In this second article in our FashionTech series, we take a step back and consider the entire value chain through which garments enter the market, and where AI has enormous untapped potential to clean up fashion, the world’s second greatest emitter of CO2.

“Anecdote” of a typical shopping process

The typical online shopping experience today is marred by the problem of size sampling: consumers order multiple sizes to try on and return the ones that do not fit. Why? Simply because there is no real way for consumers to verify which size fits them best.

For consumers, the certainty of size sampling far outweighs any cost of returns. Moreover, many brands now offer free delivery and returns with a minimum basket size, a further incentive that makes the issue far worse. Consumers keep the garments they like and return the ones they don’t. As a result, numerous wasteful and climate-damaging delivery trips are required, and these unwanted garments enter the reverse supply chain. What happens to most of them? Well, fashion companies most of the time decide it’s too costly to handle and restore each garment, resulting in perfectly wearable items clogging up landfills.

So, when a customer buys an item of clothing, they not only contribute to the emissions associated with that single piece but are also complicit in the emissions related to the production, delivery, and destruction of all the numerous other pieces they returned. This is not only extremely wasteful, but also extremely costly for many fashion brands, with companies like ASOS processing up to 30% returns on average.[1]

Sustainability in fashion is hanging by a thread

When such behaviour is applied to millions of consumers worldwide, it is no wonder that 18.6m tonnes of clothing ended up in landfills in 2020 alone.[2] Neither should the fact that $460bn is lost each year because of the disposal of scarcely worn items.[3]

And it is also why the fashion industry is responsible for 1.2bn tonnes of greenhouse gas emissions, which is more than international aviation and shipping combined.[4] If current trends continue, the sector could account for one-quarter of the global average fossil fuel consumption by 2050.[5]

Fortunately, there is already a clear path to a more environmentally friendly future for the industry. On the one hand, we know that consumers are most concerned with getting the desired clothes in the right size. However, they would gladly only buy what they need (rather than size sampling) if there was a way to do so.

Fashion retailers and manufacturers would readily adopt new tools if they helped them increase their bottom line. This creates an alignment of interests, and we are already seeing AI begin to play an important role in aligning these interests and creating a more sustainable fashion industry.

AI, the “silver needle” to mend a broken fashion industry

AI will become one of the biggest enablers of a truly sustainable fashion industry. This viewpoint is based on how AI transcends all the segments of the fashion value chain (downstream, midstream and upstream) and the amount of inefficiencies (and resulting emissions) it can eliminate.

While AI in fashion is not entirely new, we are only now seeing its emergence. AI and technology have received an average of 1.5% of fashion companies’ revenues over the last few years.[6] These investments, however, were never undertaken to drive sustainability but rather as a means for differentiation and revenue-boosting. For context, in the banking industry, average AI and technology investment is 4-5 times higher, at roughly 7%, showcasing that banks are highly aware about AI and its potential.[7]

A shift in priorities from sole profit-making to sustainability has occurred, driven in large part by consumer demands for sustainable products, and fashion brands are increasingly turning to AI for help. Brands are realising that to remain relevant today, they must significantly increase their contributions to sustainability, and AI gives a convenient and effective means of doing so without upending entire supply chains or decommissioning production facilities. Whereas technology and AI adoption were previously viewed exclusively as tools to increase profits, there is now a growing understanding that these also have enormous application and credibility in the sustainability movement.

This shift is predicted to drive investments in AI from around 1.5% today to 3-3.5% of revenues by 2030.[8] What does this mean in $s? AI spending in retail is expected to be a $19 billion[9] market by 2027, up from an estimated $7.3bn today (approx. 3x increase in absolute amount).[10]

To demonstrate the growing importance, perception, and relevance of AI in sustainable fashion, early predictions point to the potential for enterprises already employing AI to generate a 118% increase in cash flow.[11] Those that are just getting started with AI-driven initiatives could see a 13-15%[12] increase in cash flow. In actual $ terms, that adds up to $100-150bn[13] of incremental cash flow over 5 years. This is especially pertinent in mindful luxury, which is estimated to generate $1.5tn in annual sales.[14]

For both brands and investors, the potential benefits of pursuing profitability while prioritising the environment are undeniable. Companies like Amazon, which actively embrace AI are already reaping the benefits. The company’s AI-enabled features account for 35% of all consumer purchases, resulting in a 3x higher conversion rate than the competition and an additional $800m in monthly revenue.[15] In stark contrast, laggards who remain agnostic to AI and remain stuck in the status quo face a 23% relative decline in profits.[16]

Downstream fashion – what’s wrong and how AI can fix it

The fashion value chain can be divided into three major streams: upstream (manufacturing), midstream (distribution), and downstream (consumer purchasing). Given its exposure to consumers and the secondary circular market, the latter is where our attention is concentrated.

Size sampling is a key issue here. However, this problem is not limited to the primary market. Consumers shopping on second-hand garment resale platforms face the same difficulties as their ‘primary’ counterparts. There is an additional issue, particularly for luxury goods, in that purchasers are unsure about the authenticity of such second-hand garments.

AI can enhance consumer personalisation, trust, and confidence by increasing the likelihood that each purchase includes the exact desired garment in the desired size.  AI-based personal styling and sizing platforms can improve sizing accuracy. At the same time, AI-driven augmented reality (AR) and virtual reality (VR) technology can help shoppers to better understand a garment’s proper fit. More advanced AI-driven recommendations based on individual preferences can also reduce the number of returns.

Several companies in the field are already providing cutting-edge solutions. Stitch Fix is an online styling platform that employs AI/ML to personalise clothing based on the dimensions, budget, and style preferences of the customer., a fashion and eyewear style-based customisation platform, uses AI and high-quality rendering services to boost customer lifetime value and reduce returns.

The algorithms of women’s retail app THE YES uses customer data to improve product personalisation. The platform has already partnered with companies like Prada and Ralph Lauren. MySize, a virtual ‘fitting room’ app, can calculate and record measures in unique ways and is accessible via a smartphone camera. There are also reverse imaging tools available such as and Lykdat that enable shoppers to find sustainable versions of desired items by simply uploading pictures. Even established platforms, such as Rent the Runway, have begun leveraging machine learning to identify their customers’ preferences with more precision.

AI is also playing an important role in enabling the secondary market, through  the authentication of second-hand luxury items to prevent counterfeits. This is critical to ensuring goods are genuine, especially for luxury items, and is essential to sustaining the circular fashion ecosystem. As just one example, US based, Entrupy, has successfully used machine learning and computer vision through its platform to authenticate products from several luxury brands, including Louis Vuitton, Chanel and Hermes.

In addition to use cases surrounding authentication in the secondary market, it must also be reiterated that AI’s value-add to the first-hand market (i.e. better personalisation, better styling, better fit) are as applicable here. These provide another backstop preventing unwanted (but still in good condition) garments from ending up in the landfills.

The first piece of this series has set expectations on $10bn of investments being directed towards the circular economy in the next 5 years. But this ecosystem will only be as robust at the investments in AI that follow it.

Going upstream – stay tuned for part three

In the next piece, we will investigate the sustainability-related issues plaguing the fashion value chain’s upstream (manufacturing) and midstream (delivery) and will quantify the potential reduction in emissions AI could achieve.

Learn more about us. 



Say Hello