Background
The advancement in machine learning has made prediction models a hot topic in today's market. Supermarket chain MENY asked us to explore how prediction models can be used to make the shopping experience in the app faster, easier and better. This project was done trough an internship at NoA Ignite and is a proof-of-concept project.
My role
Scope
As we were two designers on this project, naturally both of us wanted to touch upon all parts of the project. However, I had main responsibility for the Figma prototype.
UX Design, Interviews, User Testing and Mapping, Market Research
Team members
Tools
Designers: Dag Takuro Hara
Developers: Lise Marie Nilsen, Jonathan Helgesen & Hary Pirajan Mahalingam
Developers: Lise Marie Nilsen, Jonathan Helgesen & Hary Pirajan Mahalingam
Figma, FigJam, Procreate, Azure DevOps
Understanding the case form both designers and developers point of view
An important agreement we had as a team was that all of us (both designers and developers) should have an equal understanding of our scope, and to agree on a concept based on the different knowledge each of us holds.
We started our process by mapping out how the MENY app is today, and where we could reach out with our solution. With this in mind, we did some idea generation through Crazy 8, and dot voted on the ideas we thought were interesting.
Our Crazy 8 session
From ideas to sketches
After the Crazy 8 session, us designers started to make some quick Figma sketches based on the ideas we dot voted on. We had frequent feedback sessions with all team members and moved quite quickly with our concept.
Understanding the user was crucial
User research was of course a crucial part of fine-tuning and validating our concept. We started by mapping out shopping habits through a survey and a couple user walk-throughs of the current app. As the project progressed, we started doing some more in-depth user testings, where we wanted to understand how the users shop in the current app, and how they reacted to our concepts. We used the feedback of each test to adjust our sketches to create a seamless shopping experience. We also supplied our findings with reports on shopping behaviours from Forbrukerrådet and Amedia.
...Which resulted in three main findings
Our solution
We looked at how we can use ML based predictions from start to finish to create the best possible shopping experience for the end user. The project resulted in a Figma prototype showcasing our proposed features. The prototype follows the entire customer journey, from finding inspiration on what groceries to shop, to completing a online order.
Themed shopping lists
Customers often have rough ideas of occasions they are going to shop for, such as specific dinners. For example, if they want to have hamburgers for dinner, they often create a shopping list in their mind consisting of simple components: a protein, a burger bun, and some vegetables. When browsing a physical grocery store, the customers can browse and discover alternatives for each of these components, and choose those they find appealing. When browsing an online grocery store, customers have to search for specific items, and miss out on the inspirational part of browsing a grocery store. By introducing themed shopping lists, the customer regains a part of the feeling of being in a physical grocery store.
With our proposed design, the customer only has to search for a specific occasion, such as "hamburger", and they will be presented with themed shipping lists that fit the search. Customers can easily add groceries from this list, or choose to swap for an alternative if they want to. The themes shopping lists are curated by ML to fit the user and create lists that are relevant. The predictions are based on:
- What the user is searching for
- Groceries already in the shopping cart
- The shopping habits of the user
- The timestamp
- What the user is searching for
- Groceries already in the shopping cart
- The shopping habits of the user
- The timestamp
Single predictions
In addition, we designed a simple addition for ML predictions just before check out to ensure that customers don't miss out on items they might have missed. The customer receives five addition suggestions based on what is already is in their cart. The predictions are based on:
- Groceries already in the shopping cart
- The shipping habits of the user
- The timestamp
- Groceries already in the shopping cart
- The shipping habits of the user
- The timestamp
Retrospective
We looked carefully at the entire user experience and decided to actively redesigned the entire app and looked at how we can use predictions from start to finish to create the best possible shopping experience for the end user. It was exciting to present our concept to MENY at their office and to receive such good feedback.
