Decision making models: applying heuristics to search results designs
TL:DR
Existing vehicle search results design didn’t help users differentiate between options - they were too similar in presentation, and decision making stalled (no pun intended). Users who already knew exactly what they wanted were fine, but users less familiar with car models were getting lost. Different vehicle manufacturers have different ways of differentiating between models, so the data sets were complex and varied.
I deep dived into decision making models, looking at the different frameworks users apply when narrowing down from a very large pool of results. I used the physiological principles to ground the logic behind the more advanced search presentation
I collaborated with experts and the tech team to identify how we could broaden the journey for users, giving less informed users options to explore and navigate by vehicle model, and identifying different search values that could help narrow down the search results to become more relevant.
The existing design - very similar results offer little to a user to shorten the list
Existing issues
The car leasing website has a limitation within the search results page. The image and vehicle name alone don’t offer enough to help a user differentiate between the search results available to them. The naming conventions used by different manufacturers are complex, and often not transparent enough for inexperienced or low-confidence car buyers to see the differences in the results.
I had concerns that the initial design for the MVP of the website didn’t do enough to enable a user to create a shortlist form results. We had introduced a wish-list feature on the website, but the search results appeared too ambiguous to allow a user to choose what to add to their consideration set. I gathered feedback from a variety of sources, both internal and external users of the website, and from analytics showing wish-list use, and confirmed that this was an issue, particularly when searching for popular models like Audi and BMWs.
Understanding how users select their shortlist based on their criteria was key
Understanding the problem
I started by understanding the data. Different manufacturers treat models in different ways, so there are a wide range of variations within each model group. I explored the many approaches used across manufacturer and broker websites. There was no consistency in the approach, and there were no clear conventions to follow.
The next step was to understand the way users approach this type of decision making. I studied research on decision making models. In particular I explored how users create a consideration set of potential options before making a decision. I identified decision making heuristics that would be relevant for a user who is shortlisting and comparing cars, for example disjunctive and conjunctive rules. This gave me a basis for how to assess any solutions.
Fully functioning prototype
Testable prototype
I arranged an informal workshop with key stakeholders. I presented the background and context, and led a discussion about the potential solutions. I had already identified some options, but used this as an opportunity to engage stakeholders in the design approach, and have them collaboratively identify solutions.
The output of the workshop was a Model page, which allowed a user to include or exclude filter options specific to their criteria., like fuel type or transmission type. It used a simple introduction to identify the generic Model features, as well as clearly differentiating between different categories of trim.
I took the sketches, comments and technical input from the workshop, and produced a clickable prototype. I first shared the wireframes with stakeholders for initial feedback, made a few tweaks, and then developed simple interactions for the prototype.
Outcomes
I worked with the tech team to identify a way of reusing existing APIs and functionality to create the desired functionality. In technical terms it would be virtually identical to the existing search, but would provide an interim step for users, allowing them to have a more manageable selection of options to add to their consideration set.
The next step will be further testing with users outside of the business, and working with the Product Owner to prioritise this story within the backlog.