x Abu Dhabi, UAEWednesday 26 July 2017

Technology puts the personal touch back into recommendations

Social recommendation combines your social network with an understanding of your preferences to deliver personalised suggestions from trusted sources.

Young woman reading next to bookshelf
Young woman reading next to bookshelf

Let's try a simple thought experiment. It's 1990, and you're stuck for something to read. What do you do? Probably, you turn to a friend and ask: "Hey, read any good books recently?" Otherwise, you might hunt down the newspaper and see what the book critics are saying this week.

As an answer to a deep-rooted desire - that is, the desire to discover new products and experiences - it was OK, but not great.

No wonder, then, that since its inception, the internet has been disrupting the old model of recommendation and discovery, as applied to books, music and holidays. But problems remain: say you visit a straightforward reviews sites such as TripAdvisor, for advice on a hotel you're considering. You're often beset by a nagging feeling that this user complaining about the rudeness of the staff is a bit of a nut.

In other words, there is a trust issue. Then there are recommendation engines such as Amazon's, which tells you "users who bought Bridget Jones's Diary also bought Pride and Prejudice". OK, you're discovering new books, but it all feels a bit algorithmic: too much like you're being told what to read by a corporate behemoth.

Now, a spate of new start-ups want to change the recommendation and discovery model again, and solve those problems. Their answer is being called social recommendation, and it combines your social network with an understanding of your preferences to deliver personalised recommendations from people you trust. That's a powerful mix: get it right and it could forever change the way we access new products and services.

Take the US start-up Vineloop (www.vineloop.com), a smartphone app that lets users "loop in" friends to provide recommendations on anything from books to drinks. Say you loop in a friend to provide advice on books: you then get access to her recommendations, as well as those of others she trusts, so that you develop a books "trustline" that starts with friends and branches out into the wider community.

Meanwhile, the new web player Trippy (www.trippy.com) calls itself a "social travel site": it allows users to get bespoke travel advice from people in their social network, and build a "friendsourced" itinerary based on that advice.

All this amounts to a powerful new way of driving people towards products, and that means huge commercial implications. Few were surprised, then, when the race-leader in social recommendations, Hunch (www.hunch.com) sold to eBay for US$80 million (Dh294m) in November. Hunch takes your answers to preference questions and your social network and crunches that through its unique "taste graph" - a vast network of other people and their preferences - to provide smart recommendations. Currently, Hunch says their taste graph holds 500 million people, and makes 30 billion connections between those people and things they like. eBay will use the technology to make purchasing recommendations to its users.

For many, social recommendation will emerge as a useful new way of finding a good book or album. But when these tools are ubiquitous, do we run the risk of becoming trapped in a self-referential taste circle, only ever receiving recommendations from people who are like us, only ever discovering new products that make sense in the context of those we've used before? After all, if we never expose ourselves to entirely new thinking, how can we ever change or grow? Difficult questions, which, for now, we must leave to the future.