Service design and AI in Sustainable Fashion

We designed an internal work tool using AI technologies for increased business resilience and better collaboration during rapid growth in the Swedish clothing rental company Hack Your Closet (HYC).
a picture of a girl scanning a t-shirt

Story

The project took place during the spring of 2021, a time when the Swedish clothing rental company Hack Your Closet was growing rapidly and got listed on the New York stock exchange. Me and my project partner were brought in at a promising time for the company when they wanted to explore ways that AI could be used to enhance the styling aspect of their business. 

We approached the project with a service design perspective and learned that a critical challenge for HYC was that their internal processes and workflows weren’t designed to cope with their rapid business growth. Our focus became to enhance the styling aspect, but also attempt to re-design workflows and internal tools that could help them be resilient through their business growth.

A view of the stock management part of our designed internal tool

Background

Hack Your Closet was a start-up company founded in June 2019. Their mission was to lower mass consumption of clothing by offering a service where customers got to rent clothing instead of buying it. The company had a large closet consisting of ~30.000 pieces of clothing that were bought through second-hand or overstocks.

To differentiate themselves from other clothing rental companies, HYC had stylists that chose clothes for every customer every month based on a style profile where the customer had shared their sizes, preferences and feedback. The customers got the clothes sent to them, kept them for a month and then sent them back to HYC who washed them and returned them into the circular closet.

Responsibilities & Process

Project management and collaboration

This was a project that I managed and ran together with a single project partner. I thereby had the joint responsibility over all of the parts of the project as well as for the directions we chose to take in the project.

The fact that we were only two required that I communicated clearly, could form good arguments and in general - collaborated well.

Learning everything and doing it remotely

As a first step, I set out to learn everything about the service and because the project ran during the time of covid-19, I had to do everything remotely.

I did the following things:

Customer experience

I read reviews and customer correspondence with HYC, had informal conversations with current and prior customers and became a subscribing customer myself.

CEO interview

I conducted a deep interview with the CEO, focusing on collaboratively mapping out all parts of the service, i.e. departments, employees, tools, processes and workflows. Through this, I was able to identify more critical challenges than the styling.

Observations

I visited the company locales digitally where they showed me how departments were connected physically and how different processes were conducted.

Competitive analysis

I did an analysis of competitors to see what was HYC competitive advantage and what was perhaps working better in other places.

Employee interviews

Based on the identification of the most critical problem departments, I conducted interviews with the heads and employees of those departments, i.e. styling, buying and processing.

Identifying and formulating the critical challenge

By getting a holistic understanding of the service I could identify that although the stylists and buyers respectively had some challenges that could be addressed, the root of the challenges arose from the departments’ poor collaboration and lack of foresightful planning.

The mechanisms of supply and demand were out of balance, a common challenge because of the immediate characteristics of services, but in this case leading to both greater financial spendings and poorer service delivery - creating a highly critical problem to address in the short-term.

Creative mapping techniques

To concretise the knowledge and to form arguments as to why the project grew to where it did, I used different mapping techniques. I mapped the workflows of stylists, buyers and processors, highlighting their individual challenges and pain points in their interactions. I mapped out the full ecosystem of the service and the customer experience.

I also did a detailed mapping of a single piece of clothing’s journey through the service as a way to understand how all processes and workflows intertwined.

A single garments journey through Hack Your Closet's service.
A single garments journey through Hack Your Closet's service.

Exploring different ideas

I noted down two ways to support stylists; 1. by helping them be creative with what they already had, and 2. by creating foresight to when good clothing is getting low and let other departments know.

I sketched and explored different ways to solve the problems. For example, automatic warning messages to other departments when a stylist can’t finish a customer’s box, perhaps depending on where estimated stock is located (in processing, laundry etc.) or if the buying department needs to act.

Conducting co-design workshops and evaluations

As me and my project partner continued to develop ideas we conducted work-shops where the employees of HYC got to be part of the designing. They were involved in deciding what information they needed and could give in different phases, what aspects could be involved in their processes and what else they would need if something were to change. Their collaboration was imperative to our design.

The visual material used for one of the Workshops I held.
The visual material used for one of the Workshops I held.

Delivered Design

Digitalising the stylist workflow

All garments at HYC had a barcode and were processed into a digital database to keep track of their placement, but this was done manually by the processing team and the stylists never took part or could use any of the digital information.

