Usability testing is a method in which representative users attempt tasks with a product while researchers observe to identify usability problems and measure performance. In moderated sessions, a facilitator guides the participant through tasks in real time, asking follow-up questions and probing for insights, whereas unmoderated testing has participants complete tasks independently, often remotely through platforms such as UserTesting or Maze, making it more scalable and cheaper. A common technique in both settings is the think-aloud protocol, in which participants are asked to verbalise their thoughts continuously while performing tasks, revealing their mental model, expectations, and points of confusion.
Three key usability metrics anchor most tests. Task success rate is the percentage of participants who complete a given task correctly and serves as a primary measure of effectiveness. Time on task measures how long it takes to complete a task, while error rate captures how often mistakes occur; together with satisfaction ratings, these provide a robust view of usability. Beyond single-session tests, teams often compare designs through controlled experiments. A/B testing, also called split testing, compares two versions of a page or feature, with the existing version as the control (A) and a modified version as the variant (B), to determine which performs better on a specific metric. Statistical significance, typically reported at a 95% confidence level or \( p < 0.05 \), indicates that the observed difference is real and not due to random chance, and the required sample size depends on the baseline conversion rate and the minimum detectable effect, often running into hundreds or thousands of users per variant. When several variables need to be considered together, multivariate testing examines multiple variables simultaneously, such as headline, image, and button colour, to find the optimal combination, though this approach requires much larger sample sizes than A/B tests.
Two complementary methods address how information is structured and found. Card sorting has participants organise topics into categories that make sense to them: in an open card sort, they create their own category names to discover structure, while in a closed card sort they sort cards into predefined categories to validate an existing structure. Tree testing evaluates the findability of topics in a site's hierarchy by asking participants to locate items within a text-only tree, without visual distractions. The two methods work well together: card sorting helps create the information architecture, and tree testing validates whether users can actually find content within that structure.
Several lighter-weight techniques offer quick insight. A usability benchmark is a baseline measurement of task success, time, and satisfaction that can be compared against after design changes. The 5-second test shows a design for five seconds and then asks what the user remembers, testing whether the key message and purpose come across at a glance. A first-click test measures where users click first to complete a task, and research shows that users who get that first click right succeed around 87% of the time. Remote usability testing conducts sessions with participants in a different location, either moderated via video call or unmoderated via a testing platform, while guerrilla usability testing takes an even more informal approach, recruiting random participants in public places like coffee shops for fast, low-cost early feedback. Throughout a project, evaluation can be formative, conducted during design to identify and fix issues, or summative, conducted after design to measure overall usability against benchmarks.