An alternate perspective on data-driven decision making

The O'Reilly Radar Podcast: Tricia Wang on "thick data," purpose-driven problem solving, and building the ideal team.

In this week’s Radar Podcast episode, O’Reilly’s Roger Magoulas chatted with Tricia Wang, a global tech ethnographer and co-founder of PL Data, about how qualitative and quantitative data need to work together, reframing “data-driven decision making,” and building the ideal team.

Subscribe to the O’Reilly Radar Podcast

TuneIn, iTunes, SoundCloud, RSS

Purpose-driven problem solving

Wang stressed that quantitative and qualitative need to work together. Rather than focusing on data-driven decision making, we need to focus on the best way to identify and solve the problem at hand: the data alone won’t provide the answers:

“It’s been kind of a detriment to our field that there’s this phrase ‘data-driven decision making.’ I think oftentimes people expect that the data’s going to give you answers. Data does not give you answers; it gives you inputs. You still have to figure out how to do the translation work and figure out what the data is trying to explain, right? I think data-driven decision making does not accurately describe what data can do. Really what we should be talking about is purpose-driven problem solving with data. You really have to figure out the hard questions with an organization: what is their purpose, what is the problem they’re trying to solve, and then how to use data to help that.

“Oftentimes, that’s a first step: is data going to help you with this? If it’s not, if it is just a problem with perception, if it’s a problem in messaging, then we need to be able to say this is not a problem that data can solve. If there is a clear way that data can contribute, then you figure, okay, what are the most important questions we should be asking? What are the questions that aren’t being asked? Because you have, ideally, a data scientist and a qualitative researcher, like an ethnographer, at the table. Everyone should be data agnostic in the beginning; everyone should be pushing the organization to think about the problem.”

The value of “thick data”

Wang has a term for bringing together qualitative and quantitative data: thick data. She outlined her thinking behind the term and the path that led her to it:

“I was thinking, in a room of data scientists, how do I quickly explain the skills and benefits that ethnographers can bring to the table? I wanted something that would speak quickly and also communicate the value of it beyond usability. One of the problems is that people associate qualitative research with usability; the problem with usability research is that it’s very downstream: you’re already given a product, and you’re there to optimize. You’re not there to actually discover and ask new questions because whatever assumptions you have are already built into the product. I wanted to quickly communicate that the value of ‘thick data,’ of research, is that if you do it up front by asking the hard qualitative questions, it can be used for discoverability.

“I wanted to show that the opposite of big data is not small, and this is the number one problem: I felt there was a perception issue among data scientists that the opposite of their work was small; it was puny data, not as valuable. So, how do we make it sexy and how do we do an elevator pitch? I thought, why don’t we call it ‘thick data,’ because it’s not small; it’s thick. It’s a thickness in stories, it’s a thickness of depth, you feel the texture of someone’s emotions.

“It also dovetails really well with Clifford Geertz, who is a very important figure in the founding of anthropology. … He said that the ethnographer’s job is to create thick descriptions. I thought, well, this is really lovely because it works with the history of ethnography, and we’re here to deliver thick data. I feel like that would give a common framework to compare big data alongside thick data. … I was just trying to do my part to meet the data scientists halfway and to make it just as sexy and to start creating a common language.”

Building the ideal team

Effectively bringing together qualitative and quantitative data, Wang said, requires bringing the right people to the table, which in turn requires companies to break traditional molds:

“The old way of organizing companies is not very compatible for agile research because usually the quantitative and qualitative people have been separated. This kind of tight integral thinking requires multiple forms of expertise at the table.

“A really tight team, an ideal team, would always have someone who’s strong in data science, someone who’s a strong ethnographer, and then designers — working with designers is really great because it forces you to always think about how to make data usable for execution. Designers bring so much to the table; a lot of problems can be solved with design, not necessarily data. You need someone to push you on that.”

You can listen to the podcast in the player embedded above or download it and subscribe through TuneIn, SoundCloud, or iTunes.

tags: , , , , , ,
  • Alex Thomas

    Could the meme ‘Deep Data’ also be useful?