Agile Usability Testing - Using AI
Conducting usability testing and evaluation for a new product is a necessary step for making smarter, data-driven decisions. However, it can also be cumbersome and difficult to execute within a startup environment with tight timelines, fast learning, and decision-making. This is where the use of artificial intelligence (AI) in the design process comes in. To succeed in providing valuable products in no time, we use Attention Insight - An AI product which allows us to obtain heat-maps and contrast mapping, Clarity scores, and benchmarking for UI screens. With its help, we can "predict" human attention, instead of traditional mouse tracking which is more indicative of the perceptual level.
Case Study
In our recent design project with a new startup in the Cyber Security industry, we incorporated a data-driven approach with the help of Attention Insight, alongside qualitative assessments and stakeholder feedback. Our biggest challenges were the organization of the main dashboard which contained big complex visualizations and their color intensity. At first, we compared wireframes to finalized UI screens to ensure the transfer between concept into the detailed phase and alignment with customer requirements. Second, we compared 2 versions of the UI screens, and by that, we could control the coloring and contrast of the elements inside the dashboard until the desired result. This method facilitated data-driven decision-making regarding UI and UX design elements, providing a structured framework for our design process. Eventually, this allowed us to conduct AI-based usability testing in less than a week and saved us a lot of time and effort in order to obtain additional data-driven feedback on the design.
Reservations
Although using agile usability testing is time-saving and data-driven, it is important to note that usability testing with real users is still necessary and crucial because of a few reasons. First, it is more reliable and representative than an AI model, especially when the data the model is trained on is different from the one examined. Second, qualitative feedback like users' satisfaction, feelings, thoughts, and sayings are very important and help us learn the users' mental models. Third, a predictive attention tool will help us better understand how and where the product attracts attention, but it doesn't explain the gap between attention to interpretation and taking action, which real-user usability testing provides.
To sum up, in the fast-paced and resource-constrained startup environment, this predictive-attention solution provides us with an additional layer of knowledge and insights over the design process and assists in the process of making efficient and effective decisions.
.png)
.webp)
