Ashwin Purohit

Essays & Articles

The Anxiety of Usability Test N°1

I’ve been working on a language-learning program in my spare time, and I just paid for the first real usability test of the product: a random person interested in learning French, who will send me a screencast of his experience with my program. It’s the 8020 solution Steve Krug espouses in “Rocket Surgery” – an online usability test that I don’t have to be present for. These people are “professional” usability testers, so they think out-loud about everything, no prodding required.

I have been working on the program for months. Developing it in Go, hosting my assets in the Cloud, using a Fab script to deploy and painstakingly linting my code. I normalized and noise-removed the French audio transcriptions. I made good technical decisions.

But none of this has prepared me for the anxiety of having a true stranger evaluate my program. While it is running, I have to escape the house just to think about something else. And now I come back, and the test isn’t uploaded yet, and I pace around the kitchen gnawing on an apple, which might as well be my fingers. I feel like I’m waiting for a herpes test.

Why am I so anxious? I think I have wagered too much of my ego on the results of this test. I don’t want the tester to say “I don’t even know where to click,” or, “What in the hell is this? I don’t understand,” or worse, “I would never use this to learn French.” Because that would mean rejection, that something I’ve created is terrible and deserves ridicule. It’s only my higher brain forcing me to do this now, cold-comforting me that “You’ll only have wasted more energy if you do this later.”

I think this is why so few small groups are keen on usability testing. It’s so good it hurts. Just a few usability tests can dismantle your entire project, prove that it sucks, and force you to make enormous changes. The architects know they should do it, but they’d rather stick their heads in the sand.

That way, when it fails (and it likely will), they can write it off on some cultural phenomenon or competitor, something intangible, rather than accepting that they purposefully and consistently disregarded the writing on the wall. I’m not going to be like that. Here goes.


After four months and fifteen usability tests, I can’t live without them. Every time I introduce a new feature, smooth an interaction, or fix a bug, I need the voice of a real person to say it’s okay. Or to not speak, in the case of a smoother experience.

I’m amazed at how watching a half-second stutter from someone using your program illuminates some problem that’s been slinking around. The highlighted problems can be facilely fixed, like when a user says, “I’m going to click here, and oh … well, I guess I should click this instead,” meaning, “that’s not what I expected, this sucks.” They can ruin weeks of infrastructure work when a user says “I don’t understand why answers are presented to me in this order,” and you have to rework your queueing system.

But if you’re trying to sell something to someone, the peace-of-mind that good usability provides you is worth the rework. What craftsman does not hone?

And if you should ever find empty the list of work-to-do, just go back and watch a recent user testing video. Within are piles of ruffled feathers that need smoothing.

Software Engineer building back-end systems for ML tasks.

Previously: Stanford E.E., senior software engineer at 23andMe, lead software engineer at Recursion Pharma.

Want updates when I write?