“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 Jeremy Hlavacek, head of global automated monetization at IBM Watson Advertising.
We have all met people who love driving a manual stick-shift car. They will tell you how much they love “feeling the engine,” “the sense of control” and “being in tune with the car.” The language used is often highly emotional and evokes a human connection with a machine. They will also tell you that automatic cars are “less fun to drive,” and they will focus on a few limited use cases where the manual shift works better than the automatic.
On the other hand, automatics dominate the US new car market, and it seems that automakers are leaning in to further refine and improve them. Having recently bought a newer car, it’s clear that auto companies are trying to add more gears and create software and intelligence to achieve both maximum full efficiency and driving performance.
I think this is the perfect analogy for the crossroads that the marketing industry is facing right now when deciding to how to deploy today’s incredibly powerful data sets.
In one camp, there is a belief that there is too much automation mixing with powerful data, leading to bad targeting (“creepy” ads), illogical measurement (cookie bombing) and potentially catastrophic brand failures (contextual ad disasters).
Marketing seems out of control, and the solution from this group is to bring more human touch to the handling of data, which may slow processes down but is ultimately a small price to pay for the concerned marketer.
On the other side, fans of the almighty programmatic algorithm and its rapidly ascending successor, artificial intelligence, believe that no one can match the raw performance that a smart machine can deliver when it has the right driver and is fed good data gasoline. These folks will admit that algorithms make mistakes sometimes, but they expect that number to shrink over time because the inevitable march of progress in processing behaves almost like Moore’s law.
The Data Market Today
Before deciding between a manual or an automatic, it is important to understand exactly what is fueling this debate. The answer is data. I believe the development of the digital data marketplace in the last 10 years is nothing short of revolutionary.
This shift began on the buy side of the business as marketers discovered the power of data captured in browser-based cookies. More precise targeting afforded by cookies led to more relevant – although sometimes more annoying – ads that clearly delivered better performance. Simple tactics, such as retargeting, became cornerstones of media plans and it didn’t take long for media agencies and ad tech companies to get creative with cookie data and start building “segments.” Did you spend the afternoon browsing Cars.com? You’re an “auto intender.” What if you checked out some sneakers on Zappos? You are “in-market” for shoes.
Today these would be considered rudimentary examples of data targeting, but after making heavy investments in data science teams and machine learning technology, marketers and their agencies are mixing and matching all sorts of data sets or “events” and running the complicated analyses on powerful computers to look for “signals.” This process is happening very quickly, and it is leading to both great successes and embarrassing failures.
Although this space is still in early innings, it is clear that marketers now believe that the right data and machines makes it possible for them to send a real-time digital message to a person or “identity” that has some associate characteristics. And depending on what happens next in the measurement results, the marketer’s machine might change the campaign on an impression-by-impression basis to influence the consumer’s behavior and drive a business outcome.
This is an incredible transformation in the speed and power of traditional marketing processes that cannot be overstated.
What To Do?
So, what should a marketer do? Revert to a less automated approach so they can “feel the campaign” and be more “in tune with the consumer”? Or embrace the power of algorithms and AI, even with its risks?
In my view, there is no compromising on this question – marketers must embrace the future. There is no going backward, and my opinion is that there is a sort of Moore’s law effect in the continuous progress of data and algorithms: They are get faster and better at an increasing rate.
Marketers need to learn how to drive these new systems today. If they revert now, it will be harder to catch up in the future, and they will be stuck piloting a horse and buggy on a Formula One racetrack.