conform to the idiosyncrasies of every business.
If you’re an agency, multiply these idiosyncrasies by the number of clients you have. It can be quite a bit. This is why, in order to get the full benefit from using a programmatic media platform, an agency needs to become a technical and development shop as well, coding into the API of their DSP to address these specific business needs. It may be data they need for their reporting, a special algorithm they need applied to each of their campaigns or even just a workflow improvement based on their internal systems.
One Wrong Keystroke
Surely you’ve heard of (or experienced) the trader who meant to input “$10,000” but instead input “$100,000.” Hopefully they caught it sooner than later. But it’s not just financials that can get fat-fingered. For example, when hypothetically setting up a campaign on the first of the month with an end date of the 22nd, the user could mistakenly hit “2” and establish the second of the month as the end date. This can result in spending your budget really fast.
Clicking the wrong site to optimize toward or away from something can lead to a lot of wasted media money. The industry has developed significant checks and balances to ensure these things don’t happen, but DSP users should be prepared for it to happen a few times before they really learn, especially if they are trading on their own.
True Operating Costs Up To Four Times Higher Than DSP Fees
I’ve seen many organizations take their trading in-house, only to tell me six months later, “Forget it. We got half the data, half the results and half the insights for double the DSP fee.”
What does that mean? Organizations see low, transparent margin information and initially believe this is their central cost to operating a DSP in-house. As it turns out, the cost to execute programmatic media internally is two to four times more than the DSP fee alone. Basic costs may include traders, the training they require and the training to replace them if there is turnover.
To really make use of the power contained within a buying platform, agencies also likely need visualization software, training, a knowledge-sharing platform, a robust data warehouse and the engineers to operate it and a data science practice to enable insights specific to their organization.
Doing It Right
For all the scary scenarios I’ve seen, I’ve also seen things done correctly. Netflix is a great example of a company that decided to take programmatic buying in-house, but did so with eyes wide open. It knew that it would need to apply significant investments in money and time to build a practice tailored for its specific needs.
And that’s the point. It absolutely can be done. But it absolutely doesn’t have to be done for an agency or marketer to be successful. It works both ways, and whether or not a DSP is too powerful for its user largely depends on the size of the investments that user’s organization is willing to put behind bringing the capability in-house.
Follow Jay Friedman (@jaymfriedman) and AdExchanger (@adexchanger) on Twitter.
