Thoughts: how data science and design research could collaborate
Prioritizing research questions with data science
I believe that in order to make good design/product decisions, we need to gather different kinds of evidence. These are some examples of how we could triangulate data points to make well-informed decisions.
Let’s say you just roll out a new feature, and you want to know what to improve. Instead of jumping right off to research, you could collaborate with the data scientist to prioritize research questions and the target participants by discussing these questions:
- Who keeps using the feature? What’s distinctive about them in terms of demographics or behavior?
- What activities are people doing when they’re using the feature?
- Who stops using the feature? What’s distinctive about them in terms of demographics or behavior?
- Are people more likely to try the new feature after seeing the announcement?
- Is there a big drop-off point?
Okay, then so what? Here’s a hypothetical situation to better explain how to use that in real-life.
Hypothetical situation
Imagine you are the designer/researcher at an Online Travel Agent (OTA) company that just recently released a “Pay Later” feature in which it allows the users to pay the installments over time rather than pay the full amount upfront.
*Hypothetical interactions
Product announcement > Pay Later landing page > Application form > Waiting for approval
Scenario #1
Let’s say you find out that only a few people visit the Pay Later landing page but you also find out that once people visit the page, they are more likely to apply. This might suggest that:
- People are not exposed enough with the feature (discoverability issue)
- The announcements are not working. E.g. people don’t understand or misunderstand them (usability issue), or the announcements aren’t engaging (desirability issue).
In this case, we could evaluate and improve the announcements or we could explore and test new entry points.
Scenario #2
Or in contrast, maybe you find out that lots of people visit the Pay Later landing page but few continue applying (stop on the application form page). This might suggest that:
- There are usability issues on the landing page or on the application form
- The feature is not relevant for them (e.g. they aren’t eligible to apply, or they don’t find it useful)
In this situation, we could identify those people and contact them for usability tests and interviews to understand why they quit.
Those are just a few examples of how data science and design research could collaborate. By spending enough time understanding the data before doing research, we could gather more informed hypotheses and prioritize the research questions better.
