DIY Recommendation Engine

Reccomendation EngineI was recently talking with a young entrepreneur that is working on an exciting project in the men’s fashion space. One of the important characteristics of the project is creating a detailed profile of a user’s preferences.

I don’t know the first thing about fashion, so I wouldn’t really know where to start if I were trying to create the framework of the user profile. The concept of using data collected during interactions on a website to personalize recommendations is not new, but it is very challenging. Netflix is famous for its recommendation engine, especially the million-dollar contest it ran to crowdsource a more accurate version.

As we talked about issues specific to his new company, my mind started to wander a bit. I started thinking about the process of learning a user’s preferences. It can be a lot of work to assume preferences based on clicks and hovers.

Maybe the value of the recommendation is strong enough to ask the user to tell us what she likes and doesn’t like. What if the process of describing her preferences was part of the fun? For items that can be visually represented, why not use a simple gamification approach like the old Hot or Not website, where a user selects photo A or photo B as their favorite?

We all know the addictive nature of looking at pictures online. It’s the core of almost every popular consumer website. If I were presented with a sequence of images and asked to pick my favorite, I would almost certainly participate. This is doubly true if I were interested in seeing what was recommended for me. Sure I’m doing the work of data mining for the website, but it doesn’t feel like work, so why not.

There are thousands of businesses that could benefit from understanding their potential or existing customers’ preferences. This would allow them to focus their marketing efforts to deliver the best reward to the ideal customer every time. One of the biggest obstacles to doing this kind of targeted marketing is finding out what current and potential customers actually want. Perhaps using an entertaining methodology of gathering those preferences would streamline the collection of this information.

If Facebook has taught us anything, it is that we are willing to publicly announce who and what we like. From friends to brands to articles, there is no limit to our desire to express our opinion about the things we are presented with. Product developers and business owners that want to make personalized recommendations can get started by making the process of expressing ourselves as fun as liking photos on Facebook.

About johnwilliamson

I like to think of myself as an innovator. From a very early age I have been obsessed with finding new ways of doing things. I’m all about efficiency and the use of technology to transform the way work gets done. I’ve created successful products for others and for my own ventures. I’ve built companies around my products, raising money and acquiring customers. I truly understand the full life cycle of taking an idea, finding a customer, developing a product, and delivering a solution.
This entry was posted in Thoughts. Bookmark the permalink.