The failure of social reviews

The development of e-commerce during early 2000s transformed our purchase behavior beyond online shopping. Big aggregators like Amazon made it extremely easy for users to share opinions about the products, giving customers more power than ever before. It resulted in the very common pre-purchase research we all do now before buying something significant (or not even so). I'm still to meet a salesperson that knows more than me about anything I buy, so I see the inevitable fate of the good old salesman: extinction. (Interestingly enough, this trend is still lagging behind in retail financial services, perhaps signaling a spot of opportunity).

This new consumer attitude was channeled by a well known feature of most sites now: the social review system. A 5 star rating system that allows customers to evaluate the product so the marketplace can average that and order by it, showing you the most regarded products first.

This system had to be tweaked over time, trying to weight the relevance of a review by letting users rate the review itself (which is how Amazon orders the comments of a product). The impact of the rating in the number of units sold is so material (trust me I have an app on the App Store), that it has been hacked many times. Competition will weight a product down with early negative reviews, or someone will hire an army of assistants sitting in a distant country to get that 5 star Nirvana.

Over time, I've lost confidence on this system. Not so much because it can be manipulated, but because it fails in the last five meters. What I mean is, the social review system is unable to show me products I would fall in love with. Films clearly illustrate this point. Go to IMDB, see a rating of 8.4 in a movie and tell me if that is a good predictor that you will love that movie. Likely, you will find it a very good movie. Now, loving it is a different league, and the rating won't tell you so.

Social reviews have been extremely effective in filtering out the crap, which is a huge win. But we consumers take that for granted now, and seek to step up the game. Technology, massive reach, averages and other algorithms are of no use here. This is where the human touch surfaces. Can technology help us in those last 5 meters? It's not about modeling taste, it's more about creating connections, identifying affinities and, again, working on relevance.

Again, another interesting (alas pretty much first-world) issue to address. Have a good weekend all!