Models pose in neutral linens for Plan C

Fashion trend forecasting: Beyond guesswork

Key takeaways

  • Historically, fashion brands and retailers have used a mixture of analytics and guesswork to make decisions in the collection planning process.
  • According to a 2020 report from McKinsey & Company, fashion brands and retailers “who leverage analytics are outperforming their competition by 68%.”
  • Data-backed trend forecasting provides fashion brands with more precision and efficiency when creating collections.

The fashion industry has always relied on a mixture of analytics and guesswork to make decisions for future collections. Seasonal runway shows and past sales data, for instance, are some of the ways in which industry insiders traditionally informed collection-planning decisions, and for many years, this was sufficient. But in a time where digital transformation has become the status quo for much of the fashion industry, it has become crucial to take a more data-driven approach to trend forecasting. 

According to a 2020 report from McKinsey & Company, fashion brands and retailers “who leverage analytics are outperforming their competition by 68%.”

But why are data-based predictive analytics so advantageous, and how does this approach go beyond mere guesswork?

Backstage at Michael Sontag
Michael Sontag

The marriage of analytics and guesswork behind traditional trend forecasting

For the past several decades, traditional trend forecasting has relied upon an amalgamation of multiple sources to determine the best moves for upcoming collections: 

  • Designer opinions
  • Panels and questionnaires from consumers
  • Sales data
  • Social listening
  • Competitors’ collections
Plan C

But today’s landscape requires more precise trend forecasting than the classic methods. Consumer desires are changing more rapidly than ever, social media ignites and extinguishes trends overnight, competition on the market is oversaturated, and brands are expected to be as much human as they are business. 

As an example, let’s take a look at why past sales data are no longer sufficient if considered by itself. Every year, the market changes: new promotions, changes in demand, new products, trending values, and more render it difficult to compare between seasons.

If by chance one product is wildly successful and the brand sells out of it, this sales data can easily be misleading for the following year’s potential demand due to the lack of reliable data. Furthermore, trends have a life cycle like anything else, and what is successful one year won’t necessarily be the following year. Miscalculating demand is indeed a major cause of waste, one of fashion’s biggest issues: up to half of all products are discarded at factory sales, shipped to other markets, or even burned.

Historical sales data on their own are no longer a reliable source of future trend performance.

Of course, fashion will always rely on some form of intuition. Guesswork isn’t a bad word — opinions and gut-feelings are important, too. Designers’ and agencies’ trend predictions, for instance, are important following a season for other designers, brand teams, and even consumers who closely follow trends. But this approach, too, has its limitations. At the end of the day, these opinions remain subjective, and more importantly, there is often anywhere from one to two months’ delay between time of observation and time of prediction. 

Data-based trend forecasting

More and more brands and retailers are beginning to see the value in predictive analytics, or data models which are created using algorithms and machine learning. In the simplest terms, this method answers the question: What will happen? 

When used for trend forecasting, predictive analytics can help brands understand what their customers really want. Brands, in turn, can put this data into effect to make decisions about design, assortment, marketing, and more to stay in tune with their customers and remain relevant season after season. 

But does data-based trend forecasting really work? 

In our case, data-based trend forecasting is based on an artificial intelligence technology which scans images on social media and identifies over 2000 attributes, including color, pattern, and silhouette. This data is then used to predict trends and their future behavior among different consumer segments, seasons, and geographies.

Let’s take a look at an example of a trend to best illustrate the power of this trend forecasting compared with traditional methods. 

Explore emerging brands, upcoming trends, and fashion e-commerce in Africa

Trend use case: Combat boots

For this Winter 2021, our client wanted to know how combat boots would behave among women in Europe. Using image recognition technology, we scanned millions of social media images to determine the trend’s behavior this Winter compared to last Winter. 

This trend forecast predicts that combat boots will…

  • Rise in visibility by +19%, which marks a strong growth compared with the same season last year. 
  • Appeal most to edgy consumers, or those who tend to adopt trends seasons ahead of their mass market adoption. 
  • Reach their high season in February, signalling to brands that this is the ideal month  during which to push this trend. 
Heuritech trend forecast for combat boots for women in Spring 2021 in Europe

These are the kinds of insights that data-based trend forecasting provides to brands — the above information renders collection planning decisions far easier and more efficient for the teams involved. Indeed, trend forecasting allows teams within a fashion brand to become more aligned with one another, an important time-saver given that the process isn’t always streamlined. 

Trend forecasting aligns teams within a fashion brand

It’s no secret that creative-minded and analytical-minded teams aren’t always on the same page.

While fashion is a fundamentally artistic field, a brand couldn’t operate without its analytical teams, like its merchandisers and marketers. 

Data-based trend forecasting has the unique ability to align these teams: it backs designers’ intuitions and guides decisions rather than making them, and gives merchandisers and marketers a quantitative reason for collection decisions. Trend forecasting can be seen as a common language which unites teams within a brand, a benefit where traditional forms of forecasting fall short. Merchandisers may not consider the opinions of journalists and designers, just as designers may not consider historical sales data — data-based trend forecasting takes the separation and guesswork out of the collection planning process. 

The future of fashion trend forecasting with social media

Ultimately, trend forecasting has no choice but to rely on data, particularly data drawn from consumers as they often express themselves through social media. Brands are no longer the only deciders of trends — consumers have taken control through social media, and they make themselves heard. Their desires, muses, and values are all written across social channels if one knows how to look. 

Trend forecasting based on social media is truly the window to the consumer in their rawest form, and this data can’t be passed by. In the words of McKinsey and Company, “the winners will be the ones that can best harness data to inform core business decisions—and advanced analytics is a critical lever to make that happen.”

About the writer: Mélanie Mollard, Fashion Content Manager at Heuritech

Mélanie writes about the fashion industry and its many features through the lens of applied data and trend forecasting technology.

Questions or feedback? Email us at

Subscribe to our newsletter

4,191 fashion experts receive consumer insights and reports each Wednesday. And you?

trend forecasting platform

Get contacted by one of our consultants