IAG media chief: Marketers must understand AI, or lose effectiveness
- 26 February, 2018 13:12
Dr Willem Paling (left)
It is imperative marketers better understand artificial intelligence (AI), otherwise they could end up measuring the wrong thing and miss the right digital advertising opportunities.
This was the clear message from Dr Willem Paling, director of media and technology at IAG, at today’s ADMA Data Day in Sydney.
Dr Paling said AI is in every demand-side platform (DSP) and will only become more prolific. But it’s almost pointless unless the marketer knows how to extract the value from the technology effectively, specifically causality.
“You must set the right goals when it comes to extracting value from AI, and many marketers simply don’t understand the goal of the system,” he said. “The general goal of marketing is to get more sales, or to maximise those sales. However, this doesn’t take into account the difference between obtaining incremental sales, as opposed to attracting sales from people who would have bought anyway.
“The goal should be to find incremental sales, or additional sales, to find compliers who buy if and only of we supply that marketing.
Typical DSPs will take behavioural and customer data and use that to build features like upper funnel on-site signals, Paling continued. This builds a model used to asses the conversion likelihood, then the model adjusts bids depending on the likelihood of customer converting.
However, there is no distinction there between compliers, always takers, never takers, and defiers, it just puts ads in front of customers who are most likely to buy anyway.
“This kind of use is naïve to intent characteristics, and doesn’t ad incremental value,” Paling said. “If our goal is to imply touch as many people as possible, out our AI system can optimise even without proper ads, even using a blank ad. With the wrong goal, a touch goal will still be optimised by AI.
“If we don’t understand the adtech and martech products and understand the goal of AI we are at risk of putting an impression in front of people who were going to buy anyway,” he said.
To better negotiate the explosion of adtech/martech, and meaningfully deploy AI and machine learning, marketers need to develop scientific literacy, Paling said. The best way to do this is not to engage with the sales force that sell the tech, but to engage with the scientists behind the technology.
“The impact of AI on marketing has, thus far, been largely superficial. For AI to improve from ‘black box magic’ to being meaningful, the marketer must develop a strong understanding of causality, understand the scientific approach, develop scepticism of vendor AI claims, and develop strong relationships with data scientists,” he said.
“The single most important question a marketer can ask is ‘How does this system address causality?’
“We need to develop a strong understanding of causality, how we measure this so we can understand what would have happened if we didn’t do this marketing."