Formerly part of SNAP INC, Fit Analytics (Ares) is a retail tech scale-up company that utilizes machine-learning and AI to help retailers reduce purchase return costs by guiding online shoppers finding correct garment sizes via AI-driven recommendations. Clients include: Patagonia, Oak+Fort, Hugo Boss, ASOS, & The North Face among others. As of April 2023, Fit Analytics diverged from Snap Inc and resumed operations in early 2024 as an independent company.
Mark & Spencer (M&S), a client utilized Fit Analytics’ Fit Finder utility to drive conversions, enhance buyer satisfaction, and reduce returns for purchased upper, and lower-body garments (e.g., shirts, skirts, pants), thus helping retailers saving costs.
Eventually, M&S sought our expertise to develop a machine-learning solution for a rather specific garment. The problem is (was) that customers seldom purchase this garment type online due to difficulties in finding the right size and fit, coupled with restrictive return policies.
Thus, M&S desired to find means to drive online shopping while reducing dependencies on brick and mortar stores.
I worked with:
Here are a few samples of the newly designed garment size recommendation utility made for Marks & Spencer (UK) for both desktop and mobile. We went for a minimal design to elevate elegance, and content ensuring shoppers stay focused on the questions enclosed within the UI.
Together with stakeholders, we set strategies and actioned requirements (business, client). We devised a roadmap with problem statements, tasks, and priorities which we tackled. Simultaneously, I set directions for research and concept design.
The process involved identifying user needs via observations, algorithm training, data annotation, iterative design, and usability evaluation (quantitative, qualitative) leading to final design and copy iterations before deployment on the M&S e-Commerce platform.
Despite our commitment to a user-centered design (UCD) approach within a DevOps Agile environment, our workflow increasingly depended on the requirements set forth by data scientists and machine-learning (ML) engineers. Embracing a pragmatic approach, we adopted a ‘data-scientist-centric’ methodology. This strategy allowed us to effectively balance the needs of data scientists with user requirements, streamlining LLM training and ensuring that shoppers receive accurate garment size / fit recommendations.
As a team, we decided it was logical to base the design and structure of the new product (bra finder) on Fit Finder (left), the flagship solution for upper- and lower-body garment sizing. This decision was driven by several key factors:
Lastly, I ran an evaluative study by recruiting 16 participants. Of this sample, 8 comprised those who took part in the discovery phase (recruited to ensure their requirements are met). In summary, the participants favored the solution enough to be described as experts for being considerate towards gender-related requirements, thanks to the subject-matter expert.
The product was later launched on the M&S website yielding the following outcomes overtime: