How data is helping the fashion industry predict purchases

The fashion industry produces an incredible amount of waste: 13 million tons globally each year, and that is only going to increase by an estimated 60% between 2015 and 2030, with an additional new 57 million tons of waste being generated annually, reaching an annual total of 148 million tons by the end of this decade. 

This is due, in part, to manufacturers overproducing; every season 30% of clothes produced are never sold, leading to 12.8 million tons of clothes being sent to landfills annually. Ultimately, more than $500 billion is lost every year due to underutilized clothing.

This is an unsustainable situation, and a crisis for both the industry and the planet. It's also one that will only be solved, like so many other problems, through the use of data, analytics, and technology, which can help manufacturers and retailers actually predict purchases, helping them understand which styles and trends will sell best, along with the longer term trends that will shape the industry in the years to come. That is a big change from having to wait for feedback from retailers on what already sold, at which point it's too late to cut down on waste.

So, why is this relatively new for fashion when so many other industries have been using these technologies for years? Simply put, the fashion industry has lagged behind other industries simply because, for a long time, this type of data wasn't available because it just simply did not exist.

"Apparel players point to several core challenges as constraints for investing in their analytics capabilities, including poor data quality, a rapidly changing assortment and competitive landscape, high SKU and logistics complexity, and limited analytics expertise within the current employee base," McKinsey noted in a report back in 2018.

Even at the time, though, things were already starting to change; as that McKinsey report also noted, "a number of successful apparel wholesalers and retailers have begun to crack the code on analytics," and that some analytics applications were already increasing the sales figures of different companies by up to 10%.

Fashion forecasting

Cut to 2023, and a number of different brands and retailers are now regularly using this type of data and analytics to inform fashion forecasting, which involves using research to predict future buying habits. 

This can include both long-term and short-term forecasting. Long term forecasting, as the name suggests, looks at the larger industry macro trends, sometimes years in advance. Those can be technological, scientific, economic, political, or cultural changes, such as the move towards more sustainable fashion.

Short term forecasting, meanwhile, involves using data to predict more immediate changes, such as what styles and colors customers will want in the coming season. This data is used by product developers, merchandisers and production managers to shape their upcoming collections, and to gain insights into what will sell. 

So, what data are they actually using to make these predictions and to forecast where the industry is going? It can be everything from weather forecasting, real-time sales data, inventory levels, purchase history, and product movement.

"Through predictive retail analytics, the retailers and company heads can use historic data to generate futuristic insights. They can predict potential sales in the next year, quarter, or the very next day, forecast trends, know expected industry activity, predict customer behavior, and much more," wrote business intelligence reporting tool and analytics software provider Intellicus. 

"Predictive analytics provides a competitive advantage by proactively informing the leadership about the potential events and outcomes and making a timely action plan before it occurs. Retailers can increase sales, understand which product will sell better, optimize the complete supply chain, and be more efficient at the ground-level every day."

Using artificial intelligence 

As with all data, it needs to be turned into something actionable to be useful. And that requires some amount of artificial intelligence to sift through it and gleam out what's useful.

There are a number of companies that are now providing these types of tools to retailers and manufacturers, analyzing this data through the use of AI, such as Stylumia, MakerSights, StyleSage, and Heuritech, a company that uses AI to predict fashion trends by analyzing social media images

For example, Shinola, a luxury goods brand, worked MakerSights, which correlates consumer feedback with historical sales data, for its Vinton watch; while it was initially designed with women in mind, through these analytics it learning that the watch appealed to all genders. As a result, the brand deepened its buy-in by roughly 70%.

“You never design by data, but the data provides a compass as you’re navigating a hunch,” Shinola CEO Tom Lewand, told Vogue Business. 

Stylumia, a trend forecasting solution company, meanwhile, uses consumer-driven trend research, meaning looking at seasonal category reports, to make fashion trends and seasonal category report. It also uses in-season and post-season performance of creations and instant actions to make fashion performance insights and predictsions; the company makes pre-season predictions of demand and grading through its analysis. 

"Using data on an internet scale, we cover brands, retailers, and runways relevant to your fashion brand and prepare an appropriate fashion forecasting guide for you. The AI technology uses the latest fashion forecasting methods to map new product introductions, pricing and discount movement, market positioning, promotional offers, and get up-to-date information on best sellers and laggards across your inspiration and competition," the company wrote in a blog post. 

"Using the Consumer Intelligence tool, one can instantly determine consumer buying behavior across markets, geographical regions, fashion retailers/brands, categories, and styles/colors – enabled by our Demand Science engine."

Stylumia's customers have increased profitability by between 30% to 50% in under 12 months with 10x return on investment.

IBM is also using Watson to help retailers, including a project with Bestseller India, using a virtual assistant that can answer employee questions.

Of course, there are still potential problems with using AI in retail; as IBM points out, though, there are still limitations in certain markets when it comes to AI, such as in India, where there is a "lack of demographic data for AI to leverage," and it only works if employees know how to use it, the company wrote.

"AI is an essential tool, but the even best technology won’t help if you don’t prepare employees to accept it. As they say, the proof of the pudding is in the eating. People may resist AI unless you develop use cases that show positive results."

(Image source: dotactiv.com)