Why it shows up in your week
Stakeholders love the new comps. Five customers from your ICP cannot find pricing. That ninety-minute session just saved a relaunch argument that would have lasted three months.
In plain language
Give a participant a goal ("find pricing and request a call"), watch without teaching, and note where the path breaks.
You do not need a lab. Remote tools and five customers from the real ICP usually beat twenty internal opinions.
Moderated tests catch confusion in the moment. Unmoderated tests scale when you already know the tasks.
Fix, then retest. One round finds issues. The next round proves you removed them.
Recruit people who match buyers. Testing only on internal staff produces polite lies.
Facts worth keeping
- Classic guidance
- About 5 users per round often finds major repeated issues (qualitative)
- Formats
- Moderated, unmoderated, in-person, remote
- Inputs
- Prototype, staging site, or production
- Outputs
- Task success, time, friction notes, severity ranking
- Best timing
- Before visual lock and before major build spend
Not the same as
- A/B testingA/B tests compare variants at scale with metrics. Usability testing explains why people struggle, often earlier and with fewer users.
- Focus groupsGroups discuss opinions. Usability testing watches individual task behaviour.
- Analytics aloneAnalytics show where people leave. Usability testing shows the confused click before they leave.
Where it bites
Skipping usability testing hurts when a relaunch ships a beautiful IA nobody outside the company understands.
Ads then amplify a confusing path, and the team debates copy while the structure was wrong.
What to check
- Which three tasks represent real revenue paths?
- Are participants from the ICP, not only internal staff?
- Which findings change the sitemap or form before build continues?
- Is there time booked for a second round after fixes?
- Are sessions recorded with consent for the team that was not in the room?
Common questions
What is usability testing?
Observing real users as they attempt defined tasks on a prototype or website to find friction and failure points.
How many users do you need?
For qualitative finding of major issues, about five users per round often surfaces repeated problems. Run another round after you fix them.
When should we run usability tests?
Before locking visual design and before expensive build. Also after major IA changes.
What should you check first in a test plan?
Define revenue-critical tasks, recruit people who match the buyer, and agree which findings can change scope before visual polish.
Is usability testing only for big companies?
No. A half-day remote round with five customers is cheaper than a failed relaunch month.