Background

Aim

The system designed was meant to enhance the shopping experience for Net-a-porter clients by using virtual or augmented reality (VR/AR). Our project was an open-ended ‘greenfield’ project in which we explored what we can achieve with virtual reality in the context of shopping and retail.

Initial Problem Statement

“Virtual Reality / Augmented Reality shopping experience. This is more of a greenfield project on how e-commerce can embrace VR / AR. investigate how this could change the shopping experience."

Our Client

This project is being developed for the Yoox Net-a-porter (YNAP) Group, the world’s leading online luxury fashion retailer. The YNAP Group is a result of the merger between the Yoox Group and the Net-a-porter Group, both previously leading online retailers. YNAP now runs several online websites for luxury fashion retail targeted at different demographics.

More information on our client can be found through their website (http://www.ynap.com/).


Requirement Gathering

Interviews

Our team did not conduct any interviews to potential end users because most of the requirements could be gathered from interviewing Irina and Robin (our clients). They explained what they wanted us to create in more detail and our discussion led to a list of draft requirements which was then revised. The main idea was a virtual recommendations engine that pulls clothing images from the YNAP websites and combines them into given outfits. People would then have the possibility to "like" outfits and their preferences would be updated.

Surveys

Even though our requirements list was drafted when interviewing our client, the prioritisation of requirements was still an unknown and we had a couple of decisions to make. To resolve these unknowns we carried out a survey with two questions with a sample size of 27.

The first question concerned the type of input the user preferred for moving between items of clothing in the app. The result suggested a strong preference for hand gestures over voice commands so hand gesture recognition became a higher priority requirements and voice recognition remained a "could have".

The second unknown was regarding the behaviour the app should have when a user liked an outfit. As this is a recommendation engine, we thought it should update the outfits to display next as soon as the "like" button was pressed. However, we also feared this might be too invasive. We did not reach a consensus on this with our client so the survey results were the deciding factor. We opted for updating the list of products displayed to the user every time they "like" an outfit.


Use Cases

Personas

The typical end user for this recommendation engine would be loyal Yoox Net-a-Porter clients, or EIPs (Extremely Important Person) as YNAP staff refers to them. This is because the HoloLens application would be offered to them as a novelty when they are invited to visit YNAP offices or in networking events. As such, the persona in queston values high User Experience, responsiveness, simplicity and elegancy in applications which is what we aim to provide with our app.

Use Case Diagrams

Based on the personas above, our system should fulfill the following use cases:

  1. Browse outfits by changing back and forth between top and bottom items.
  2. Add any given outfit to your list of favourites.
  3. Navigate to the Favourites section from the Home menu.
  4. Navigate to the Home section from their Favourites.
The flow of operations of the above cases are depicted in the following diagram:

Scenarios

Bianca is a well known fashion blogger and her website recently reached 2 million followers. With this occasion she decided to do something unique for her followers and the luxury brand Net-a-Porter offered her the possibility to try their new augmented reality app. She thought that this is an amazing opportunity and it would be a good story for her blog. Bianca is using the HoloLens in order to create some outfits using clothes from Net-a-Porter’s website. She is using voice command when she wants to change the top or the bottoms. After a few minutes Bianca already created a few outfits and saved them all. Now, she can go back, decide which is the best outfit and present it to her followers. Bianca successfully completed her goal.

Madison is a popular high fashion model that is going to represent an important clothing company which has a contract with Net-a-Porter. They offered her the possibility to choose her outfit for the next runway which has as a theme street style. Madison decided to create her outfit using Net-a-Porter’s app as it would be easier for her to make a decision by seeing different outfits in 3D. She is using hand gestures to change the tops and bottoms until she finds the perfect jeans and a matching shirt. She saved the outfit, presented it to her managers and they approved it. Madison successfully completed her goal as she is going to wear this outfit on the runway.


MoSCoW Requirement List

Based on our multiple client meetings, interviews, and the surveys carried out, the requirements for our system were refined to the ones below, and prioritised using the MoSCoW method. These were approved by our client.

Must have:

*Functional requirements are marked F, and non-functional requirements are marked NF.

  • The ability to pull any clothing image from the YNAP clothing APIs based on Product ID. [F]
  • Hand gesture recognition to view next clothing item - the item which changes depends on where the user gaze is. [F]
  • Mapping of the 2D clothing images into 3D space. [F]

Should have:

  • Live preference updating in app (based on outfits the user can “like”) - liked outfits stored in memory. [F]
  • The ability to view all liked outfits in a list after adding them to your preferences. [F]
  • Intuitive structure and flow of the application to facilitate user navigation. [NF]

Could have:

  • Integration of voice commands as an additional navigation and input method. [F]

Would like to have:

  • Integrating haptics using the GloveOne to allow to users to feel the texture of the products. [F]