I worked on the Personalization team at Amazon from 2000-2004, and more specifically on things like “Personalized Recommendations”, “Customers who bought this also bought…”, “The Page You Made” (a page of recommendations based on what you’ve looked at in your current browser session), and “Your Store” basically a slice of Amazon that was entirely build from what we knew about you (what you’ve bought, what you’ve looked at, what product lines you’re most interested in, where you live, etc). The team was a bit infamous within the company for replacing editorial jobs, though some of that was probably not entirely our fault. One thing was true, though: the numbers were in — the computers were better at finding content and products for you (a mashup of customer reviews, new releases, and statistics) than the humans were.
When I read this fantastic article about how the DNA of certain companies spreads out into future companies (their example was Facebook and Paypal), I thought about my own DNA. It was forged largely on that Personalization team. My first project around that same time, for example, was All Consuming. It’s different now, but at the time it would crawl all 2,000 blog posts that are updated PER HOUR (how times have changed) looking for links to Amazon, IMDb, and a few other online venues that were keyed off of ISBN or UPC. I created a zeitgeist of what people were reading, right now. And I created weak-link groups around everyone who was reading the same book as you at the same time, across the early social web.
My DNA is built something like this: use technology to improve real-world experiences. Whether it be the books you’re reading, or your life goals, or (now) your health habits, it’s all about supplementing core experiences in our culture with cyborg limbs that can some things a lot more effectively than we can alone.
The first signs of my larger goals with Health Month got pushed out over the weekend. Dubbed the Rule Wizard, it’s the first tiny step towards being able to recommend small tangible steps based on both your own data and the aggregate data of everyone else playing the game.
Some notes about how it works:
- At the moment, this is only available to people who are on their 2nd+ month. Right now I’m only using data about your own behaviors in previous months to help you improve your rules for next month. Of course, there is a lot more that can be done here on the algorithm side.
- It recommends you to make your rule more difficult, easier, or to keep it the same depending on how you did with it last month, and will suggest a level of rule that fits your actual behavior.
- It will also give you some guidance on whether you should take more rules, fewer rules, or keep them about the same. There’s a lot more to do here, as I’ve found that sometimes a rule-heavy month is best followed by a rule-light month. But I’m waiting to have enough data to support that claim more solidly.
Baby steps, but just wanted to start talking about this since I think it’s an important part of where I feel Health Month is headed.
Read a bit more about how it works on the Health Month blog.