Linx+Neemu+Chaordic is the lead provider of personalization for ecommerces in Latin America. Its solutions impact directly in user engagement and conversion, helping consumers find and discover products relevant to them. Busca Personalizada is one of the consolidated solutions responsible for this, providing a search mechanism that reads product information to rank them in an specific order and feeds an autocomplete field. The Product Experience team was briefed with the mission of delivering three new features in a tight deadline of two months to best attend the company's business interests. Me and another designer formed then a smaller team and we've conducted the process in lean UX design sprints.
Two of the new features were closely tied together by the need of giving our clients visibility about how the search mechanism ranked the products in order. One thing we needed to keep in mind was not giving away too much, since Busca Personalizada's algorythm was part of the "secret sauce" of the solution and what made it so great. The third feature had an clear overlap with an older feature and our precautions here were headed into making sure our users understood the difference between them.
INTERVIEW, ITERATE, TEST, REPEAT
Designers working in pairs on a process they're both used to it has the advantage of switching roles as the project goes to make it more dynamic. Per example, if I was busy making a script the next round of interviews, my colleague was already recruiting users and making sure we had a conversation scheduled. If I was compiling the last usability test, he was already sketching on white board the new solution based on the insights we've got. For each of the three features we worked the same steps as shown below and each sprint one of the features was handed over to the development team.
broadening the value without changing the scope
These features were mainly aiming for a specific client's needs. We could have taken this client words for it and delivered exactly what he asked for. Instead, we decided they should be useful in different scenarios for various users. That's why our sprints included multiple interviews, scenario-mapping and writing of user stories based on the new scenarios. Even if we couldn't spend time developing them at the time, it also gave the features a greater sense of purpose and a roadmap vision. Most importantly, what we would have developed anyway was now useful for a broader range of clientes because their needs were also listened.
Transparency and control
These features shared the similar value of giving the user control over the ecommerce search results. One important goal was that we should provide just enough control, not too much, not too little. Finding this balance was a difficult path, since too much customization from the user meant the algorithm was being subscribed by user input, and thus bringing negative impacts on conversion along with it. To give this sense of control turned out to be extremely tied to bringing visibility of the results the customizations had in the system. Each time an user changed the weights of his search results criteria, we had to show in real time how the result order turned out with clear communication of "this is before/this is after". Only by transparency the sense of control was fully achieved.