Trend forecasting has become integral to the fashion industry, giving brands valuable insight into the colours and styles that will be most popular in future seasons. It can act as a roadmap for brands designing their next few collections, although the long gap between publication and products reaching the market introduces an element of risk.
As the pace of fashion speeds up and brands shift to smaller drops instead of bigger seasonal changes – enabled by smarter manufacturing with shorter lead times – traditional models for trend prediction and analysis will also need to adapt. In this rapidly changing landscape, a new startup founded by University of Leeds researchers aims to disrupt the colour forecasting market.
“It is really difficult to make predictions a year or more in advance, so it’s no surprise those traditional colour forecasts aren’t perfect,” explains Professor Stephen Westland. “Our data-driven method is based on what’s actually out there doing well now, fine-tuning the forecasting process now that we’re at the point of sale.”
Colour Intelligence, a spin-out company founded by Stephen and Dr Kaida Xiao, is building on years of colour research to develop more accurate colour trend reporting.
The company can rapidly analyse thousands of digital images from social media channels and fashion shows, using a combination of proprietary algorithms and machine learning techniques. The result is what the founders call a ‘nowcast’, currently published once per quarter with individual colour palettes for womenswear, menswear and interiors. June 2021’s nowcast has just been published.
Accurate image analysis and innovative approaches to colour design have long been central to the founders’ research, giving them years of expertise to invest in the new company.
Stephen is Professor of Colour Science and Technology at the University of Leeds, where he has been dispelling myths and misconceptions about colour for students and industry partners alike. As leader of Future Fashion Factory’s data-driven design research, colour design has been an important part of his work.
Along with his research group, he has continued to make progress with Colourpedia, a platform which uses similar machine learning techniques to generate a palette of colours associated with any given word, giving designers a valuable creative insight into how colours are understood by different markets to support their creative decisions.
That understanding of data-driven decision-making is combined with Kaida’s expertise on colour accuracy and imaging. As Associate Professor of Colour Imaging and Science, Kaida has worked extensively on improving the quality of colour communication for applications from medical imaging to online shopping.
Powerful insights drawn from the latest data will empower designers to thrive in a rapidly changing industry, Stephen explains.
“We believe that brands using smart manufacturing methods with vastly reduced lead times will be able to use our reports to sell the right colour at the right time,” he says. “If we are right, it will allow companies to sell a greater proportion of their stock without discounting.
“Ultimately this will reduce waste, and we’re motivated by contributing to work that can make the fashion industry more sustainable for the future.”
Additional nowcasts are being produced for events such as major runway shows and product launches, and the first ‘cheat sheet’ for London Fashion Week was produced this week. In time, the model could even be used to produce bespoke reports for clients.
Kaida says the model has huge potential to diversify as the company grows, focusing on different countries and product categories. “Colour forecasts might predict that, for example, blue will be popular next year, but small nuances in the hue can make or break the sales for a product,” he adds.
“Nowcasting is a powerful tool to help reduce the margin for error.”