Jean-Marc Chauve is currently a consultant, director of the Imane Ayissi brand and artistic director of IFA Paris. He studied marketing, fashion fesign and socio-semiology of fashion, and used to work at Nelly Rodi and Maison Martin Margiela.
The development of e-commerce and the importance of the internet in terms of fashion purchasing decisions allows the collection and analysis of huge volumes of data. Can this "big data", this enhanced awareness of consumers thanks to new technologies, help to anticipate trends and finally make the volatile market of fashion predictable?
Big data is trending in the fashion industry. Articles, debates and round tables are proliferating on the subject and experts predict that trend offices (and sometimes even stylists) will soon be replaced by algorithms.
The fashion industry is undergoing profound changes that are not yet complete, in particular in terms of acceleration and digitization. For mass market brands such as H&M, the horizon is no longer the half-season, a product must be created and produced within a few weeks, while the traditional trend offices usually reason in terms of the season. In addition, their production volumes and their low margins don't leave much room for risk-taking, each garment produced must be a best seller.
But the share of e-commerce in fashion is increasing steadily, representing almost 17% of the textile clothing in France in 2016 according to IFM surveys, more than 20% in the United States, even if it covers important disparities according to the types of products and the brands' locations. And if, according to Nathalie Remy, associate director of McKinsey who was speaking at a conference organized by Farfetch on the future of shops, 75% of the purchases will still be done in stores in 2025, 99% of them will be influenced by the digital. The temptation to rely on the mass of data collected by the digital is therefore big, whose analysis by algorithms would allow a perfect knowledge of the consumer, to anticipate future best sellers.
Thus, Google now has a "trendspotting" division, and regularly publishes its’ "fashion trends reports" based on the search trends of keywords of internet users. Studies of limited scope when it comes to clothing, since words cannot account for visual or intangible data, aesthetics and style, that are fundamental in fashion. WGSN, the leading online trend agency, has created INstock, a service for analyzing past sales to predict future sales. Even more ambitious, Lauchmetrics, a launching agency for new products that collaborates with brands such as Dior, Levi's, L'Oréal and Alexander Wang, proclaims on its website: "Data becomes the most loyal allies for brands. "Imagining" trends and what your consumers need is no longer necessary. From now on, the data is enough to predict and anticipate their expectations "
With big data, did the prediction of trends become an exact science?
As recognized by Edouard Fonkenell, founder of the agency Claravista, a "multichannel" marketing agency, during a round table organized by Lectra and ESCP Europe during the last Fashion Tech Week that was held in Paris at the beginning of October: "It's never perfect, an algorithm makes mistakes all the time". Every online consumer can see it for himself : between offering the exact same product that we just bought or a completely unsuitable one, if your journey on e-commerce sites is not completely linear (women's clothing when you are a male consumer for example), the effectiveness of push marketing of e-commerce has yet to be proven.
But above all, this vision reflects a lack of knowledge of the deep springs of fashion that does not respond to any "need" and no "expectation". We cannot talk about demand in the classic sense of marketing: in our developed societies, the closets of all the consumers are oversaturated ... Fashion is based on desire and trends, more precisely on the weariness of the same, the déjà-vu and the desire of novelty, of the unknown. As Nicolas Santi-Weil says, Managing Director of the AMI brand, "Creating something that nobody needs today but everyone wants tomorrow"
Data analysis is therefore certainly very useful to detect when a rising trend becomes a mass trend, or to anticipate the decline of a trend. So it is a tool that's perfectly adapted to the mass market sector which is never at the origin of the appearance of a trend and works in the short term. In any case, knowing your customers better is a necessity for any brand, but it is far from enough. We do not see how the analysis of past data and in large quantities can allow to detect new trends in advance, which originally concerns only a small part and usually atypical market. Does the future hold the appearance of artificial intelligence endowed with intuition and capable of analyzing a multitude of very complex weak signals? Perhaps, but in the meantime, wanting to secure the fashion market based solely on big data may instead accelerate the decline in fashion consumption by offering only "predictable" products and therefore not very desirable ones...