Monday, August 1, 2011

Recommendation engine


A few months ago a very good friend of mine sent me a link to a British newspaper review of a compilation album of early seventies music recorded in the southern states of the USA.

Now whilst I’m a child of the seventies, and love much of the music from that period, so called southern rock has never been of interest to me. Way too many beards and confederate flags.

However the friend who sent me the link is someone I trust when it comes to music. So if he recommends an album, I’m prepared to investigate further.

So off I went to Amazon, where the album had a series of glowing four and five star reviews.

Now seriously tempted, my next stop was iTunes, where I purchased the album. And it is indeed a great collection of music, by bands I would generally avoid like the plague.

This morning an email arrived in my inbox from Amazon suggesting albums I might like, based on my viewing of the southern rock compilation.

Whilst I love the idea of recommendation engines, my Amazon email proved to me that they can’t always be trusted. Well not by someone with specialised taste in music anyway!

In order to work really well, like good research, they need to make recommendations based on a broad source. An email after looking at just one album really can’t be accurate enough.

I should point out though, that I regularly get emails of this type from Amazon, and they’re generally pretty good in their choice of album and book suggestions.

Anyway, before I go way off track, let’s go back to the beginning of this month’s column – a recommendation from a friend.

Otherwise known as word of mouth this is what Bill Bernbach referred to decades ago as the most powerful form of advertising.

Bear in mind that when Bernbach made this statement the internet had yet to be invented. And it is via the internet that word of mouth goes in to overdrive.

It is why social media has to be a priority for any marketer today.

Because as exciting as it may seem to be involved a big budget TV ad with a hotshot director behind the camera, the recommendation of a friend is far more powerful.

And if you can get people to recommend your brand or product to their friends, then you’re definitely onto a winner.

Which brings me back to Amazon.

You see their recommendations are based on things they think you’re interested in. Which obviously sounds like a great idea, but based on the recent email they sent me is far from perfect.

I however have another problem with the recommendation engine model.

If everything you buy is based on things you have previously purchased or expressed an interest in, then how on earth are you going to discover something completely new?

And by completely new I mean something different, unexpected or outside your usual area of interest.

You can’t stumble upon something by accident if everything you buy is based on your shopping habits.

Which means that your shopping habits can very easily end up stereotyping you. The same applies to many of the smart new digital magazine formats.

I love the idea of the recently launched Zite – an iPad based magazine that learns which articles you do and don’t like, then over time personalises the content to suit you.

Sounds brilliant, but you could easily end up with everything you read being based on what you have read in the past.

Which, just like the recommendation engine, prevents you from stepping outside of your comfort zone and discovering completely new things.

Now that narrow minded digitally programmed way of living may suit some people, but it sure doesn’t suit me.

And hopefully it won’t suit you either.