Tapping the Google Play Store icon on my phone earlier this week I was faced with a horrifying sight. Sat staring at me in the ‘Recommended for You’ section was the official Tottenham Hotspur app.
It’s lucky Google isn’t in charge of selecting my Christmas presents. As a man sporting a (tasteful-ish) Arsenal tattoo, I’m on the verge of suing for slander.
The personalisation of the web has taken great strides, with big data helping to draw detailed pictures of who you are based on where you’ve been, but how do companies find the right balance between trying to deliver based on what they know, and what they assume?
As our smartphone activity, online card purchases and use of social media continue to rise unabated, we fill the big data bubble with information that’s informing billions of pounds worth of marketing and web design decisions.
However, my wife and I share eBay and PayPal accounts, I tend to use her ASOS details, and we’ll buy from whoever’s Amazon account was logged into last. I’m sure we’re not the only ones in the consumer form of an open relationship.
How are companies expected to track any behaviour patterns based on that?
Who does the web think I am? Fortunately, Google makes this information freely available.
It has my age and gender right based on my browsing history, but that’s where the accuracy seems to end. There are some very odd topics under ‘Interests’ including make-up, East Asian music and Volvo. I have nothing personal against any of these categories, but they sit very close to the bottom of a list of things I might consider buying.
Despite my online life being ensconced firmly in the Google ecosystem, the company seems to know very little about me. Many would argue that the custom-built experience of the now defunct iGoogle was far superior to that of the presumptuous Google Now.
Response and conversion rates to online advertising are high enough to suggest that many companies do in fact get the approach right via those channels, using behavioural retargeting and other tools to deliver poignant messages to the audience at the right times, in an interesting rather than intrusive way. When used correctly this can be an immensely powerful way of generating attention and revenue.
But people react differently to onsite content as opposed to advertising, and if you’re planning to have personalised segments of your site, you need to be much more precise with how it’s delivered to give your users what they want. Here are a few things to consider:
Employ experts who understand your market
At some point you need to stop with the questions, seeking that magic piece of information that will solve everything – just make a judgment call. Your customers aren’t expecting or even asking for you to know everything about them and partly the reason for coming to your site is to benefit from your expertise.
Behind every great restaurant, there’s a head chef who has drawn on all their years of experience to develop a menu to tempt and tickle your tastebuds. Their craftsmanship, coupled with a proficient front-of-house team, makes for a great gastronomic experience, and they’d never think of asking each diner what it was they fancied to eat that day and cooking up individual dishes.
Allow some tailoring of the experience
The BBC homepage experimented a couple of years ago with allowing people to move each element of its homepage around; according to their dev blog, few took advantage the feature and they have opted for simple content filtering instead.
This is a sensible option that many sites should take notice of; use data and editorial expertise to show off what you think will have most appeal, but give people the option to quickly cut down the options of what they want to see, hear or buy.
Use ratified information rather than user inputted data
A “Hello ggdahsgd!” message on your homepage or email looks very shabby, but with many users opting not to offer up proper information when completing sign up forms, this may be what you’re displaying.
Only offer personal messages if your login is hooked up to a Facebook or Google API where there’s a higher likelihood people will use their real details.
Create user scenarios, not personas
Meet Jane. She has 2.4 children, likes Coldplay, lives in Milton Keynes, and bears no relation to anyone who actually exists. Personas are often used in marketing to segment audiences into manageable buckets, but rarely do they translate to the real world. Instead, consider example scenarios of why people might be landing on your website.
Meet Jane, she’s just found out the theme for her Christmas party and wants a new dress that screams ‘1920s’, or meet a different Jane who has just been invited to the wedding of her ex-boyfriend and wants something a bit sassy to show him what he’s missing out on. Your messages are much more likely to connect with people using this approach.
Author Michael Buckley once wrote “You can’t judge the many by the actions of the few”; flipping that quote, I’d argue that “you can’t judge the few by the actions of the many.”
If companies want to add ‘Recommended for you’ content to my web experience, they should take a bit more time to get to know what I’m into, although I’d accept that ‘Recommended for you based on cookie data that we’ve collected and a complicated categorization matching algorithm’ doesn’t quite have the same ring to it.