research methods
Guerrilla testing
What is guerrilla testing, and when is it good enough?
Guerrilla testing is quick, cheap, informal usability testing - grabbing a few willing strangers (in a café, a hallway, a high street) and watching them try your design for ten minutes each. It trades the rigour and control of a formal study for speed and almost no cost, on the bet that a handful of fresh users will still expose the biggest, most obvious problems fast.
Also known as: guerrilla testing, guerilla testing, hallway testing, quick usability testing
Prefer to watch?
Watch the recap 0:53
The demo
Same goal - find the usability problems - two very different ways to get there. Flip between the formal study and the guerrilla test, and weigh what each buys you and what it costs.
What this demo shows (text version)
Two ways to find usability problems are compared across the same dimensions. A formal study takes weeks, costs a lot, recruits the right target participants, runs controlled tasks, and yields reliable findings and metrics. A guerrilla test takes an afternoon, costs about the price of a few coffees, grabs whoever's around, runs loose tasks in a noisy spot, and yields the big obvious problems fast but no trustworthy numbers.
The takeaway is to match the method to the question. Guerrilla testing wins for catching glaring problems early, cheaply and often - which is most of what early design needs - because the first few fresh users surface most issues. Its honest limits are a convenience sample, no task control and no reliable metrics, so step up to recruited, structured testing when you need the right people, controlled comparisons, or numbers you can defend.
You don't need a lab to find the worst problems - you need a few real people and a coffee shop. Guerrilla testing catches the big, obvious usability issues quickly and cheaply, which is most of what early-stage design needs. Its limits are real: a convenience sample, no controlled tasks, and no reliable metrics - so use it to find glaring problems fast, not to validate niche audiences or measure anything precisely.
An afternoon and five flat whites later, you'd watched five strangers all trip over the same button - a problem a six-week study would have found too, for a hundred times the cost and time. For catching the glaring stuff early, "good enough and now" beats "perfect and in two months" almost every time.
The case for it is the economics of finding problems. Most serious usability issues are obvious enough that a few fresh pairs of eyes will hit them, and the first handful of users surfaces the majority of problems. So a quick round - five people, ten minutes each, today - returns most of the value of a formal study for a tiny fraction of the time and money, early enough to actually act on it.
Its limits are equally real and must be respected. You get a convenience sample (whoever's about, not your actual target users), little task control, a noisy environment, and no trustworthy metrics - so guerrilla results are directional, not definitive. They're great for "is this obviously broken?" and useless for "what's our task success rate?" or testing a specialist audience you can't grab off the street.
So match the method to the question. Use guerrilla testing early and often to catch glaring problems cheaply and keep a feedback habit alive; step up to recruited, structured testing (or quantitative methods) when you need the right participants, controlled comparisons, or numbers you can defend. Cheap-and-frequent and rigorous-and-occasional are partners, not rivals.