What is Guerrilla Research?

Guerrilla research (guerrilla testing, hallway testing) is informal, opportunistic research conducted quickly and cheaply outside the controlled conditions of a planned study: approaching people in public or semi-public places, asking for a few minutes, showing them a prototype or a concept, and learning from their immediate reactions. The name signals the trade: minimal recruitment, minimal setup, minimal cost, in exchange for minimal control over who participates and under what conditions. It is a member of the usability testing and concept testing families at their most reduced, and its legitimacy depends entirely on matching the claims to the method's modest reach.

What It Is Good For

Catching the obvious early. Confusing labels, unfindable buttons, a concept nobody understands from the first screen: problems so basic that almost anyone reveals them, and so cheap to fix at sketch stage that waiting for a formal study is waste.

Testing comprehension and first impressions. "What do you think this does?" asked of ten strangers with a landing page is a fast signal on clarity, precisely because strangers bring no context.

Building the habit. Teams that have never watched a person use their product learn more from an afternoon of hallway tests than from a quarter of dashboards; guerrilla research is often the gateway to the real thing.

Rehearsing the study. A rough run-through of tasks and questions with whoever is nearby is a legitimate pilot for the formal version.

The Hard Limits

The wrong people. Café patrons and colleagues are a convenience sample of whoever was around, not the target audience; for any product whose users have specific expertise, context, or needs, guerrilla feedback measures the reactions of people who would never use it. The more specialised the product, the less the method can say.

The wrong conditions. Ten minutes, standing, in public, with a stranger watching: nothing about the setting resembles real use, and complex tasks, sensitive topics, and anything requiring the participant's own data are out of reach.

The wrong claims. Guerrilla findings are directional and design-level; they cannot support prevalence ("most people..."), preference between polished alternatives, or anything a stakeholder would fund a roadmap on. The failure mode is not running guerrilla research; it is presenting it as something else.

Consent and courtesy. Public intercepts still require clear consent, especially for recording, and an easy no.

Keeping It Honest

1. Use it for the questions it can answer: comprehension, gross usability, first impressions, and rehearsal.
2. Screen even lightly ("do you ever do X?") so the most irrelevant participants are excluded.
3. Keep a fixed script and record reactions consistently, so the results are at least comparable across participants.
4. Label the findings as guerrilla in every readout, and let them commission the proper study rather than replace it.
5. Notice when the cheap alternative has become the better one. Much of what guerrilla research was invented to do (fast, cheap feedback on a prototype from real people) is now done faster and with the right audience by remote unmoderated studies: a Ballpark study can put a prototype in front of screened participants within hours, with recordings, which delivers guerrilla speed without guerrilla sampling. The hallway remains useful for rehearsal and for the very first reactions; for anything the team will act on, the screened remote round is the honest default.

The Takeaway

Guerrilla research is fast, cheap, and uncontrolled: excellent for catching obvious problems, testing comprehension, and building the habit of watching users, and unable to say anything about specialised audiences, real conditions, or prevalence. Use it for what it can do, script it, label it, and let it point toward the study that matters. When the same speed is available with the right participants, take the right participants.