90 percent of everything : Usability Blog

Written by Harry Brignull

There’s a persistent myth about guerilla user research that it’s perfectly OK to grab just anyone to act as a proxy for your users. Perhaps it’s something to do with the whole low-cost, lo-fi ethos that makes this myth so easy to believe.

Actually if there’s one thing you shouldn’t cut corners on, it’s your recruitment. User testing should mean real end users, not Bob from accounts or some random dude walking down the street. Let me explain why this is so important with a case study.

Picture this: a large institution designs and builds a decision-making app for their customers. Their customers are sales-staff in small, partner companies whose job it is to resell the larger institution’s products to joe public, along with a commission. Now, during the early stages of design they thought they were doing all the right things. Wireframing, low fidelity prototyping, and usability testing - but here’s the crunch - they did their user research on their colleagues within the institution. They included middle management, data entry staff, the receptionists, and even some of the staff in the canteen. In other words, everyone but the sales staff in the partner companies who the product was really for.

And so they ended up with a product that was really usable - even a naive first-time user could use the system to punch in the data and get product recommendations out.

At this point, having spent about half a million dollars, they were sure they were onto a winner. So, they decided to get some real end-users in for face-to-face research sessions. The first few users were very polite since they had strong professional and political relationship with the institution making the product. But then one of the users, an admin assistant, commented that they would never use this system in her workplace. Her boss would do all of this work on paper and in his head. Then, he’d give her his calculations and she’d double check them in Excel. Apparently, his figures were almost always spot on.

Sure enough, when all the other users were asked about their current practices, they admitted that they also did all their decision-making in their heads. In fact 3/4 of them said they’d never use the app, because it would actually increase their workload. The owners of the product suddenly had a crushing realisation. The app was highly usable but simply not useful - it was solving a problem that didn’t exist. In other words, they were screwed.

And so ends today’s lesson. Always carry out research on real users, or you might end up like them.

Out of Box Experience (OOBE) design is a discipline that consists of designing the details of the first few moments of owning a product, from the point of peeling off the shrink-wrap to the point of powering it up and trying it out for the first time.

I carried out a few OOBE design projects while I worked at Flow Interactive and Amberlight. A few months ago I got chatting to Pete Gale of CogApp, and it turned out he’d done some similar work for different clients. So, we got together and wrote this presentation for UXbri back in May. I’ve finally got around to putting it on Slideshare.

If you’re interested in how iterative research & prototyping is used in industry to create great User Experiences, then you should find this pretty interesting. (Using a feed reader and can’t see the presentation? Click here)

Clicktale seems like it’s matured quite a lot since it launched. They have an interesting take on analytics - instead of going for a pure stats approach like everyone else, they are trying more of a click & keystroke logging approach, where they track your users’ mouse movements and key presses, reconstructing this data in various ways for you on your Clicktale dashboard, including video playback of each user, heatmaps, and now, form analytics.

I originally tried Clicktale when they first launched, and although the video playback tool was very impressive, it also felt quite raw. You ended up drowning in video data - not quite knowing which videos to watch, nor how to get to the bottom of it all. I haven’t looked at their service in months, but glancing at it now, I can see that they’ve been working very hard at addressing this very issue.

I’ve been chatting to Kevin Arthur of Touch Usability (and Usability Dude at Synaptics in California), who has been filling me in about some of the new developments in tactile feedback. The little rant I was having about “before the act” tactile feedback is actually something that researchers in the field are very aware of and working on. There’s some really interesting research in the CHI’08 paper shown below, which is written by a Glasgow University research group (Hogaan, Brewster & Johnston). Check out page 3 where they talk about “tactons” (tactile icons) and fingertip-over, -click and -slip events. Kevin has pointed out that this approach can’t really be bolted on to the iPhone keyboard UI given the predictive text / hit area adjustment technique it uses. Interesting stuff.

You may have noticed that this blog has gone pretty quiet over the past few months. This is because that’s because I’m working pretty damn hard at the moment, consulting on the user experience design of the Madgex Job Board platform. If you’re wondering what that is, it’s a platform that powers a fair number of major job boards, including The Guardian, The Times, and about 100 others across the world. It’s various installations across the world receive in total roughly 2 million job seekers per week. In a sense, I’m the person who’s responsible for ensuring all those people are happy.

Job boards have much more to them than meets the eye - as well as the job-seeker facing UI, you also have a recruiter UI (where they post jobs and search resumes) and a very substantial CMS where the job-board owner can configure their site, manage marketing email campaigns and so forth. I’m currently applying an iterative research and design methodology each area in turn.

So, as you can imagine, my day to day work involves planning, running and overseeing a lot of user research, analysis and prototyping; plus whenever I have a spare breath I’m preaching the user-centred word. I’ve been doing a lot of interesting work on lazy registration, faceted navigation, and advanced search, but I’ve had to keep it all under wraps until the new version of the platform is well under way.

On 23rd September I’m giving a talk on Job Board User Experience Design at OnRec China. I know it’s unlikely I’ll see you there since it’s a conference for recruiters rather than interaction designers, but I’ll be posting my presentation on Slideshare in case you fancy a look.

Oh, and by the way, thanks for keeping me in your feed reader during the quiet past few months.

Quite a few new handsets are offering tactile feedback on their touchscreens, like the Blackberry Thunder, pictured above. I can’t help thinking that because the manufacturers can’t do multitouch properly, they are opting to enrich their feature lists in other ways.
Is it really a valuable feature? In concept, tactile feedback is wonderful. It is the reason that real keyboards are way easier to use than on a touchscreen. However, the execution is quite different. Consider the micro-interaction of pushing your finger down on a key on a real keyboard or keypad:

1. You line your finger up as best you can
2. The tactile landscape of the device tells you whether you have lined up your finger properly or if you are overlapping with other keys
3. As your finger presses down, you get a satisfying ‘click’, letting you know that it worked.

In other words, you get tactile feedback before the act as well as after the act. Most important is the feedback beforehand, since it allows you to adjust your action accordingly. The reason why all these new handsets are missing the point is that they only give after the act feedback. It’s a vague confirmation. Essentially it’s saying “Yeah, you pressed a key. You wont know which one, though, until you look at the screen.”

So, what should they be doing instead? Concentrating on honing the primary features. Adding enhancements that offer deep, lasting value rather than just an additional bullet point on the promotional materials.