The help of my friends

Henrik Schinzel

The help of my friends

By Henrik Schinzel, February 24, 2011

Facebook’s valuation just continues to rise. The latest number: a whopping $60 billion.

I can’t really judge if a company where one in 13 people on Earth log in each day should be valued at 6, 60 or 600 billion dollars. But I do think there’s some confusion as to where this value would come from. What can the The Social Network do?

Max Levchin (Paypal, Slide) and Bill Gurley (Benchmark) explain it well in this panel discussion:

While it is clear that Facebook has become the de-facto white pages for finding people worldwide, and a communication network to rival even email and texting, the story is different when it comes to commerce.

What Facebook knows when it comes to recommendations today is primarily your social graph – who you know. What it knows a lot less about is your interest graph: what you are interested in, what you have bought, and what you’re considering buying.

Gurley cites a Netflix project where they used customers’ social graphs as the basis for their movie recommendations, but quickly found out it wasn’t as effective as their existing algorithms.

A lot of retailers are pioneering Facebook recommendations; widgets that tell you what your friends have “Liked” on a certain site. There are two problems with that:

1) As Gurley puts it: “It seems intuitive that friends recommendations would be powerful motivators. But when you look a little bit deeper, you hang out with people who have different very different tastes than you.”

Consider your average Facebook friend – do you share their taste? In what context? Going to myself, I may have one or two friends who I’d trust for product reviews. But to rely on their taste in books and movies – no way. When it comes to retail recommendations, you are what you buy, not who you know.

In fact, the most likely situation where a friend recommendation would work is when it’s an item with very broad appeal (say Dan Brown). But those recommendations add very little in incremental sales and can be accomplished with a simple top-list.

2) It requires explicit input. There’s a well-known Internet rule saying of all visitors to a site, 90% are lurkers who never contribute, 10% are occasional contributors and only 1% are active contributors.

In Facebook terms, this would imply only 1% of all shoppers can be counted on to reliably tell you using the “Like” button what they prefer. The rest of us, you will never know or get very scattered input (a like on a song, a picture of a dog, and a college Alumni page).

As a retailer, rest assured that you still own a much more detailed and actionable interest graph than Facebook. It’s just a matter of putting it to practical use.

Be Sociable, Share!

    Leave a Reply