• by Wayne Friedman , Staff Writer, August 19, 2015

Speaking at MediaPost’s Programmatic Insider Summit, Morgan Vawter, programmatic marketing & data management lead of North America for Accenture, said marketers are increasingly using competing algorithms to find a better algorithm.

Vawter calls this “ensemble” method modelling. She notes that Netflix, for example, uses this well when it comes to figuring out what a TV consumer might like to see.

“When you look the way your recommendations are served on Netflix, and how you are going to rate something, it’s not just one algorithm making those decisions. It’s a number of algorithms competing against one another. And then they are building this better and powerful algorithm.”

She says: “Interestingly, attribution vendors are seeing this as a great competitive advantage. If they can give you an ensemble method for doing this-- one, it makes them completely transparent.”

“And, two, if you have a great team of data scientists (or one), he [or they] can come in a build a model. [It] can compete against their models in real time to create this better marketing attribution model."