Data Driven Personas - UPA 2007 Tutorial
Today I attended the full-day tutorial Data Driven Personas with Todd Warfel. Quite a relevant topic since I’m building personas for three clients at the moment and I really want to refine the technique.
Initially the thing that has struck me was there seems to be quite an emphasis on data, not surprising considering the title of the tutorial, but I mean quantitative data. Well at least I think that’s what everyone in the audience meant, as opposed to qualitative “data” (I use quotes here because I think knowledge might be a better word for this). It could be because typically US organisations have a greater maturity around the practise of collecting real quant data from CRM systems and the like. So this forms a foundation for the design research and persona development that goes on.
My clients don’t usually have anything this sophisticated—but I am willing to concede that they are atypical and it’s just like my luck that I have several atypical clients at the moment—and my design research is primarily qualitative with a bit of web stats to support it. So as we approached the lunch break, I was thinking this could be a problem for me, or it could just be a terminology issue whereby Americans say data and aren’t thinking just quantitative data.
On another note, Todd is quite a good presenter and his approach to consulting seems very similar to my own (and that of Step Two). Almost everything struck a chord in terms of methodology and the reason behind it. So much so, that something I have been thinking through over the last few weeks was kinda spoilt today…in that I thought it was rather innovative but turns out good people have already thought of it (stay tuned).
As it turns out, most people in the tutorial were thinking of quantitative (read “hard”) data, but Todd revealed that he actually meant a combination of qualitative and quantitative research. It’s a tightrope you walk when it comes to design research, and other IA/UCD techniques, and eventually the conversation turned to statistical significance and that’s when some people turned. Comments I heard signalled to me that the audience was a little disappointed by this and thought it was all a waste of time because it’s not “real data” or “proper research”. I’m not sure what they expected, did they think Todd was going to give them a silver bullet for personas? In that regard he made things difficult for himself by titling the tutorial as he did (and in fact he changed the title on the slides to “Design Research Personas”). I’ve seen this debate many times but to be honest I still don’t see why some people get so hung up on statistical significance and hard data, especially since they are not statisticians and probably wouldn’t know good data analysis if it bit them on the behind (note US PC lingo). Why do these people get transfixed by “data”?! I think it removes the responsibility for design…or maybe from doing any thinking at all.
I for one think the approach Todd talked about is the best approach. After all we’re not using personas for computational analysis, we’re using them to help designers design and to help businesses serve their customers. So it doesn’t matter that we build them based on good ol’ conversations with people rather than mining data from a “proper sample size”. We’re not dealing with machines, we’re dealing with people. Anyway, as Todd reminded us, it comes down to “garbage in, garbage out”. Having hugely complex data sources to drawn on is irrelevant if it’s not good data or if you can’t learn anything useful from it. That said, I was interested to hear about analysing customer service emails using a language analysis tool to discover issues and themes.
Rather ironically, the afternoon involved an exercise in creating personas, in groups, but without any research to work with. I think Todd’s idea was to focus on the actual writing of personas, rather than the creating of personas from research. So after complaining about people being fixated with data, I’m now doing it myself since I don’t have any data. We would have got better value by simply talking with Todd, rather than this exercise, which essentially encouraged what we have been trying to stop designers from doing.
Comments
Thanks for the thorough review, Pat. As someone who comes at this from the point of view of balancing theory with practice, I too struggle sometimes with the purist academic viewpoint of hard core statistical significance in data. It’s a balancing act in the real world. Additionally, at Messagefirst we’re continually walking the tightrope of balancing qualitative and quantitative data in our work. It’s a real shame that people today can’t recognize the value of both of these types of data.
Thanks for point out the areas it could have been better. As a practitioner, I’m always striving to improve what we’re doing. I’ll definitely take these things to heart for the next time.
So, next time, I’ll have some real data for attendees to work from. And I’m going to have them begin by developing personas at the start. Then we’ll introduce some techniques and data and have them evolve their personas. That will be the format used throughout the course, as I think there’s a great value in seeing them evolve through the process - something else that was part of the persona lifecycle.
[...] Yesterday I taught a day long tutorial at UPA 2007 on data-driven design research personas. Patrick Kennedy has written a nice good review of the tutorial. I say “good,” rather than “nice,” because it’s honest. And honest isn’t always all positive. While most of what Patrick says about the tutorial was positive, he does highlight a few oversights, or areas the tutorial could have been better. And for that, I thank you, Patrick. I’ll take honesty any day over being nice and PC. [...]
Post a comment