The Day After: We Got It Wrong

As you’ll probably know by now, we did not predict the outcome of the 2016 US election. We believe in owning up to failure and learning from it. We knew this would be a close election, and we are proud of sticking our necks out and making predictions. But on the big one, we failed.

Why did the method not work, and how can it work better next time? System 1 Politics is an experiment in applying to politics a method that works for brands. We’re looking back at our data to see if there were any tell-tale signs that would have pointed to President Trump, and we’ll be back with a post-mortem once we’ve done that.


For instance, back in January our pre-primaries model of the candidates showed Trump marginally ahead. But our prediction of shifts in the market suggested Hillary Clinton would overtake him. So our initial read reflected the actual outcome better than our prediction.

That’s the kind of thing we’ll be looking into as we refine and re-tune our method to study the French and German elections next year. Thanks for reading and being part of our project.