Being data-driven has become a synonym for being new-age. The best thing about data and its analytics is that it can show us what people do on their own. But what it doesn’t tell us much about is the context, motivations, and intent behind the data to make it useful enough. We explore how design can inform the process of collecting and creating data that can help in finding insights about user behavior, and how product designers can leverage this phenomenal dynamic between data and design to create compelling experiences.
Mohita Jaiswal & Hari Nallan - April 2021
A good product amalgamates design, empathy and data
Netflix backed by its algorithms, is continuously learning to calculate and suggest the kind of content we would like to see. However, it often fails in identifying and having information about one’s real experiences of the content. Netflix knows that you watched four episodes of Westworld on Saturday but it doesn’t know that you were making and also eating a wonderful salad as you watched. Likewise, it couldn’t listen in as your manager and you chatted about the show the following morning.
With limited information, the algorithms can only paint a distorted picture of its users and fail to engage and entice its audience beyond a point.
Ted Sarandos, the company’s chief content officer agreed, “It is important to know which data to ignore,” and added that his decisions were 70 percent data, 30 percent judgment. Not knowing the why behind user actions, the current efforts in the collection of data and gathering of information are limited and ask for a transformation.
Why Designing for Data matters: Moving beyond Data-driven design
The idea behind gathering data is that the more data there is, the better the design can be. Data-driven design or designing with data became popular on the premise that designing backed by the findings from data, could help improve a product by inculcating validation into the design.
But the challenge is that although data can point one in the right direction, it does not necessarily always prescribe what the solution should be.
Various pitfalls emerge which indicate that sometimes it is better to be data-informed than data-driven. To evaluate high-level user motivations, expectations, perceptions, or emotions, a data-driven approach might not be the only method to make design decisions.
Data-driven design or designing with data on one hand tends to be reactive whereas Designing for Data would mean taking a proactive approach to collect/gather data through primary user behavior/actions where those user actions themselves are guided by the design we create.
The idea is to capture the data with intent. A/B testing is one very relevant example that captures user data through intent. Comparing a variation against a current experience(in an A/B test) lets you ask focused questions about changes to your products, and then enables collecting data about the impact of that change.
Example of an A/B test: Changing the customer testimonial logos, from their original color to black and white, led to an increase in form submissions.
Your infinite loop of data and design – with insight perched at the center
Imagine you get a message on your Android watch and want to respond with a frown. Scrolling down an entire list of emojis, would you actually want to respond back a frown?
Android Wear has a very interesting way of handling emojis. Android Wear lets you draw an emoji with your finger tip. If you can’t draw particularly well, the software simply guesses which emoji you were trying to draw, and plugs it in for you. Comparing your drawing against a repository of collected data of common emojis entered by users, your emoji gets mapped to the right emoji using a best-fit algorithm.
True to a design that intends to capture data, a very human insight is discovered about users here, allowing data created to inform the design experience continuously.
Infinite your own loop
1. Collect intent-driven data
The two main types of data that you’re likely to collect are about what people do and what people say – behaviors or words. This data gets derived from design, a measurement strategy, evoking users to reveal their motivations, needs and pain points, focused upon a question, a metric which needs to be captured, thus intrinsically embedded in the design.
For example, imagine reading a blog. A user clicks (data) on a Read more CTA (design element), indicating the preference of the user for a certain type of content (insight).
2. Convert data insights to decisions
Data that is converted to insights need to be acted upon to make it useful. One can use tags and computations to derive insights from actions while tools could be used to capture decisions from insights. Insights that do not get acted upon do not lead to solving problems.
For example, you received insight on the user preference for a particular type of content; now a decision could be made to probe further – did the user actually find it useful or not?.
3. Act on decisions by transforming them into design
Act on decisions to convert concepts into executable items to implement a data-driven design. It does not help to validate a hypothesis by creating additional user behavior data but allows the introduction of changes in your design that will solve problems, and measure the impact of your changes.
For example, adding a question and yes/no button to probe deeper into the insight uncovered.
Finding the right ‘balance’
Facebook has the largest amount of data in Hadoop (manages data processing and storage) in the world, yet they base their decisions on more than just their data. User Experience (UX) researchers and Human-Computer Interaction (HCI) researchers at Facebook seek to deeply understand and improve the experiences of the over 2 billion people around the world who use Facebook every month. This becomes possible as one keeps learning through data, without data taking over.
Basing a product decision solely on a data point like “conversion” is not wise. Basing a product decision on a couple of customer interviews is also not wise. It’s essentially about making decisions under uncertainty and updating yourselves as new evidence emerges, constantly readjusting your beliefs when you encounter new data. You never say that you are absolutely certain because then you wouldn’t be able to revise your beliefs and what gets designed in the light of new information.
