For this recap, we welcome Girl Geek and September Dinner attendee Shan Lian as our recapper. She kindly agreed to step in when Melle was off sick and PJ was swamped managing logistics. Thanks, Shan!
Communitech graciously hosted us for the evening as we settled ourselves in with special guests MaxMixology, a local startup at the Velocity Foundry that is developing a “Keurig for mixed drinks”. We served as user testers as they served us whiskey sours and margaritas made by their machines. Beta testing should always be so festive!
Entitled “Data, Go Home, You’re Drunk”, Danielle’s talk led us through the dangers of letting data take over design. Especially in today’s data-driven world, it is common to think of data as the ultimate source of truth – neutral and correct. But Danielle warned us of the ways data can be biased, can mislead, give us hollow answers and have no conscience. It is up to people (designers, in this case) to use data as a tool to aid in decision-making, but never to let data lead.
There are two types of data, qualitative and quantitative. As a product designer (especially one in a data-driven company like Facebook), Danielle will always end up turning to quantitative data to justify her decisions. She jokes that designers will become obsolete in the next few years – as they will have just become requirements gatherers, turning assets into systems based on what the data spits out.
But she argues that a designer’s years of experience and intuition can’t be all wrong. Despite all the goodness that we often think data is, data is not the end-all to our problems. In fact it can cause more problems than it helps – just different ones.
Data is not neutral. It is biased because the people who collect and interpret the data are biased. There are many ways we can manipulate data, even if unconsciously. Incomplete data collection can mislead; reading data in isolation can tell a different story. Remember, data itself is only just a moment in time. “If you torture the data long enough, it will confess to anything.”
Data is misleading. Data only shows what we ask it to show, and it is narrow and focused on one task. Often, peripheral influences are missed and outcomes are dramatically changed when data is followed blindly. It is up to designers to watch the entire ecosystem to gain insights, using data only as sources of knowledge to help them lead.
Data only asks the what, never the why. Data does not show the emotions behind the numbers. It cannot explain what happened and it cannot decide what happens next. Don’t mistake data with insight; it is up to designers to ask the “why” and discover reasons behind what the data is saying. Not everything is measurable.
Data has no conscience, no taboo. Data shows what you ask it to show. It is important to establish values for the company or product, and it’s important that the way data is being used supports the values that are set. Even if it means proactively not collecting the data to begin with.
We must be wary of data and avoid its pitfalls by designing how we collect and use it. Establishing clear goals is number one. From that we must identify biases, validate, and segment appropriately. Too often the emphasis is placed on being measurable, instead of ensuring that important things are measured. Data can’t tell you what to build, it can only tell you the results of what you built.
Data should not be the only factor swaying decisions and guiding direction. Use qualitative data to gather requirements and new ideas that will help build your product, then use quantitative data to test what you decide to build.
And with much food for thought and a more critical eye toward our gathering and use of data and decision-making processes, we concluded the evening’s festivities.
Our October Dinner is coming up next week on October 22nd. There are still a few tickets left, so join us and learn all about the Internet of Things. How our stuff is connected, what it can do, how we should manage security, and more!