In our service design we suggest that the stylists use scanners to streamline both their own work and can remove the processors’ task. The stylists can then immediately be noted if a customer has had that garment before or has preferences that don’t coincide with it, etc.

Digitalising and organising the stock

At the time HYC had ~30.000 unique garments in their closet - some put away due to season, but most rotating over four weeks in customers’ closets. In the closet garments were only sorted based on size and type of clothing, and if the stylists were looking for something special, there was no way of knowing where a garment was.

As the stock grew, it needed to be detailedly organised - each garment placed in a numbered slot in the closet and the possibility to digitally search for garments’ locations.

Internal digital tool

The two aforementioned digitalisation and organisation strategies were pre-requisites for the digital internal tool we designed. To enhance collaboration, we argued that one single tool specifically designed for their purposes would be beneficial.

We focused on developing four different functions which could help HYC cope with their critical challenges.

I also created a video to showcase how the tool could be incorporated into workflows in Hack Your Closet.

A video created in After Effects.

We designed an enhanced customer profile with all customer history and preferences in one simple overview. With growing numbers of customers the stylists can no longer remember every individual customer’s history and preferences.

Their work can be more personalised and effective with a better style profile.

The new customer style profiles for easing the stylists' work.
The new customer style profiles for easing the stylists' work.

In the same view we have designed a styling tool that uses AI. Because of the better organising of the stock, a system can be developed to overview the entire stock, sort out what is unavailable and present suggestions of what garments to pick for a customer based on their preferences and history.

Because the personal styling by a stylist is a part of HYC unique selling point, we did not want to hand over the process fully to an AI, but rather design an assistive tool the stylists could use as a support.

The simple assistive AI tool that suggests clothing that match the profile and informs stylists of where to find them.
The simple assistive AI tool that suggests clothing that match the profile and informs stylists of where to find them.

To create a better collaboration between stylists and buyers, we designed a communication tool that works like a combination between a chat and a bulletin board - what we called a Challetin board. Stylists could send messages about what they needed and the buyers could save them on a note to a place in the system where it would be relevant to them.

The communication tool that works as a combination between a chat and a bulletin board.
The communication tool that works as a combination between a chat and a bulletin board.

The largest focus for us was to design a stock system that could foresee a future stock to support buyers in knowing what they would need to buy in advance.

The most important part of such a stock system was the adjustability of views to be able to look for different potentially problematic low stocks.

We designed an idea for a weekly stock view that would use statistics and AI to estimate what stock would be in the closet every week. This includes new garments and garments returning from customers with an estimated reduction for lost, destroyed or late garments. This is sorted based on sizes and is compared to the sizes of the customers that would need to get garments from that week.

This could provide the buyers with information about what clothes they need to buy weeks in advance. The buyers can go into more details about type of clothing and certain tags that are in bigger estimated demand.

Weekly stock view with indications of what is needed weeks in advanced.
Weekly stock view with indications of what is needed weeks in advanced.
Detailed view of weekly stock with recommendations.
Detailed view of weekly stock with recommendations.

In addition, we designed a general overview of the entire current stock that could be sorted on either type of clothing or sizes and could be filtered down to what is in the closet currently and what is out with customers.

All over colors are used to indicate what garments are in highest demand - low stock numbers does not mean more is needed, high demand does.

The general stock overview with colours warning about low stocks.
The general stock overview with colours warning about low stocks.
The detailed view with filtering functions to overview the entire stock.
The detailed view with filtering functions to overview the entire stock.

Implementation and impact

Unfortunately, our project involvement ended with our delivery of a prototype and suggestions for improved workflows and processes. I don’t believe much of it was implemented due to high costs and organisational unreadiness.

Less than a year later they filed for bankruptcy.

Reflections

Holistic approach

Since we worked on the project I have gained more experience on working with holistic design. Service designers uses a holistic approach and look at the bigger picture as well as to the details, but even if a service is neatly designed over its entire spectrum, the business structure that aims to deliver the service also needs to be aligned and work.

I realise in hindsight that what we set out to do was more about redesigning the organisation and such an approach call for other deliverables. Today I believe that Hack Your Closet was in a position where big investments into new tech were needed for them to handle their growing service. But I also believe that before that, they needed to redesign their organisational structure to be better prepared for the new tech. If I were to go into the project today, that is what I would focus on.