Tuesday, February 26, 2008

What's the Problem? It's the filter...

There's a big problem today online. What is this big problem you might ask? How can I find recommendations that are tailored to my preferences? Like if you want to find a new sushi restaurant tonight or a hotel to stay in while visiting Bangkok, how would you do it? There are a lot of options out there right now and many more gaining speed, but I'm still not sold that any of these (other than Vyoo of course) can solve this problem eloquently.

Here's a look at what's out there and what's wrong with these options:
1) Social Networks - circle of friends (facebook, myspace, etc). These websites are great for getting people together and sharing media and pretty much telling people what you're up to. But they are not great at compelling people to make purchases (for advertisers) or for finding recommendations or reviews (for consumers). I'm including them in this discussion because we're adding a bit of a social network to our solution - but in a much different way.

2) Recommendation engines - you all know these, TripAdviser, Amazon, Hotels.com, Netflix, whatever it is that tries to advise you on a purchase. They are great at giving you lots of options, but they are terrible with accuracy. Netflix has had a competition going on for a few years to improve their recommendation engine by 10%. The prize, a million bucks, hasn't been given yet and there has been a lull in activity since the 8.5% improvement rate was reached last year. I personally think it's incredibly difficult to trust the ratings I receive. Other people may feel differently, but I think there's no doubt that there is a lot of room for improvement.

3) Collaborative filtering - as discussed last post, involves large amounts of data and multiple inputs to use as much information as possible to make decisions. The place I could see this working is when people go crazy over a product - like an iPod. I could see them recommending one of those to you for an mp3 player. But that doesn't exactly give me what I want - it gives me what's popular. Collaborative micro-filtering was covered below, so I won't go into that.

4) Search engines. Great for delivering large quantities of information left up to the user to consume. Not great for purchasing behaviors or tailoring different results to different people.

So there seems to be a void that we can fill or that someone can fill if they can figure out a way to give people what they really want.

No comments: