“Post-truth,” data analytics, and omissions–are these design considerations?


Post-truth. The 2016 word of the year.

Yikes for some of us.

This got me thinking about UX around data, analytics, and information, and what it means when we present conclusions or advice based on quantitative data.

Are those "facts"?

If your product generates actionable information for customers, then during your design phase, your team should be asking some important questions to get the UX right:

  • What risk is there to our customer if the data is wrong or could be interpreted incorrectly [easily]?
  • What information might we want to include to help customers judge the quality of the information the product generates?
  • If our technology analyzes raw data to provide actionable information, are there relevant analyses that the product did not run that the customer might need to contextualize the conclusions drawn?
  • Is our product (and company) being genuine, honest, and transparent when appropriate?
    (That's how I roll at least, and few scenarios suggest this ever is bad advice.)
  • Is the display of supporting data considered and as unbiased as possible?
    (Notably: did you design the presentation of the information before coding it, or did you just dump it into a charting tool?)

Part of getting a product's design and UX right is knowing what questions to ask.

Let's take a quick example many of us without pensions can relate to: retirement savings.

Let's say you work at a financial institution and you're supposed to design an online tool that can help customers understand how much income they need for retirement, and specifically, what monthly savings target they should have in mind to reach that goal. A wise design-thinking product owner will be considering issues beyond how the UI works, the sliders and input fields on the forms, and the way the output charts look.

If we're talking about "truth" in the context of design, I'd hope the product team considered:

  • How confident are the displayed estimates?
  • Since this is a predictive tool, did the app run one than one type of simulation before generating advice?
  • Did the app factor in unique characteristics of the user, such as their own behavior to date (if known)?
  • Does the design clearly mention relevant variables that the tool cannot control for, and also how much those variables might affect the predictions that are shown?

How much, and how loudly a tool answers these questions depends on the content, risk, customer problems, and needs.

Sometimes these things don't need to be "answered" literally in ink because not all customers will care, or they might just assume that your calculator already does all this magic.  And, there are times when all the ink might just be noise (e.g. weather forecasts).

All that said, I am not sure "post-truth" fits in anywhere with good product design.

Want More Insights on Designing for Analytics?