Predicting Cannes winners has long been a sport for the creative industries' top minds, but now it can also be done by computer algorithm, thanks to the Oracle from Decoded, which promises to identify likely winners using big data instead of creative judgment.
Rather than take their word for it, Ad Age put the Oracle to the test, pitting it against the Leo Burnett team, which claims that its Cannes Predictions reel has an 84% success rate in forecasting Lion winners over the last 26 years. Leo Burnett has only missed a Film Grand Prix twice, and one of those years was 1997, when none was awarded.
The shortlist for the Innovation Lions, only in its second year, has been released ahead of the festival, so Decoded's Oracle and Leo Burnett's team set to work identifying the top two contenders from the 30 hopefuls.
Man and machine came up with very different answers. Leo Burnett picked its own Samsung "Smart Bike" – created by Leo Burnett Milan -- as the top contender, followed by Intel's "Project Daniel," which manufactures low-cost, 3D-printed prosthetics for victims of violence in war-torn Sudan. "Project Daniel" was entered by The Ebeling Group in the U.S.
Mark Tutssel, Leo Burnett's chief creative officer, worldwide, said via email, "If we look to last year's inaugural [Innovation Lions] for guidance, we see a focus on ideas that have a real human impact. If this year's jury follows this direction, I expect big things from Samsung "Smart Bike," a next-generation bicycle that is arguably the world's safest."
Leo Burnett's methodology is purely human. The team looks at global award show performance, monitors online buzz, and taps into its own global creative network for local insights, as well as listening to gut instinct.
The Oracle made less human choices. It picked "Beats Music" -- the music streaming service entered by R/GA London -- as the top contender, with a 77% chance of taking home the Grand Prix. "Samsung NX Rover" – a user-controlled camera bot entered by Cheil U.K. -- came out second favorite, with a 58% chance of victory.
Decoded's predictions for the Innovation Lions were made purely on the basis of last year's data, looking at which agencies, which regions, which cities, which holding companies, and which kinds of campaigns had the greatest success.
Because the Innovation category uniquely requires shortlisted entrants to do a brief in-person presentation of their entry for the judges--so finalists have to know in advance to be in Cannes--it's the only category that releases a shortlist before the festival starts. In its second year, the Innovation category drew 206 entries, and the shortlist contains 30 finalists from 16 countries, including 9 from the U.S. Innovation winners will be announced on the festival's last day, June 21.
By the time the Cannes Lions International Festival of Creativity kicks off on June 15, the Oracle should be able to factor in more elements. It is designed to be a flexible tool, so that it can process relevant information for the different shortlists. YouTube and Vimeo views will be used in calculations for the film contenders, but for other categories, Oracle will mine different data. Pinterest, for example, can be used as a measure of how much people are creating their own content based on campaigns.
There are plans to take it even further, and factor in elements like which judges are known for killing off work that is expected to do well.
John Ridpath, head of product at Decoded, which teaches Code in a Day and Data in a Day courses, said, "Creativity is such a human thing, that to reduce it to numbers and data can feel cold – there's a tension around the subject that makes it interesting. We want to get people talking about data."
Decoded is curating a day of "Data, Insight, Strategy & Creativity" as part of the festival. Mr. Ridpath added, "Rather than just give a pitch for Decoded, we wanted to do something around data -- and prediction is one of the most valuable things you can do with data."
The Oracle is already issuing a few warm-up statistics ahead of the festival. It has found, for example, that IPG entries are 78% more likely than WPP entries to move from the shortlist to winning a Lion trophy – a shift that is particularly pronounced in the film category.