Customisation rather than personalisation

You may remember a viral tweet from 2018 that had e-commerce marketers rolling on the floor.

“Dear Amazon, I bought a toilet seat because I needed one. Necessity, not desire. I do not collect them. I am not a toilet seat addict. No matter how temptingly you email me, I'm not going to think, oh go on then, just one more toilet seat, I'll treat myself.”

This echo chamber effect is the antithesis of what online vendors want. Try to guess your customers’ spending habits too hard, and you could risk a little faux pas like this.

Of course, there is a place for these filter bubbles – the social media echo chamber, for example, shows users what they like based on their previous activity. Sports, political views and so on. And indeed, we see artificial intelligence in action every day, for example:

• Amazon personalises user journeys based on previous actions

• Google’s Nest thermostat tracks residents’ movements around the house and tailors the temperature based on their preferences.

But what if Nest owners have visitors? Are we locking our customers into patterns that follow an algorithm, rather than what they really like?

We need to ensure we’re not forcing our users into reductive content cycles. We may make promises based on how much data a user is willing to give us – but we don’t want to risk frustrating them.

There’s a fine balance between a good user experience and privacy or autonomy. The key is to customise, not personalise.

Customisation rather than personalisation

Perhaps we’re being hard on Amazon. Consumers generally enjoy the convenience of a personalised experience, but we also need to understand how to avoid echo chambers.

This is where customisation comes in: giving people freedom to set preferences, rather than making assumptions based on previous data. The team at Harvard Business Review conducted a study with some eye-opening results: 93% of users have customised a site based on their preferences.

They cite a great example in the Yahoo! homepage: here, we can set our display to show widgets such as the weather forecast in our local area or our horoscope. We’re also seeing it more often on publisher sites – for example, Apple News and the Guardian offer us the chance to follow certain topics.

The benefits of individual customisation

Empowering customers gives us access to zero party user data: essentially, that which they are willing to share with us, such as their homepage preferences. Of course, as ever, we need to be compliant and respect this data from a privacy and security perspective.

We can also reduce anxiety from information overload. A regression analysis at the University of Maryland discovered that customisation of news environments was associated with reduced anxiety towards current events.

Allow an edit button

So, how else can we avoid the echo chamber effect? As much as we should empower our users, we should also inform them. A digital privacy and preference centre will allow customers to understand and manage their data choices.

How does this benefit us? We’ll see higher conversion and retention rates by serving our customers long-term, particularly those whose preferences change. This is ideal for educational platforms, where learners may progress from beginner to advanced.

Recommend blended content

To achieve the right balance, we can blend content. We shouldn’t assume everybody wants the same thing based on user attributes – remember the age-old example of two 73-year-old men, Ozzy Osbourne and King Charles.

Rather, we should follow the 80/20 rule. Netflix does this brilliantly: recommending 80% of titles based on user activity, and filling in the other 20% with something they may like. In a training platform scenario, this is ideal for getting more courses in front of a diverse customer base.

So, why are echo chambers bad?

In many contexts, making reasonable assumptions based on customer activity can lead to more sales. But we mustn’t force our users into algorithmic patterns that may frustrate them.

Instead, we need to remember that personalisation doesn’t trump good feedback. We should offer features or options that empower them to personalise their own content, rather than rely on algorithms. Strong user feedback may even lead us to create features we’d not considered before.

Of course, this comes with challenges – particularly if we’re scaling. The more users and content we have, the more patterns and customisations there are on offer. We need to present these options in a digestible way to avoid overwhelm.

Zero party data will help us to understand our users, rather than assuming King Charles and Ozzy Osbourne are the same person. Using a data-compliant zero party policy will help us to come up with our own conclusions – people may like to see the weather on their homepage, for example.

We live in an age where users want to feel like they’re in control. Let’s empower them, not constrain them.