There has been discussion in fashion ecommerce for over a year about the roleartificial intelligence will play within the entire industry. One of the most widely read articles in 2016 published by Business of Fashion highlighted a scenario of convergence between smart technologies and the fashion world: just as happened with the internet,artificial intelligence will rapidly contaminate every aspect of this industry. From the alignment between supply and demand to automated conversations via chatbots to predictive intelligence to suggest new trends to brands and designers: today,artificial intelligence is no longer a futuristic key but a present driven by already mature technological solutions.The cornerstones of the relationship between fashion retail and artificial intelligence are as follows:
1.Aligning supply and demandStilltoday, fashion brands work with limited data to determine how many products to order and how much to discount them to sell out. If the prediction is wrong, the result is a loss of revenue due to excessive mark-downs and wasted resources in the sell-out. By analyzing a broader spectrum of data, from customer history to the entire market, artificial intelligence could support retailers in deciding with what priority to show and sell products.
2. Competitor study and analysis of market-wide data.The ability to make predictions can be extended, with the support of non-proprietary data. Crowlers exist that can analyze competitor ecommerce sites, with the goal of returning broader spectrum information-for example, identifying which colors or materials are most successful in a certain country or even in a single city, identifying trends and microtrends. This data, which was once unavailable, can take on a key role in defining brand strategies.
3. Personalization of the conversation. Specially for niche brands, which can only be purchased online or in selected boutiques, it is crucial to build one-to-one customer service. The Internet often returns a poor and impersonal experience that can instead be enriched not only by the personalization of the proposition, but also in the pre- and post-purchase brand relationship phase.
4. AI in assistance to designers. There are artificial intelligence systems today that compose music, write, create. As Pedro Domingos, author of The Master Algorithm says, “It will happen that artificial intelligence will not replace designers, but will become an indispensable tool for them. “Late last year, Forbes magazine published a list of 8 trends that will reshape the scenario of the world of luxury and fashion retail.
These include reconfirming some of the points made in the BoF article for example personalization and less friction between online and offline experience. As always, these are data-driven predictions.Fashion Ecommerce and Artificial Intelligence: the case of Stitch FixInfashion ecommerce, the case of Stitch Fix is one of the most interesting because it has translated the natural consultative vocation of the fashion industry-understood as offering services to push products-by relying primarily on the use of artificial intelligence.The company offers a subscription-based clothing sales service that delivers clothes and accessories directly to the home. In reality, users do not turn to Stitch Fix to buy products – partly because the portal does not allow them to do so directly – but to receive outfit suggestions combined with their tastes, daily needs, and current fashion.Customers, following the steps of a short questionnaire based on sizes, tastes, and spending orientation, provide all the information necessary for the system to be able to suggest purchase proposals. Products selected by Stitch Fix are delivered directly to the user’s home, at a frequency set by the user (every two months, for special occasions, etc.) and can be purchased or returned.
Machine learning is at the heart of this business: customizing a proposal every day for thousands of users with different needs and tastes would definitely be time consuming and certainly not scalable. Therefore, machine learning takes care of “digesting” all kinds of information (including unstructured data such as images saved on Pinterest and notes left by customers) and provides some initial results. These profiles are then communicated to a fashion stylist who selects 5 products from many brands and sends them to the customer. The company uses AI to increase the productivity and efficiency of workers by also supporting them in insights and anticipating certain trends.
The Stitch Fix case confronts us with some fundamental questions that can no longer be ignored by fashion commerce. The importanceof personalization even online, understood as adviceIn thecase of Stitch Fix, extreme personalization of the proposal, driven by AI, is the service that the fashion market is now able to offer in retail (thanks to the skills of a good salesperson), but which has yet to be refined in ecommerce. If you want to build a high and consistent experience on the web as well, it is essential to equip yourself with tools that can interpret the need, suggest and advise starting with user recognition.AI-based advice is an asset, but it must be done wellStitchFix uses advice as a value proposition to finalize product sales. Missing the promise by sending the wrong outfits can turn into a point of weakness. The machine learning system, in addition to learning in real time every possible variation, must be constantly monitored to aim for the goal: to provide solutions that are closer and closer to users’ tastes to the point of anticipating them. The winning key for Stitch Fix is the mixed approach between artificial intelligence and human touch. The AI enables the use of all available data and provides an initial skim, the fashion stylist adds the extra touch or corrects any inaccuracies. It is critical that human intelligence enters the business process. In this, machines are not fully autonomous. The competitive advantage generated by AIL Artificial intelligence is a competitive advantage for early adopters who will enjoy all the benefits associated with scalability and the ability to collect multiple data sources, structured and unstructured, returning readable answers and insights that could not be detected even by good data analysts. Fitch Fix’s machine learning enables brands what would not humanly be possible: to fully customize their offerings on a per-market or even per-person basis.
