Unless you have been living under a rock, you are well aware of the growing skew towards math and science in the digital media industry. We are operating in the age of the algorithm. So let’s put digital marketing technology algorithms into context, says Jason Heller on MediaPost.
So, what exactly is an algorithm?
There are debates over the actual technical definition of an algorithm, but Wikipedia has as straightforward a definition as any: “an effective method expressed as a finite list of well-defined instructions for calculating a function. Algorithms are used for calculation, data processing, and automated reasoning.”
Algorithms can be simple, or they can be incredibly complex (think Google’s search rankings). They can be static and updated manually, or they can be dynamic and “learn” based on inputs. In our case, the inputs would be observable consumer behavior and multivariate media campaign performance. We rely on powerful algorithms every day.
Since companies invest significant amounts of money, time and energy into developing algorithms that are specific to their businesses, these algorithms are heavily guarded as intellectual property, thus creating “black boxes” – their proprietary secret sauce, if you will.
Why are we obsessed with them?
Black boxes are alluring. They represent the potential of new technology that is smarter, faster and better at driving digital media performance. The vast majority of algorithmic systems are designed to optimize conversion and revenue. However, applications span across dynamic content, social engagement and beyond.
With the continued drive to increase automation and create operational efficiencies in the digital media process, black-box-driven technologies have proliferated over the last several years. There is a significant amount of VC money behind these technologies as well, providing infrastructure and increased visibility in the marketplace. Dynamic bid management, RTB, auto-optimization, intelligent profiling, dynamic creative, yield management, artificial intelligence, and predictive modeling are all powered by black-box algorithms in one way or another.
Tough decisions for marketers and agencies
The problem with a black box is that there is truly no way to prove that one partner or vendor’s algorithm is better than that of their competition. Only in-market testing can prove the efficacy for your particular category, products and objectives. So be prepared to invest in systematically testing multiple technologies, platforms or vendors.
An algorithm doesn’t work in a vacuum. There are no silver bullets; no “set it and forget it” buttons. A balance of technology and human oversight is still required to maximize digital media performance. Of course all companies pitch the best technology and a client service focus, so marketers face tough decisions on selecting partners to work with. Sometimes the best algorithms and client service teams don’t win. The best sales and marketing teams do.
Algorithms and media buying
Demand-side platforms will be enjoying an increasing percentage of digital spend for the foreseeable future. Most DSPs have real-time biddable access to the same pool of inventory and data (theoretically) but employ unique algorithms to maximize performance of each buy and campaign. For a nascent and growing category, DSPs have been portrayed as the saviors of display media. That’s one big, hairy, audacious promise.
For certain categories, buying audience has proven to be fruitful. However, for a number of clients, it’s not quite living up to the hype yet. That said, the market will certainly improve over time. It’s still early.
Maybe the lack of liquidity and high quality inventory in the ad exchanges restricts performance to certain categories. Maybe the value of certain targeting data has been overpromised. Maybe the algorithms aren’t as sophisticated as they profess to be. But maybe context, media environment and creative messaging still matter as part of the equation. Remember, DSPs and exchanges may have cut out the ad network margins, empowered buyers with RTB, provided transparency, and lowered media costs, but they didn’t magically improve the quality of what is essentially the same inventory. So we keep throwing more data and math at our problems. It can be hit or miss. Again, be prepared to systematically test different DSPs and the various data overlays available to find a model that works for your brand/product/category.
Are smart algorithms threatening to dumb us down?
As the digital marketing world continues to polarize between direct response and consumer engagement marketing, algorithms are becoming more prevalent and important across the spectrum of tools, providers, and media platforms that we use to reach, influence and engage consumers.
We all love sexy new technology. It makes for a great story. The good sci-fi movies and TV series are always the ones where the technology we create rise up against humanity. Let’s never forget the required human element. We might live in the age of the algorithm, but a great algorithm is only as effective as the strategy, brand, product, and manpower behind a campaign. Viva la algorivolucion.
Jason Heller is CEO of AGILITI, a consulting firm focused solely on client-side digital marketing operations management, helping clients foster productive relationships with their agencies and empowering clients to take control of their digital strategy. Follow him at @JasonHeller
This article republished by kind permission of www.mediapost.com https://www.mediapost.com
