Unlike other evaluation methods that require the judgment of one or more interaction design experts (so-called “inspection” techniques, such as heuristic evaluation ), the user test leaves the last word to people who through the real use of the product will identify the critical aspects of the interface (such methods are in fact called “empirical”).
During a test, users are invited to try “physically” a product (finished or in prototype form) by carrying out some tasks (called tasks). The difficulties that users will encounter during the performance of these tasks represent the critical aspects of the interface on which the design team will have to intervene to make the product easier to use.
For example, if during a usability test for an e-commerce many users will have difficulty concluding a purchase due to the poor visibility of the button to add products to the cart this element will represent a critical interface that will be corrected in order to improve the site.
When do the tests have to be done? Usability tests can be performed at any stage of product development. As mentioned above, it is possible to carry out tests both on existing and functioning products (this is the case of a test aimed at the complete redesign of a site) and on products still in the design phase, ie on prototypes.
Performing tests during the design phase allow the design team to identify the critical issues before the final release. The goal of the test is just that: identify in advance the problems in using the tool avoiding to run into expensive changes to be made after the release. The whole is therefore based on a concept (and a life lesson) very simple and familiar: prevention is better than cure!
How to design and develop a usability test. A usability test consists essentially of 3 phases:
Definition of the test protocol. The design team lists a series of tasks (tasks) that the user will have to perform on the product that is being redesigned. The set of these tasks represents the test protocol that will be administered (what a bad word!) To users. The tasks will be chosen on the basis of the more general objectives of the test but must, in any case, be thought about the most “important” (critical) points of the interface. In the example of an e-commerce, some representative tasks can be the purchase of a product, the purchase of a product as a gift box, the registration to the site, the identification of information related to the payment methods.
Test sequence. The test is conducted by an evaluator (or conductor) who submits to the participant (user) the list of tasks to be performed (the test protocol). The test normally takes place in a quiet room without other people. Usability tests are usually conducted using recording methods that capture both what happens in the interface and the user’s expressions. There is several more or less expensive software that allows you to record the test sessions. However, keep in mind that good results can be achieved even with simpler devices such as a camera and a clipboard. All you need to do is identify through the actions of users what goes and what is wrong with the product interface.
Analysis of the results. At the end of the test, the evaluator summarizes the critical issues emerged in a report accompanied by the user comments. Based on the number of users who took part in the test, some quantitative measurements can be made with respect to the different performances (how many users have successfully completed task 2?). However, what matters in this type of test are the qualitative pieces of evidence: what is it that users have made the most effort to understand and use? And why?
How many users are needed for a good usability test? And here we are at the $ 1 million question! The most famous answer is provided by Nielsen who states that with only 5 users it is possible to intercept 80% of the criticalities of an interface.
In reality, there are no exact answers and obviously, a lot depends on the type of project, on the objectives and not least on the available budget. In general, the number of users can vary from a minimum of 5-6 to a maximum of 24-32. If there is a need to test 3 different prototypes on a product intended for an international audience, it will be necessary to involve a large number of users. Everything depends on the cases, but what is important to understand is that even one test can be useful to bring out small or big problems that could have escaped during the design phase. The recommendation of a practitioner like Krug is, in fact, to do more test cycles with few users rather than a single cycle with many.