Why marketers need to learn to trust the machines by Jason Trout
City A.M. previously published an article titled “Why marketers need humans behind the programmatic steering wheel”.
The author highlighted how automation has transformed the marketing industry, and argued that for marketers to achieve successful automation, humans need to lead as machines lack the ability to predict or detect subtle changes or circumstances that make consumers respond to ads.
It also argues that ads that are designed solely by machines won’t perform as well due to their inability to produce creative ideas.
Whilst humans undoubtedly need to be part of the ecosystem, machines deserve much more credit within the digital marketing process. Of course the technology isn’t perfect and if used incorrectly, will result in negative results.
There is good and bad practice and most issues surrounding poor creativity, poor context and poor content are the result of bad practice. Whilst the technology may not be perfect, it does have its undeniable benefits for marketing professionals.
Re-targeting your audiences
Re-targeting continues to present a significant challenge for marketers. On one hand, it is universally accepted that re-targeting customers who have expressed interest in a particular product is beneficial.
“Machines may not be completely accurate all of the time but due to amount of pages they process, they are vastly more reliable for making judgements”
But one of the most frequent complaints when it comes to adtech is the ‘pair of shoes that follows me around the internet’ and there have been incidents in which customers are targeted with the same ads even when the product has already been purchased.
Of course, this can be irritating for the customers who are essentially being spammed with the same messages. Nevertheless, it’s important to remember that it is not the technology at fault.
It is simply doing as it has been instructed. This means that either the company that provides this product-level retargeting hasn’t bothered to set up effective frequency caps or their statistics show the ad does yield a better-than-average return regardless of whether it is irritating a small number of people in the process.
Can machines make sound judgements?
Ensuring ads don’t appear against the wrong kind of content would seem like something humans would outperform machines at. The ability to understand nuances of a page and the context of recent events to judge whether an ad should appear seems an inherently human task.
However, this is not the case. When operating at internet scale, automated page-level processing is the only way to understand what each page is about and handle the content accordingly.
Humans can, and do, make mistakes. Machines may not be completely accurate all of the time but due to amount of pages they process, they are vastly more reliable for making these judgements – particularly when it involves making millions of decisions each day.
With the evolution of programmatic technology, marketers are able to track audience behaviour in real-time and cross examine a number of their interests for intelligent targeting.
As the industry continues to serve millions of ads per day, there has been very few reported examples of bad placement which is a testament to the improvements in both execution and technical implementation.
While one cannot disagree with the overall thrust of the City A.M. article, I think it is guilty of misjudging the relative importance between man and machine in the advertising technology industry. The article concludes, “Machines and algorithms have a crucial role to play, but we still need skilled drivers to win in this race.”
To emphasise just how fundamental, and successful, machines are in operating the internet economy, I suggest this would be better phrased as “without machines and algorithms, there’d be no race – but humans still need to set the rules of the race”.