Designing with Data for a Personalised Experience
Understanding users through interaction and observation
Whether on Spotify, Netflix or Amazon, it’s no longer surprising to see an experience that is becoming more and more personalised based on likes and clicks. In a hyperconnected world where there is an infinite amount of content at our fingertips having a personalised experience is key. Personalisation is providing content that deeply resonates with each individual and having experiences that are tailored to specific wants and needs based on a thorough understanding of behaviour and intent.
So how do we, as designers, provide such content? Where do we begin when developing a product concept for a personalised experience? What are the ingredients we need to consider?
1 — Data as the voice of the user
Personalisation is about the individual as opposed to segments (groups sharing common characteristics). It is about recognising those behaviours that are unique to each individual, deciphering their underlying wants and needs.
One way to gain a better understanding about every individual is using implicit and explicit data.
Explicit data
Explicit data is data that a user deliberately shares for the purpose of customising an experience. For example, “following an artist” is a deliberate action taken in order to see more content from a brand one likes. It’s a preference that is shared intentionally so that it’s reflected in the content experience.
Implicit Data
Implicit data, on the other hand, is data that users don’t consciously share. It’s the data that arises from how one engages with products and experiences: the little click to play a song, to pause or stop, the various interactions involved in viewing or buying a product, opening a newsletter, or sharing an article. All of these interactions have indirect consequences: by opening a newsletter you’re showing interest in the content, by viewing a product you’re probably considering a purchase…
Every user interaction is a representation of a personal preference, a conscious need or a vague intent. If used correctly, implicit and explicit data should help us discern those intentions and personalise the experience accordingly.
In an early product design phase involving personalisation, we could begin by asking the following questions:
- What implicit and explicit data is already available in the current product experience?
- Is that data being leveraged to deliver any sort of personalised experience?
- What other data is not yet available but may be important to consider?
2 — Solve: Matching the data to the use case
The next step is discerning how data can be meaningfully used. The right data needs to be leveraged for the right use case in order to improve core user experiences, solve critical pain points and anticipate a users’ intent in order to provide a delightful user journey.
Let’s take an example.
When you go to Netflix as a user, you can run through various movie channels in a home feed. There, you will notice recommendation channels like “Continue Watching” and “Watch Again”. These are a form of personalisation as they are based on your personal viewing history.
Both provide different user values and solve different user pain points based on various user history data parameters: “Continue Watching” leverages history data that is relatively recent, assuming that you would want to keep on watching from where you stopped. “Watch again” suggests a personalised selection of movies that you finished watching in the past. This is assuming that you watched them all already and may be interested in seeing again.
In a similar way, we could contemplate the following:
- How might personalisation play a role in solving user pain points? Using which implicit and explicit data?
- How might the data be tailored to improve the product’s core user journey and anticipate intent?
- Are there other variables concerning the data that could affect our understanding of the use cases: recency of the data, location from which the data is shared, weather condition, special occasions, etc…?
3 — Delight: Showing what matters in the right moment
A valuable personalised experience is one that understands a user’s intent and shows the right content at the right time.
In the case of Spotify and Netflix, personalisation plays a large role in their respective “Home” pages. The minute you land on that page there is a stream of personalised content available.
E-commerce shopping sites like Amazon or Zalando provide other valuable examples.Typically, users who go to an online shop will start searching for products they are considering to purchase. They will browse around, view a couple of products they like, and keep on eye on a few to make the purchase decision later.
Amazon focuses on this precise user journey in order to deliver a tailored experience: by looking at your recent views, Amazon sends out a personalised email 2–3 days later featuring the products that you checked along with other relevant items.
Some further questions to consider:
- Which moments in the user’s journey should be personalised?
- Which channels could be used? Website? Email? App Notification?
4 —Learn: Delighting and/ or improving with a solid feedback loop
Let’s assume we found a perfect use case for personalisation and the right moment to deliver it. At this point, we hope that our users are impressed with how well we anticipated their intent and nailed their preferences.
Consider, for example, Spotify’s “Discover Weekly”, a weekly personal playlist based on the songs you play and artists you follow. Spotify users have been incredibly pleased with how “ right” the selection was every week. We’ve all probably been a little creeped out by how much Spotify seemed to know about our very personal musical tastes.
However, the opposite outcome is also possible:
- When you see a movie that is recommended specifically for you but you dislike it
- When you receive an email with products recommended just for you but you already bought them and are no longer interested
- The so-called “Personalised For You” section where you see suggestions of content that are simply irrelevant to you in every sense of the word
- When ads are targeted to you based on content you interact with, but end up feeling so random
In order for the personalisation process to foster trust, rather than harm the user experience, it is necessary to build a consistent and solid feedback loop to enable the engine to improve over time. Because it’s unlikely that a personalisation process will be perfect for every individual from the beginning. Mistakes like those discussed above are inevitable, which is why there must be consistent oversight.
The feedback loop could look like this in an ideal “personalisation” eco-system:
On Instagram, for example, users are able to give feedback via tree dots in order to improve the quality of suggested ads. Users are presented with 3 options:
- It’s not relevant
- I see it too often
- It’s inappropriate
Feedback suggestions should be directed to an engine, which, ideally, would get better at suggesting appropriate ads.
So a few more questions to ponder then:
- What types of feedback are needed in order to holistically improve the personalised experience of the product?
- When to ask for this feedback?
- How to ask for this feedback?
That’s about it.
I hope this article serves as inspiration to get started with data personalisation for your product. I am also curious to hear your thoughts on this topic. Feel free to comment!
In the next article, you will learn about the 4 human-centered design principles for designing with user data!
Clementine Jinhee Declercq is a Principal Product Designer at Zalando, experience designer loving to design digital things for people’s use.
