Data is part of the fabric of our day-to-day existence. Notable companies like Netflix, Amazon, and Google use data extensively to drive their design processes. Undeniably, there are benefits with using quantifiable statistics when constructing and tweaking a user experience. Regardless, sifting through the data to make meaningful analyses can be a challenge, especially if we were to take into account the vast masses of data available. In her workshop, Designing with Data, Pamela Pavliscak demystifies this process and gives us insights into how we can glean new meaning from data to develop a better understanding of everyday experience.
What Does Big Data Really Mean?
Before we dive into what data really means, it is key to first reframe our understanding of big data. A common misunderstanding is that data deals solely with numbers. Though this frequently appears to be the case, what numbers fail to express is how, why, and what influences them. Pamela re-examined our perception of data in her workshop by making a comparison between data science and human archeology. Like human archeology, data gives us a sneak peak into the lives of otherwise total strangers and thereby allows us to roughly paint an “every person story.” This should come as no surprise to most of us. By now we are well aware of the popular use case for data thanks to exposed government surveillance programs and targeted advertisements trying to sell to us. We leave an endless stream of data in our tracks. Just open up floodwatch, a chrome extension that allows you to view ads you see in a day. You’ll be baffled-and perhaps slightly creeped out-to see how much a stranger looking at your advertisement history can learn about your lifestyle and behavior patterns. The data we generate speaks volumes of our lifestyles and how we go about our days. Despite its eerie use cases, big data is really about the human relationships and interactions created by and through it.
The Problem with Data
“Not everything that can be counted counts, and not everything that counts can be counted.” - Albert Einstein
By nature of being creations of human design, bias is intrinsic in data sets. No matter how big the data is, its objectivity can never be guaranteed. For instance, in a research study of social media metrics during Hurricane Sandy, it was discovered that twitter and foursquare use were concentrated in the Manhattan area thereby providing the illusion that the disaster occurred in that area. The social media data collected during Hurricane Sandy only showed one segment of the population, specifically those on twitter using a certain hashtag. This clearly demonstrates signal bias or the bias of omission brought about by collating data from a single avenue. It also highlights how research biases data; just because something can be counted doesn’t mean it is important or relevant.
Why Do We Want to Collect Data?
Oftentimes, people look to data as a means to prove a hypothesis. This usually has to do with business metrics such as conversion rates, bounce rates, and various other key performance indicators. Pamela stated that “99.9% of the time the main goal is to grow revenue.” In cases like these, the data presents us with hard, tangible metrics from which we can make strategic decisions for improvement. Data directly related to profit margins and revenue is without a doubt important. However, they are only loosely tied to UX. In order to fully gain value and insights from data, we must first ask ourselves what exactly we want to know more about.
Where Do We Start?
Analytics is typical in our industry, whether that be A/B testing, surveys, or user studies. In these scenarios, we are left with a lot more data than we know what to do with, and it is common for us to look to the mountainous data for all the answers. An important thing to keep in mind is that data is really about people. It is impossible and foolish to try to decipher a complete user experience from the data. Pamela said, “data represents an approximation of the user experience, not a matter-of-fact truth.”
“They came <something something>. They left. You can’t single source your metrics. Data can be too granular.” - Pamela Pavliscak
When trying to decipher what data points to measure and how to go about collecting all of it, begin by asking yourself what you already know and what you don’t yet know. This does not necessarily have to rely on existing data and is purely hypothetical. Pamela suggests considering measurable moments. Since experiences are never static and are often based on the past at the time of recording, experiences can be viewed as a continuum; there are expectations before, experiences during, and memories after.
Where Do We Go from Here?
There is no such thing as a right or universal way to use data to inform our processes. Admittedly, there are loose guidelines that we can use to incorporate designing with data into our processes. Pamela described a few of them for us to consider.
1. Get Curious
Numbers don’t tell us the whole story, but they provide a context from which to start asking questions. Start by thinking with the existing data, look at existing analytics, and narrow down on certain details that you find intriguing. Categorize your findings into common themes or questions from which you can further your investigation.
2. Think Bayesian
When examining and evaluating data, remember that you are dealing with degrees of probability. Be prepared when new findings from your research thwart your expectations and hypotheses. At the same time, understand that correlation does not necessitate causation. The high concentration of Nobel Laureates in countries where a lot of chocolate is consumed does not mean that chocolate consumption leads to better IQ.
3. Be Mindful of Bias
Social media only represents a portion of the population, while user studies might cause participants to change their behavior because they are being observed. To prevent your data from limiting your analyses, diversify your sources - A/B test, conduct surveys, create heatmaps, gather sales and CRM data, and broaden your research as long as it is within your means. Each new approach we take to gather information, when compared against what we already know, makes the picture a little more accurate.
4. Embrace Complexity
Designing a better user experience with data is ultimately about being people/users first. The most useful data explains the behavior, interactions, feelings, and attitudes of the user. Conveniently enough for us, this is extremely difficult to measure and quickly becomes a matter of ungrounded subjective perception. To combat this, combine qualitative and subjective data with quantitative, more objective data. That way you can gain a more balanced insight into your data.
5. Tell a Story: Add Meaning to Your Data
Ultimately, data is meaningless unless there is an actionable impact to it. It is just a series of numbers, figures, and bar charts. Creating meaning with data is not just about drawing connections and making relationships between data points. It is also about giving meaning to the data and using it for emotional impact so people feel compelled to make a change or continue as they were.
“Designing with data has to go beyond algorithms, automation, A/B testing, and analytics.” - Pamela Pavliscak
Data is about the Human Experience
Big data is no crystal ball; it doesn’t give us all the answers. However, it provides a starting point from which we can gain a better sense of our users and their behavior patterns. With passing time, data is becoming more and more closely connected to our day-to-day interactions. No matter how much we try to make it about numbers and figures, data is becoming more closely tied to human experience. We risk alienating our users if we choose not to consider the human factor in big data. As a result, it is key for us to ask the right questions of our data and continue being curious about it to gain actionable insights and thereby create better user experiences.