Heuritech: AI for fashion trends

Heuritech is a cutting-edge fashion technology company that offers predictive analytics on trends and products using real-world images shared on social media. The company was founded in 2013 by two PhDs in machine learning who believed that images on social media could provide valuable insights if one knew how to read them.¹

“Social media is disrupting the fashion industry, shifting the power structure that shapes and influences trends, and therefore sales, away from a few glossy magazines to millions of consumers. By capturing and analyzing these new voices in real-time our AI technology allows brands to predict fashion trends months before they happen, adapt their designs and better plan their merchandise mix.” Tony Pinville, CEO of Heuritech

Today Heuritech’s proprietary technology scans 3 millions of social media images and videos each day and analyzes more than 2000 fashion details (colors, textures, patterns, prints). Results of the analyses include current trends and forecasts about fashion markets and products. This informs the strategies of leading brands, including Louis Vuitton and Dior Moncler, and Havaianas.²

Heuritech offers its services through a subscription-based model and creates value for fashion brands in multiple ways. Through Heuritech’s analytics, brands can:

• Optimize collection strategy and launch timing

• Improve inventory management: reduce overstock and waste by identifying declining trends and improving markdown strategies

• Support international growth by understanding local market specificities, discerning differences in top products, and consumer adoption.

• Foster a data-driven decision-making culture among employees

• Promote smoother communication between creatives and analytical teams

The methodology to forecast fashion trends is structured on four steps.³

Image 1: the four steps of Heuritech’s methodology

Step 1) Define audience panels on social media

The first step is defining what and whom to analyze. Heuritech selects publicly accessible Instagram accounts using the following criteria: i) accounts demonstrating interest in fashion; ii) accounts that post regularly; iii) accounts that are representative of different style segments of the fashion market.

Contents from the selected account are sorted into three categories:

Edgy accounts: They are people with distinctive style and can be professionals (stylists, journalists, influencers) or industry authorities (luxury, sports, fast fashion, high street, beauty). Their content is very niche and they tend to post often. This represents the smallest segment.

Trendy accounts: People looking for the latest styles and that tend to help spread trends across the market. They are more numerous than edgy accounts but are much more rare and qualified than mainstream accounts.

Mainstream accounts: People looking for clothing choices and who tend to follow a trend rather than contribute to its creation. They post lee frequently and represent the largest panel.

Additional geographic panels are used to understand market dynamics and regional differences.

Image 2: Heuritech’s consumer panels

Step 2) Analyze trends

Once the panels have been determined, Heuritech’s proprietary image recognition technology analyzes images to categorize more than 2,000 details: shapes, attributes, fabrics, textures, prints, and colors. These components are then combined to allow Heuritech to precisely identify trends for each panel. By looking at the volume of each feature, Heuritech quantifies the past growth and evolution. This information helps merchandisers and designers understand how a particular product is being received, by whom, and in which markets around the world.

Step 3) Use of machine learning forecasting algorithms to predict trends

The next steps concern the prediction of future trends. Heuritech uses a series of forecasting methods.  and a master algorithm, whose purpose is to select the best combination of forecast methods, based on general information about the trend.

To be able to forecast trend evolutions that may emerge from niche trends, Heuritech developed an in-house deep-learning approach that analyzes what is considered “early signals”. They are slight changes in activity among edgy influencers that may influence future trends. Using these early signals, these algorithms detect the onset of new emerging trends before they reach the mainstream and realize their full market potential.

4) Enter the data into Heuritech’s platforms

The final step gives fashion brands a comprehensive view of market trends and product performance. The data collected is entered into Heuritech’s platforms:

• Market insights platform: customers can analyze the predicted visibility growth of multiple details (shapes, colors, prints and fabrics) and their forecasted market share. Quantitative analyses of trends are complemented by fashion experts’ qualitative analyses of fashion weeks, emerging designers and brands. Platforms’ users can also compare forecasts in different geographic areas for each style and detail.
• Product insights platform: for each product of their mix, brands can access data on the brand adoption among the target audience. Brands can also compare their products’ performances (visibility, scope and desirability) with those of competitors.

Image 3: The product insights platform

In conclusion, given the positive results that Heuritech has had among the best brands in the world⁴, it can be said that it’s really succeeding in revolutionizing trend forecasting in the fashion industry. Heuritech has managed to go beyond the classic trend forecasts imposed by designers during fashion shows, quantifying the impact that new generations of trendsetters have on the industry through their activity on social media.

References:

1.        Trend forecasting specialist Heuritech raises €4 million, eyes international expansion. Accessed April 18, 2022. https://ww.fashionnetwork.com/news/Trend-forecasting-specialist-heuritech-raises-4-million-eyes-international-expansion,1132820.html

2.        Trend forecasting with AI in 2022: Fashion’s way forward. Accessed April 18, 2022. https://www.heuritech.com/trend-forecasting-fashion-ai/

3.        About Heuritech – The analytics platform for fashion brands. Accessed April 18, 2022. https://www.heuritech.com/company-about-us/

4.        ShopTalk: Heuritech Uses Artificial Intelligence To Predict Fashion Trends From Millions Of Images. Accessed April 18, 2022. https://www.forbes.com/sites/jeanbaptiste/2018/03/20/shoptalk-heuritech-uses-artificial-intelligence-to-predict-fashion-trends-from-millions-of-images/?sh=27c7645d2a90