Four principles for respectfully designing with user data
Principal Product Designer, Clementine Jinhee Declercq, shares some principles, using metaphors and examples, for respectful personalisation.
Every day we’re surrounded by more and more personalisation. We’ve come to expect it, and even demand it. Especially in online retail, where we want brands to remember us and deliver personally-tailored experiences tuned in to our tastes and preferences. This depends on data, deployed to gain a better understanding of users, and deliver a more personal experience.
All fine in principle. As long as this personalisation is relevant and thoughtful (just because other customers bought a hairdryer with the same book on gardening I just purchased doesn’t mean I’m likely to, too); and as long as platforms don’t lose sight of empathy and trust, anticipating needs while respecting the owner of the data. There are lines not to be crossed. These lines protect and build customer trust, and they show respect. To preserve these lines, personalisation needs principles.
Principles for personalisation
Trustworthiness. Trust is what’s at stake in how data is used (or abused) in design. Teams exploring ways to make product experiences more personalised identified clear patterns emerging, which would help to maintain this trust. These patterns were then used to develop our four principles for personalisation.
Each principle codifies respectful design practice, providing actionable guidelines that any tech practitioner can follow when working with user data. They are designed to build on each other and work together as a set.
Here are these principles, illustrated by metaphors, with some real examples and challenges we addressed in defining them for use.
Principle I : Don’t be intrusive
Imagine a gigantic eye is staring at you the whole time. Or a huge object rolling in front of your car when you’re driving. Personalisation can be just as intrusive if it is not used sensitively. It can get in people’s way or make them feel uncomfortable.
For instance, we want to help users to find the clothes they are looking for, and could achieve this by accessing their photos, detecting their taste to suggest similar products. But that might feel intrusive, if not invasive, and a little bit creepy.
We can make our recommendations less intrusive by asking users to share one fashion photo to find products in a similar style. It’s less invasive than asking upfront to access their fashion screenshots.
There’s a fine line between personalised and too personal. To avoid crossing that line, consider how best to build respect into the experience and help users to feel comfortable.
Another example of an intrusive experience is asking too many questions at once in the hope of collecting user data! Be selective with the questions, so they don’t get in the way of your customer’s journey or risk becoming just noise.
As these examples illustrate, there’s a right moment to ask questions to collect user feedback; and a simple Yes or No answer is preferable to multiple choices.
Principle II : Communicate transparently
Personalisation should always be visible, transparent and honest. So it’s best to avoid any small print or ambiguity. Magnify it, and say it out loud so everything is clearly understood and relatable.
If a customer bought a red t-shirt last summer, we might want to suggest t-shirts they might also like to consider for this summer. If our suggestion is based on what the user already owns, we need to make that clear. Users often trust things they already know, and they are more likely to trust platforms that are transparent about the way personalisation is applied.
Users can often be skeptical about recommendations. You can overcome this with clarity. Are you stating clearly how and why you are using their data in your suggestions? Be clear and upfront about this, and what the effects will be when they provide input. Vagueness can create misunderstanding and lead to mistrust.
Avoid over-confident assertions, such as “Your filters”. These are only suggestions based on usage. It’s more polite to suggest than assert in this context.
Principle III : Reward with clear benefits
Imagine being asked to carry a boulder up a mountain. You’d expect some reward for your effort when you reached the summit. Similarly, every user interaction demands an effort, and every effort should be rewarded.
Before asking for any user input, it’s best to communicate up front what the benefits are for your users. Ensure the benefits can be seen quickly after the effort is made and that the value exceeds the efforts invested.
Even when an interaction only requires a small effort, it’s best to provide information about the benefits. Cause and effect is critical to building trust in personalised experiences. Just like when you throw a stone in the water, you can immediately see the splash and ripples.
In the end, there should always be room for serendipity — delighting your user in an unexpected yet relevant way. Think about the last time you were truly delighted by an online experience. What made it memorable? Can you apply this thinking when you design?
Look for opportunities to use personalisation to anticipate journeys, and provide a rewarding surprise. For example, instead of just indicating that a size is not available, what about proactively suggesting other similar t-shirts based on what is available in the user’s size?
Principle IV : Give control
Data can be valuable and rewarding for brands and designers. The insights it affords can help us build stronger bonds of trust and loyalty. Yet no one likes to feel that someone’s pulling their strings or robbing them of agency. So it’s always important to remember: we do not own the user data. The bond breaks if we believe we do and make this obvious.
For example, if you use user data to suggest targeted sponsored ads, then you should consider providing a way for users to control this.
Another danger of personalisation is creating a filter bubble, by selectively guessing and limiting what content the user may want to see based on their data. If they only see part of what’s out there they might start to feel like they’re trapped in a tiny house determined by data.
So, it’s important that we provide a clearly signposted exit to the wider collection that is not delimited by user data.
Lastly, remember any personalised experience you attempt to build will go wrong at some point. Personalisation is based on assumptions that are in turn based on user actions. These can be misinterpreted. Or, things simply change — especially in fashion! To improve this, create a feedback loop. Provide users options to give feedback on your suggestions, if they are relevant or not, if they like them or not. With this feedback, you can fine-tune the engine even more, to make the experience better, smarter and more human.
Applying the principles
All of these principles are designed to work together. It’s about finding the right balance to create personalised experiences that are respectful and trustworthy. If users see clear benefits they are likely to trust in recommendations, and trust can erode any potential discomfort. As long as we are transparent about how data is used, users still feel they are in control. This is how we consider them in our practice, applying them as a reference point when designing with user data.
How about you? If you are also working in the realm of personalisation, what other principles do you work with when designing with data? Is there anything you feel is important that is missing here? How do you ensure a human-centered design with data in your practice?
We’d love to hear your views.
Special thanks to Robert Mighall, Rachel Bundock, Hertje Brodersen and Jay Kaufmann for making this piece even better!
