Developing and launching an application is one thing. Seeing that it will be functional and beneficial for the target audience is another independent discipline that should precede it.
"Alena, do you click on the green box and nothing happens?" Well, these are the events we need to avoid.
Developing and launching an application is one thing. Seeing that it will be functional and beneficial for the target audience is another independent discipline that should precede it. User testing helps detect errors and tune the product to the satisfaction of its users. How do we see him?
What is user testing (UT)?
This is a method of analyzing user behavior when using a particular service. In our case, these are mobile or web applications. Sometimes UT is also referred to as usability testing, which is more accurate. Usability is the main priority of proper UX (i.e. user experience) design.
How to implement UT?
Testing can be done one at a time or in groups (laboratory testing). There are also online tools available today to simplify the work of test creation - Maze has been a good fit for us.
Online tools and questionnaires are suitable for quantitative research, we can even send the link to 100 respondents. In addition, respondents can complete the test in the comfort of their own home, which is a great benefit nowadays.
Laboratory testing, interviews are more qualitative. They allow you to promptly ask more questions and really go into the depth of the problem.
Testing can be done on an existing product if we want to discover its shortcomings, or on a so-called prototype in the case of a new product (1).
,,When testing an existing product, imagine you have, for example, an e-shop with natural cosmetics, few orders and you don't know why? At this stage, we will create a test on the current e-shop with potential customers and validate the result of the redesign."
In the case of a new product, both a wireframe (or also a low-fidelity prototype) and a high-fidelity prototype can be created and tested:
Wireframing (WF) takes less time. With WF we primarily focus on concept, structure, and functionality instead of colors, fonts, images, etc. Thus, we primarily test the user flow, the product idea and want to get first feedback from potential users. If we find flaws in WF, it is not time-consuming to make adjustments. The rule of thumb here is that the earlier we discover the flaws, the more we save time and therefore money.
High-fidelity (Hi-fi) prototype to be designed functional and interactive. Such a prototype is most similar to the final product and therefore more user-friendly for the user and will also present itself better to sponsors. With Hi-fi we will get more detailed feedback on specific design elements. The downside is that editing takes much more time than WF.
Based on the results of the UT, an evaluation will be made with a proposal for specific steps to be taken.
What is the goal of UT?
The primary goal is to identify and then eliminate the biggest errors and shortcomings of the service, based on feedback from potential or actual users.
Next, it is important to find out:
how users understand the app,
where they are lost,
if they find everything they're looking for.
Practice or How it works in our company
When it makes sense to us, we like to recommend testing to our clients.
One such practical example is our client who needed a mobile app to manage bike lockers. The project is huge. Therefore, we evaluated that it would be best to create a prototype of the mobile app and test it with hardware. Thanks to UT, we came across the most pressing issues of the product and then fixed them by modifying the design.
And what is our procedure?
The whole process can be summarized in the following points.
2.Test methods for data collection
Before choosing a particular testing method, we should clearly understand the target audience, the available resources (time and money), and the research objectives. This information will not only help us to select the appropriate test method but also to tailor questions and tasks for our test participants. (2)
A selection of methods that have worked well for us:
Guerrilla testing - we ask random people on the street, in a restaurant, in a café for their opinion on the main idea of the product, we collect feedback on the wireframe. It is advisable to come up with it at an early stage of the project.
Laboratory testing - we perform it in special environments (laboratories) and under the supervision of a moderator (testing manager). Either one by one or in groups.
Session Recording - this is a recording of interactions with the product by anonymous users. The recording data helps us understand what content/features are most interesting to users (through heatmap analysis) and also what problems users face when interacting with our product.
Maze app - the aforementioned online testing tool that offers a variety of data collection methods - from the heatmap, A/B testing to bounce rate. Each user has a unique ID, so you can evaluate the data of an individual, but also the whole group, Maze uses clear infographics.
3. Script preparation
The script preparation depends on the chosen test method. For laboratory testing, we devise a set of tasks that respondents will perform with a clearly defined goal. We have also found it useful to devise additional questions as the UT progresses.
Respondents are ideally selected from the target group that will use the app. The optimum number of respondents is 5 - reportedly they will detect up to 80% of errors. From 6 test subjects onwards, the error increment decreases. If we don't have that many respondents, the rule of thumb is that one is better than none. (3)
If you're testing an app that has been around for a while, it's worth testing with someone who has used the app.
Respondents are usually rewarded financially or with a voucher, for example. It is important to draw up a contract where the rights and obligations in the relationship with the respondents are stated.
5. The course
After thorough preparation, the actual implementation is the long-awaited icing on the cake. The procedure varies depending on the method chosen, but some attributes are best left unchanged.
Real-life illusion - It's always good to put testing in context - as if it were a real-life application. For example, tell respondents to imagine they have a bicycle that they want to lock somewhere safe because they need to go shopping. How would they use the app for this purpose?
Impartiality, unbiased, no influence - The test leader should not guide respondents to the correct solution, nor comment or evaluate anything in progress. Neutrality and reassurance to respondents that everything they do is neither right nor wrong is important. There is simply no right way.
Correct display - We test on the device for which the app is intended. There is always a test leader in the session with one respondent or with a group. We recommend recording the entire event while recording the results in an excel spreadsheet, for example.
6. Evaluation of results
Data is best evaluated using charts, heatmaps, or infographics. Of course, it depends on your budget, but if you have a bigger budget, we recommend infographics.
Infographics is a form of visual communication that uses graphic design to visualize data and information. These can then be easily and clearly read by the client.
We hope that a look under our hood has been an inspiration and an incentive for your own action. User testing should not be underestimated.
Helča wrote this article for you - our Product Designer.
The control question - what are cookies? Choose the correct answer.
Cookies are not sweets, but text files
We want to have an overview of how it goes on our website. But you have the power to affect how much we know about your visit.
As application and web developers, we are very interested in analytical data, so we will be grateful for your absolute consent.
Select your preferred cookie permissions, the basic ones are necessary for operation, others we can use only with your consent.
Your personal data will be processed and information from your device (cookies, personality identifiers and other data collected) may be stored.