Building a Practice of People First Data Systems

Mike Oren
Mike Oren
Jul 25, 2017 · 3 min read

Data is often looked at in aggregated forms to try to uncover the patterns that allow for predictive models to be built. It is often kept in isolation from the context of use. The high-level problem might get communicated from the business to the data science team. But the behavioral patterns that will ultimately decide the success of the product are kept in a black box.

Different organizations are “solving” this dilemma in different ways. Sending data scientists out in the field to collect the context on their own or partnering one or two members of the UX team with the data science team. These solutions are steps in the right direction. They are ultimately band-aids that aren’t sustainable. Machine learning is quickly becoming the backbone of an increasing number of experiences. They place the algorithm first, the human second.

Data science has a huge potential to increase efficiency and unlock new value. Without considering the human first, we risk creating a future of soulless algorithms. But that statement in itself will not help solve the problem. So in a tried Internet fashion, here are three tips to help you re-center your systems on the human.

  1. Collaborate when determining what predictive or prescriptive models will add the most value. It should be a collaboration between business, UX, data science, and engineering. You must understand the business value, user value, data possibility, and engineering feasibility. While a single individual can easily pretend to know all of these, the reality is that they’ll be decent at two of these at the most and likely over confident about the other dimensions.
  2. See users in their real environment, not just conversations on the phone. Again, this shouldn’t just be the UX person but a representative from each specialty. Each has a unique view of the world and that lens needs to be brought into the process. At the same time, observing the user in their real world environment helps build and promote empathy. Including everyone on every trip is not practical. At some point though every core team member should observe the user as early as possible.
  3. Tie everything back to a use case that came from real users. It’s not enough to just talk to users when deciding what to build. Nor are traditional usability studies necessarily the best approach. Although Wizard of Oz does make a nice comeback to reduce dev work, you need to be adding new twists to old tools in order to help understand. Do users know what action to take? Do they have enough understanding to have confidence in the action? Do users trust the system? And does the system and action fit into the larger social ecosystem of interaction?

These are things that must not be done in isolation or in one off efforts but should become part of the practices of all teams building data driven systems.

Suggestions for the next entry are welcome, but I’m currently leaning toward expanding the ideas of the third point. Especially the thinking about your system in terms of social context. This is part of a longer series that hopes to deliver on the larger promise of the title.

Originally posted on LinkedIn

Follow me @mikeoren on Twitter or linkedin.com/in/mikeoren for the latest.

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade