« Voices in your hand: one step back is two steps forward | Main | Flocking to Flock »
October 19, 2005
User-centered design for enterprise software
I just came across an excellent article that proves that user-centered design methods can work, even in the so-called “enterprise” space of corporate software.
Clean, cutting-edge UI design cuts McAfee’s support calls by 90%
Some key quotes from McAfee’s Development Manager reveal something crucial about their development process:
“We knew that the sweet spot [for ProtectionPilot] is the sysadmin who’s managing 250 machines,” Ries says. “These guys have a lot of different jobs, and they don’t have a large staff; they generally have one IT guy who doesn’t have a lot of time.”
“At the outset, we had two big concerns: We designed ProtectionPilot so that the person using it would need to spend less than hour a week maintaining his organization’s antivirus protection.”
That “sweet spot” of course, is exactly what most companies cannot or will not identify: how many users will use this software at a given customer? How many machines do they need to support? Exactly how much time per week will they use the software? In so many cases, a product manager will hedge their bets: “we need this software to work for people who use it once in a while or every day. And it needs to be “intuitive” whether the user has one or two peices of content they care about, or 10,000.” Of course, at that point, the design of the product is pure guesswork, with no useful information about how to prioritize features, or help users with tasks.
“Second, because of the large number of users, we wanted to keep our support calls low. No matter how trivial, every support call has some costs associated with it. Installation should be easy, and initial use should be intuitive.”
Yes, a clear business goal: keep our support calls low. Ok, that’s probably an goal of every software business on every release, but here it’s not simply a vague wish, it’s an actual metric. How many calls are the target number? About 2 per day. What product features accounted for most of the calls in the past? Installation and initial use, so the UI should focus on those very strongly.
Despite the title of the pice (“Clean, cutting-edge UI design”)there’s really nothing “cutting-edge” in the product’s interface (at least that you can see in a screenshot). And, of course, that’s exactly as it should be. Once-a-week virus protection maintenance is the entirely wrong place to experiment with new interface ideas.
The rest of the article is 23 “tips” for UI design from the McAfee designers. I’d probably call these “development process sea changes”, but that’s nitpicking. :-)
Some of the tips—more than a year after the article’s publication—stand out as abject heresy in enterprise software design: Involve users during product development, long before beta or even alpha. Show users prototypes of ideas before even one line of code has been written. “We sent the mockup to real IT administrators and asked them to play around with it…. A week later, we asked them to answer some questions….” “Over the span of the development, we gave them four drops of the product to try out in their environment.”
Most of these tips, to me, really feel aimed not at UI designers per se, but at product managers, who often remain the gatekeepers of knowledge about customers, handing out little tidbits about “our users” they might recall from month-old phone conversations, or noisy trade show booths.
Posted by Andrew at October 19, 2005 09:52 AM