AI in Fashion is much more than a fashion phenomenon!

Whether it is to create more economic value, to produce less but better or to improve working conditions, AI is now a must in the fashion world. It’s an active and effective ally to designers and operational teams alike.

AI can also improve customer satisfaction by enhancing the Fashion Shopping Experience, which is essential to sustainably engage consumers with brands.

But above all, it will make it possible, in the short term, to reduce the environmental impact of production processes and to reduce the destruction of value due to overstocking and aggressive promotions.

IA AT THE SERVICE OF THE CUSTOMER

In a highly competitive market, it is very difficult to recruit and retain consumers without providing a high quality customer experience. To achieve this, a brand needs to offer a seamless and as personalised as possible shopping journey to each of its customers, regardless of the interaction channel. To properly address this combinatorial explosion, many fashion players are now exploiting the formidable capabilities of AI to offer innovative solutions to the customer. For example, the French start-up Veertus allows you to virtually try on a garment on your smartphone using the barcode of an item of clothing, thus doing away with the need to try it on in a shop. Based on machine learning techniques, the in-store shopping experience is simplified and digitised, two valuable assets during the Covid period.

AI is also increasingly used to make customer service more fluid, thanks to chatbots capable of quickly answering many questions such as: tracking an order, the availability of an item in a shop near you…

For marketing, AI has become essential to significantly improve customer knowledge (Buyer Persona) as well as marketing campaigns, by proposing only items that correspond to the customer’s tastes, budget and morphology.

Ultimately, AI not only contributes to consumer satisfaction but also helps retailers mitigate the costs of returned goods (web and shop) which are known to cost billions of dollars.

USES AND BENEFITS FOR THE PRODUCT

As fashion is constantly changing, detecting trends and accurately predicting sales volumes is another strategic challenge for the textile industry, in order to manufacture only the necessary quantities, at the best price and at the right time.

In addition to trend books, AI is now helping designers to more systematically capture the trends revealed via social networks. For example, Heuritech, another French startup, analyses millions of Instagram posts every day allowing them to pick up early signals from fashion influencers and consumers. Their visual recognition technology provides predictive analysis of the trends and products that will be in fashion in the coming months. A real asset to ensure that the next collection will be successful.

AI is also at work to enrich item information, with rich and precise descriptive attributes, automatically generated from a simple photo. This is called Deep Tagging (https://www.pixyle.ai/).

Increasingly used by online shops, AI-based visual search can understand the content and context of an image to provide useful information and recommendations. With a smartphone, it can easily take a picture in the street and search for the desired product online. If an item is out of stock, the search technology also allows brands to suggest similar items or related to a theme or style to shoppers without going through a difficult to formalise textual query.

Finally, AI can significantly improve sales predictions for each item in a new collection, 6 months in advance, by analysing past sales of products identified as similar, after correcting for “promotion” effects and stock-outs. This is how solutions such as Demand Forecast, which also incorporates machine learning models to correct seasonality, balance sizes, expand or contract distribution areas or the number of sales weeks. Here AI makes it possible to improve the accuracy of the initial quantities purchased by up to 30%, with the corollary benefits of reducing the discount and increasing the rate of sale. The ultimate objective, to manufacture only what the brand is able to sell, is then achieved!

IA TO OPTIMISE STOCK MANAGEMENT & STORE PROCUREMENT

The supply chain and especially the optimisation of warehouse and shop stocks are being profoundly transformed by the systematic use of AI.

For the supply chain, AI greatly improves the ability to predict delivery times and therefore to anticipate delays or anomalies in the transport of goods around the world. The company Wakeo offers, thanks to Machine Learning, to analyse the performance of a distribution network by geographical area, carrier, forwarder or operator. The retailer can then finely optimise its supply chain by identifying interdependencies, recurring deviations from forecasts, hidden risks, etc. These optimisations are now essential in order to keep up with the race to deliver within 48 hours imposed by Amazon and its bonus programme.

AI is also increasingly seen as the best solution for optimising store stocks. The challenge is to take into account the specificities of each point of sale (surface, location, competition, customer profiles, commercial operations, weather, etc.) to adjust daily the ideal stock quantities for each size and colour reference (SKU). This increase in the availability of products in the right place accelerates the clearance rate and thus reduces the average level of discounting at the end of the season.

These same algorithms can also be used to stop restocking a shop after a particular SKU has been sold, as the store has a statistically low chance of selling it again. The products remain in the warehouse and are available for other shops and especially for the ecommerce website.

AI is also at work to reduce web order returns, which are very costly for the merchant and for the planet. This growing phenomenon (in Germany 50% of online sales are returned) reflects the behaviour of consumers who have several sizes and colours delivered for an in-home fitting session before returning anything that doesn’t fit. Thanks to supervised self-learning that crosses web paths, purchases, customer profiles and product repository, it is possible to detect and isolate these consumers and thus simulate stock-outs on the ecommerce website for products and sizes that the AI identifies as having a high risk of return. The complexity of such a model lies in its ability to differentiate between a basket with risk of return and an order for all family members.

IA FOR FASHION ZERO WASTE

Fashion is the second most polluting industry. It generates 20% of the world’s wastewater and 10% of its carbon emissions. This can’t continue!

Moreover, it is a declining market which has lost 15% of its value in the last 10 years. There are two reasons for this: overproduction generating unsold goods and value-destroying promotions, but also a change in the behaviour of customers who aspire to more personalised, responsible and measured consumption.

The fashion industry must therefore accelerate its transformation towards a new Zero Waste Fashion Business by addressing two major challenges:

• To produce only what a brand is able to sell, by accurately estimating the quantities to be produced for each new collection
• To adjust the stocks of each item in each sales channel and in particular in each shop, taking into account the local specificities of the store and the expectations of the consumers.

This complex equation, which generates a real combinatorial explosion, can now be solved thanks to the AI embedded in more and more business solutions.

AI therefore offers the fashion industry the opportunity to thrive again in a consumer-driven era while becomingenvironmentally responsible and thus green and profitable!