Seen on Twitter:
I’ve tested applications like that, where the Web pages are filled with tabs full of edit boxes crammed into the space with inscrutable labels that, I’m assured, the users know what they mean.
Sometimes, these interface designs come straight off of some crowded paper form that a worker would fill out with pen, checking boxes and putting tic marks or numbers in boxes. On paper, this is as quick as moving your eye, moving the pen, and pressing down. With a screen, it’s a little different, as it involves tabbing or moving the mouse and clicking and then typing something, scanning the form, moving the mouse, clicking, typing some more, and so on.
Other times, these interface designs pretty directly capture what the worker saw in a mainframe application or in a terminal window connecting to a mainframe application. With Windows or Web-safe colors instead of amber text on a black background that the worker. A lot of needless tabbing because interfaces could not easily branch. If you check this box, then these blank spaces become relevant. No, paper couldn’t do that and mainframes couldn’t do that, so the new Windows or Web application won’t do that. Because the users are used to it.
- The workers (“users”) today aren’t the users of tomorrow; if you’re not designing the interface right because it’s good enough for the grizzled greybeard who’s been around forever, you’re not appreciating how much easier you could make the process for n00bs. That is, probably most of the users. Especially if you’re writing software for a company that’s okay with this sort of interface. I imagine it has a lot of turnover and a lot of people getting trained to do it the hard way just because it’s always been done.
- Notice that we use the term “users” a lot in relation to people who work with the software we build. That’s defining them in terms of their relationship with our software, but their main jobs are doing something else. If your software design captures workers and traps them into being users too much, it drags on their productivity. Computer software should make their jobs easier and more streamlined, not slower than working with pen and paper. Sure, you can say that the data collection for analysis on the back-end is the driver for the software, but that doesn’t mean you should ignore other efficiencies you can introduce with a good (or at least better than this) design.
I read somewhere recently about hiring tester with test skills rather than domain knowledge, and sure, that’s the right balance, however, domain knowledge is what allows you to spot these sorts of problems. You might be hired because you’re a good tester, but you ought to study up on the industry whose software you’re testing. Me, I’ve been known to refresh myself on the basics of chemistry to better test chemical modeling software and to grok at least a little bit of the workflow of a warehouse when testing order fulfillment software.
Because otherwise you’re only logging the defects qua defects like “The Tare Weight edit box allows alpha characters” and not the higher level concerns about why you’d expect a worker would enter the total shipping weight before the number of items to ship.
Domain knowledge gives you the insight about the worker’s starting point in your software and what he wants to do to get done with your software. And that will give you the possible paths for his interaction without having to make all the possibilities available on one screen in tiny print.