Making decisions based on gut feeling and obscure market predictions are in no other industry as apparent as in the fashion industry. Fashion brands produce apparel that simply does not match customer demand. The price tag of this error? $210 bn per year. That is the loss in the global fashion industry resulting from the mismatch of supply and demand for clothes and accessories.
As much as half of all products in the fashion market are thrown away at factory sales, relabeled and shipped to other markets or even burned. This huge waste is not only an immense environmental burden but is also very costly for the industry players.
The difficulty lies in the broken process of designing and procuring fashion. Fashion designers and procurement agents rely on their feelings, experience and reports from market research organizations to forecast future trends. These agencies base their predictions on a wild array of sources. Field research, data from web shops, social media profiles or even indicators of the state of the economy and information on the current political sentiment are analyzed with the hope of finding out what consumers will buy.
Limitations of the currently prevalent approaches in fashion market research
This approach, however, has strong limitations. First, the agencies’ observations remain to a certain extent unstandardized and subjective. Second, there is a considerable time lag of up to two months between the time the data had been collected and the insights generated. The latter aspect is particularly problematic when considering the highly dynamic nature of the fashion industry.
Moreover, using economic data to help with the prediction can actually lead to some wrong conclusions. The arguable relationship between the state of the economy and the length of hemlines (i.e. the Hemline Index), or the relationship between the economic climate and the demand for lipsticks and the height of heels might be deceiving. Contradicting the theory of heel heights increasing in economic downturns, in 2011 IBM predicted that the negative relationship between heels and the economy will invert — at least for the time following the analysis. This surprising finding however shows the problem that fashion-trend forecast agencies face when basing their predictions on such simplistic generalizations.
However, with the emergence of big data and advanced analytics, leveraging existing and new data has the power to completely transform how fashion brands decide which clothes to create or procure.
The transforming power of data in the fashion industry
The antidote to the traditional method is a more data-driven approach. A first step in this direction is the collection of data from an array of obvious sources while trying to achieve objective evaluation and prediction of market trends. Customer behavioral data such as terms entered into search engines, user activity on social media and consumer purchasing and consumption behavior is already being used for new insights into emerging fashion trends. Also, sociodemographic and psychographic data or data reflecting the macroeconomic and political situation can be included.
Several companies in the fashion industry are starting to leverage the power of data: The Italian fashion brand Miu Miu has identified new opportunities for their collections by analyzing the behavior of over 300.000 influencers on social media. Also, some fashion trend forecasters incorporate data gathered online (such as customers’ clicks on certain websites) in their evaluations of changing fashion trends. Using advanced analytics techniques, ASOS was able to increase its revenue by as much as 33% and McKinsey reports that some analytics applications regularly increase the sales figures of different companies by up to ten percent. The potential is much higher, of course.
The status quo
The fashion industry as a whole, however, has been particularly slow in embracing the potential of big data, predictive analytics and the insights that these tools can provide. Here’s why: according to McKinsey, one of the major reasons for this is a lack of high quality data. The opportunities for a deeper understanding of market trends and the potential for making sounder business decisions therefore remain largely untapped.
The following chart gives an overview of data types in fashion market research:
In the left box we can see the data available today. What sticks out here is troubling: the fashion industry has close to zero data at hand on what their consumers actually do with their products. The right box shows what market researchers consider the best possible type of data, the so called “behavioral objective data”, e.g. what Google Analytics tells you about user behavior on your website. In fashion, however, the product is the garment (or shoe, etc), not the website. The problem with behavioral objective data is that it’s the most difficult to get, particularly for physical products. The IoT development plays a very important role here or, in other words, the digitization of physical products.
Outlook — changing the game
Once the sources of high quality data in fashion are explored, big data and predictive analytics will become a game changer in the fashion industry. The benefits of utilizing objective usage data based analytics in the fashion industry will not be limited to predicting shifting consumer trends, however (predictive analytics). In the same way, insights into what’s going on right now (descriptive analytics) are likely to change fashion forever. Seizing descriptive analytics to their full potential will encompass determining the best marketing strategies, making much more accurate product recommendations to customers or making pre-market launch product tests easier.
In the near future, when the behavioral objective data has truly found its way into the apparel industry, market research companies, fashion trend forecasting agencies as well as fashion brands themselves could greatly enhance their understanding of how fashion is being used, how trends unfold and how the tastes of their specific customer base change in real time. The insights would not only span broad market trends but could also provide actionable insights even into the most niche segments or remote geographic locations.
Amazon is trying to source high quality objective usage data with Echo Look. It’s a selfie camera connected to Alexa, capable of taking pictures of you wearing outfits. It then compares two outfit selfies to recommend the one which Echo Look deems as more stylish. But so far Echo Look is poorly adopted by consumers and the data set it creates doesn’t have many variables.
Combyne is both the largest social dressing and fashion community on iOS and android as well as a market research company. Over 500,000 monthly active users create over 1.5 million outfits per month from millions of items, share them with their friends and provide styling advice.