What retailers can learn about personalization from #TheDress

Who cares what color you see? We care about what color you like.

The Dress. Dressgate. The dress that broke the Internet. Ruiner of friendships and starter of fights.

Whatever you’re calling it, and whatever colors you see, The Dress has set the Internet ablaze. It’s a testament to the power of social media, and how quickly word travels and opinions form.

It also brings to mind an interesting theory – what if retailers could predict the colors you like, and magically change the colors of the dress to suit your specific tastes?

Sounds sci-fi freaky, but it’s not actually too far from today’s reality. Data science and solid engineering, of course, hold the key. Retention Science’s machine learning algorithms, for example, can predict with 96% accuracy if you have a higher purchase probability of a white and gold dress or a blue and black dress.

It won’t have the same divided-faction effect of the poorly-lit dress, but it’s a pretty amazing way to shop.

It’s Not Magic: It’s Data Science

Our data science modeling is based on analyzing user data, product data, and behavioral tracking. Typically we look at 215 variables across 7 data tables. On average, we are analyzing over 100k signals per day across all our sites to make our optimizations and predictions. In plain English? Because of how we track customer data and build profiles based on that data, we can effectively predict not what colors you will see, but what colors you want to see.

The Dress broke the Internet because people couldn’t decide what they saw. But imagine what would happen if retailers could provide each dress to the exact customer that wants it. Now that is personalization taken to new heights. (For the record though… it’s definitely blue/black.)

Screen Shot 2017-07-19 at 1.42.20 PM