Sweating the Small Stuff
Are Biometric Measures Right for Your User Research?
Over the last few years, biometric measurement has burst onto the scene of market and user research. Everywhere you turn it seems there are new advancements allowing you to measure things like heart rate, facial expressions, brain activity, and eye movements in real time, while users interact with your product.
You may have heard some bold claims about the payoff of using these biometrics measures during research, and each new technique is presented as the most useful tool available for tapping into users’ deepest thoughts and feelings. So, how do you wade through the noise and figure out whether one of these measures is useful to you and your research goals?
In this brief article, we offer some guidance for using some of the most popular biometric measures, helping you decipher when to use them, what they mean, and what additional information you might need to make them useful.
Before getting into specifics about each measure, there are two critical big-picture points to keep in mind:
- Biometric measures are just another tool in the research toolkit. The key is incorporating them in a way that enhances your ability to tell a story about the user’s experience. For the most part, these tools provide insight into either attention or emotional responses (but see below for some subtlety).
- Biometric measures will always need additional context in order to be meaningful. That is, while you can generally take a user’s think-aloud comments at face value regarding the placement of a button, the same is not true of that user’s sudden increase in heart rate. To make sense of the latter, we generally need to layer in additional survey or interview responses on to our research strategy in order to place users’ emotional responses in context.
One relatively easy entry point into biometric measurement is to capture general arousal. Changes in heart rate and galvanic skin response (GSR) both indicate an increase in sympathetic nervous system activity. So, if a user sees the new version of your mobile app and has a spike in heart rate, you know that something about the experience was cause for arousal.
However, increased physiological arousal is nearly impossible to interpret without additional context. That heart rate spike when viewing your app could indicate either strong positive or strong negative reactions. In some cases, the mental effort of orienting to a new interface is enough to trigger an increase in heart rate.
The best analogy is to think of the pitfalls of using a polygraph (“lie detector”) in a criminal case. If a suspect is being interrogated about stealing money from the cash register, her heart rate may skyrocket because she did it, or because she’s terrified of being falsely accused. In both cases, the solution is to pair arousal measures with a careful examination of thought processes. What was the user doing when his heart rate spiked? What did he say? Did it happen in parallel with a frustrating dead end in the app, or with a delightful new shortcut? Accurately recording the responses to such questions can provide the context needed to draw insightful conclusions about users’ biometric reactions.
Barriers to entry are relatively low for general arousal measures. The equipment is inexpensive, and the concept is more or less intuitive. However, if you do not have access to an expert in physiology on your team, a brief workshop can help you understand the subtleties that underlie changes in arousal.
Measuring Facial Expressions
Another category of biometric measurements involves automatically coding subtle facial expressions to infer emotional states. This work is based on a rich research tradition in social psychology, showing that emotional states map onto distinct facial expressions (see Paul Ekman’s work, in particular).
Using this tool to understand your users requires specialized software that uses artificial intelligence to analyze facial expressions in real time. Given the subtleties of designing the research and interpreting the emotion data, it is best to have an expert involved in the project. If done well, the payoff can be fantastic; facial expressions can provide a more “pure” glimpse into emotional reactions to everything from political campaign ads to package designs. That is, your test user might tell you both packages are equally appealing, but this software will pick up the unmistakable expression of joy over package B (and perhaps disgust over package A).
Measuring Eye Movements
Eye tracking, as the name implies, tracks users’ eye movements as they examine a stimulus, whether it be a website, app, automobile dashboard, or automated teller machine. These “tracks” provide unmatched insights into several key questions: What first captures people’s attention? How do they scan the page? How long do they have to look around to find the search bar? Eye tracking is particularly useful in cases where these questions are critical to your product development.
Eye tracking has recently become more popular in user research, as the equipment costs have decreased from astronomical to merely high. Jokes aside, many research vendors have purchased the required equipment and can incorporate it into projects for a reasonable fee. As with facial expression analysis, eye tracking also requires the involvement of an expert (either in-house or as part of the vendor’s team). Specialized software can pull out quantitative metrics such as time spent in defined “areas of interest,” and the amount of time it takes before the user first “fixates” on one of these areas. One alternative that both simplifies and cuts costs is to limit analyses to a “qualitative” approach—getting an impressionistic view of the visuals, rather than crunching all the numbers to understand areas of interest.
Measuring Brain Activity
Brain activity is most frequently measured to determine the mental effort or attention required by a user interface. It can be much more sensitive to small differences between designs than common non-physiological measures such as self-report ratings of difficulty, time on task, and the number of errors produced.
There are multiple ways to measure brain activity. Techniques such as fMRI (functional magnetic resonance imaging) or PET (positron emission tomography) scans allow researchers to image both the structure of the brain and the blood flow within the brain. Researchers ask participants to complete tasks while these brain images are recorded, and thus can determine what areas of the brain were most active during those tasks. These techniques, however, require specialized equipment and trained personnel thus typically cost prohibitive for most UX or market research.
EEG (electroencephalography) is a technique that allows researchers to record the electrical activity of the brain via electrodes attached to the scalp. Participants are presented with stimuli or are asked to complete tasks while the EEG is recorded. Patterns of activity at different locations are then examined, depending on the goals of the study. For example, you might examine brain activity in areas related to emotion regulation or attention to understand reactions to a new interface design. Increased activity in an “attention area” (a slight oversimplification) would tell you that users were increasing their focus on the new design. While EEG does still require specialized equipment and trained personnel, the equipment and software needed for such studies are orders of magnitude less than for other means of measuring brain activity, thus, there is a much lower barrier to entry for this type of study.
In sum, biometric measurements can add valuable layers to your user research by providing insights into attention and emotional reactions while users interact with your product. But remember our big-picture points from above: 1) They should be used when they help tell a story about the user experience, not simply because you feel pressured to try the newest technique; 2) They should always be used in context, so that measures of arousal, eye movements, facial expressions, or brain activity are interpreted alongside survey, interview, and behavioral data.
Have questions about how to get started? I’d love to help!