There’s a slightly dorky game show inspired digital assistant that lives on my PlayStation called Max. This Siri-like character guides you through choosing what to watch next on Netflix and comes with his very own game where you rate movies and TV shows on a five-star scale.
I would be willing to hazard a guess that the algorithms powering Max’s content suggestions are the same ones Netflix uses for its regular content recommendations, it’s just that Max’s suggestions feel so precise.
Last weekend I was in the house on my own, eating beige-coloured food and looking for something dramatic. Max offered up Zodiac, a Jake Gyllenhaal movie I had previously overlooked, but it was so right for a quiet Sunday night. This feeling of precision seems to come from two places.
Firstly, Max offers fewer recommendations – only half a dozen or so, while the regular Netflix screen offers dozens. Giving the user fewer recommendations makes them feel more precise. Secondly, you interact with Max, answer his questions and play the ratings game with him. These interactions build a sort of trust between you and the machine, again adding to that feeling of precision.
Max isn’t the only bot on the block to pull the friendly card. Eugene Goostman, the Turing test passing chatbot, wasn’t successful because he had the best algorithms or used most data. Eugene was successful because his makers framed people’s interactions with him with a backstory: Eugene is a 13-year-old boy from the Ukraine. Within this context the answers given by Eugene felt more precise, more real, and more human than they would without it.
Another company using the same sort of friendly data techniques but in a much less dramatic way is Google Consumer Surveys. When you look at significant differences between proportions on Google Consumer Surveys it’s not percentages and p-values that are displayed in bold, it’s short sentences, like “Among people in the US West, women picked Fetch with a toy more than men.” This is a lot easier and a lot more convincing than just displaying the numbers.
This trend towards friendly data harkens back to something that marketers have known for a long time and it’s part of the reason why marketing, insight and innovation should be deeply involved in data & analytics: it’s not what you say that makes a difference it’s how you make people feel. The lessons we’ve learnt about communication from TV spots, packaging and banners also apply to data, as big and unfriendly as it may seem.