Corridor tests
на русском · in english
What are these tests for
So, at this stage you already have prototypes and formulated hypotheses. Now it remains to shake this bag and make sure that everything works as intended. Or rather, check the interface, icons, signatures, accents, and so on.
It’s simple
We have already formulated hypotheses, so we just write them out. We formulate tasks for users.
We are preparing a clickable prototype, if it is needed. We fix the “gates”, in other words, the conversion marks of the interface. For example: opening a form, successful completion, sending, and so on.
We are looking for participants for testing. We conduct testing and mark all the most important things. We draw conclusions and analyze them. Bingo, 100% percent you will change your attitude to your ideal prototypes.
1 step. Define hypotheses
Hypotheses are assumptions that we are going to test in the course of the study. It is better to formulate it according to the template: “If we do (the idea), then he will be able to positively influence (the criterion of success), since (why is this idea good)”.
2 step. Choose a testing method
All testing methodologies are divided into unmoderated and moderated according to the degree of independence of the task by respondents, as well as by the location of the respondent and the interviewer at the time of testing.
First click. The research participant follows the link and gets a very simple task. For example: “Where would you click to find information about leasing programs?”, “Choose leasing to purchase the selected car”. Only the first click is recorded. If the majority of respondents click “wrong way”, it signals that there are usability problems on the screen.
Side by side. The essence of the method is to compare two options and choose one based on the principle of better visibility or recognition of the target object. Respondents are asked to compare two images of the interface or select a specific element on two versions of the image. For example: “Find and select the A1 payment method”, “Which of the banner options is more recognizable for the material ‘Cars for leasing on ave’”.
For patency. The method is closest to moderated testing. It assumes the presence of the interviewer next to the respondent at the time of the test. We select conversion points and give the respondent a task that should lead him along the path you need. As a result, the facts of passing the points are recorded and notes are made about the difficulties encountered by the user.
Moderated testing. With such a check, the user performs the assigned tasks in relation to the product or service, while the researcher or moderator watches him in real time. Each of the above methods can be moderated. The reliability of the results and the depth of the insights obtained during such testing is considered better than when using methods without moderation.
3 step. Respondents’ choice
For testing, we need representatives of the target audience who are faced in real life with the tasks that are described in the testing tasks. However, the corridor test makes allowances in this matter, and absolutely all direct or indirect Internet users can become respondents.
Five to eight respondents are enough to identify interface problems. If all respondents face the same difficulty— this is a reason to stop testing, make edits to the prototype and start testing a new version.
If the resources and time to work on the site are very limited, we are guided by the principle: “less is better than nothing.” Three respondents are better than none.
4 step. Task formation
The quality of the results and their objectivity directly depends on the formulation of questions and tasks in testing. Checklist for the formulation of the task:
- Described accurately.
- Announced in full.
- Relevant to the respondent’s experience.
- It is possible to perform without prompts.
5 step. Preparation of the prototype
A test prototype can be:
- Interactive clickable prototype.
- A test version or a working website.
- Just a picture.
Checklist for testing the prototype:
- Scripts are executed.
- There are different options for executing the script.
- Texts, numbers and visual content are similar to the real ones.
- There are no errors and typos in texts and figures.
- There are no hints.
- It is possible to quickly return to the beginning of the script.
6 step. Testing process
Properly conducted testing will be less stressful for the respondent, will be easier to moderate and, therefore, will give more relevant results. Advice – shut up, listen to the respondent and do not prevent him from not understanding anything.
The respondent speaks most of the time during testing. The moderator speaks only if absolutely necessary. Any intervention by the moderator affects the course of the experiment and distorts the data.
If the respondent has stopped “thinking out loud” – remind him about it: “So… So, you think that... What do you see here? Tell me what happened.”
Questions often force the respondent to take a defensive position. You can replace questions with motivational phrases like: “Tell me a little bit about it... Describe it in more detail... Share your feelings... Let’s talk about it... Help me understand...”.
In what cases to help:
- The respondent should feel that you are encouraging him, not the product.
- If the respondent has tried several methods of action and asks for help.
- The respondent thinks that the task is completed, but it is not.
How to do it:
- Focus on the task, remind the purpose of the task.
- Make a general hint: “Remember how you started doing the task”, “You have already seen it”.
- To say directly what to do next.
It is important to record the moments where the respondent encountered difficulties. After testing, we ask the respondent a few questions. You don’t have to follow the direct recommendations for changing the site that the respondent will express during the conversation. Conclusions and recommendations are usually made from observing behavior, not from a conversation.
7 step. Recording of results and conclusions
Even at the stage of hypothesis formulation, we recommend creating a table containing decomposed hypotheses and tasks for testing with the content of user steps or beaten conversion points.
After each meeting, record the test result in the table: whether the hypothesis was confirmed, whether the respondent coped with the task without errors. We also add insights and other notes describing the user’s behavior at a particular step of the task.
Based on this document, we draw conclusions about the confirmation or non-confirmation of the hypothesis, as well as formulate new ones, if necessary.