What’s next in fashion has never been as fuelled by outside forces as it is today. Decentralised from a pool of incumbent brands, publications, and long-standing figureheads, the power to dictate trends has rerouted to consumers, and the inspirations behind those trends are being found not on runways, but in the context of their everyday lives. 

Last year, that context changed for us all. The fashion landscape had become unpredictable prior to 2020, but COVID shone a light on just how far the source of future trends had pivoted away from the carefully planned collections and cyclical deadlines that the calendar had been built on before.

As a consequence, a genuine demand – some might say a necessity – has emerged for a model of fashion that runs incredibly close to the pace of consumer expectations.  And to meet that demand, retailers and brands require a whole new level of insight and intuition.

Spurred by the pandemic’s once-in-a-lifetime disruption to retail, in May last year two separate campaigns calling for a shorter timeline from fashion week to store delivery were spearheaded and signed by leading RTW designers. Both acknowledged a fundamental thing: that traditional fashion calendars lacked proximity to more immediate market demands.  Both also revealed a critical fear: that consumers were falling through the gap between runway and retail, by either losing interest from the point a style was revealed to the time they could actually buy it, or by satisfying themselves with a fast fashion copy that arrived in half the time.

At the same time, the last year has borne out the opposite idea – demonstrating that growing the ability to react to rapid changes in consumer sentiment, or building a hypersensitivity to cultural fluctuations, can equip designers with the tools they need to succeed.  Timely responses to these sorts of shifts has quickly become the transformative strategy for industry winners.  Where agile brands innovated rapidly last year to capitalise on the tectonic shift in what people were buying, slower companies struggled to shift unwanted stock as customers began to shun reliable well-performing categories such as footwear, accessories, and evening wear.

For trend services, whose task is to provide brands and retailers with accurate foresight, the fundamental unpredictability of the fashion market means their roles are more significant yet more scrutinised now than ever before.  Nobody can claim to know the future inside and out, but in retrospect, 2020 was a year filled with revisions and reactive decisions, which support a sceptical view of the industry’s heavy reliance on past-data and third party prediction as the primary tools for attempting to predetermine future performance. 

Yet if ‘unprecedented’ was the defining word of last year‘, how far can forecasters be blamed for the discrepancy between what they thought would sell and what actually sold? Should they, or we, have been able to predict the impacts both big and small that these changes would have on the future market, and could they have been expected to preemptively help clients to plan their collections for an eventuality that hadn’t occurred for more than a century prior? And should the industry be looking for alternative, more flexible ways to try and predict the future?

In hindsight, it’s easy to say that the solution rests with shorter, sharper, and less inflexible responses to real-time data rather than human intuition or prognostication. But the task of pinpointing and coordinating the ‘right’ data sources remains really complex. That is why we’re beginning to see a new crop of fashion trend services use AI to reinvent the traditional data analytics process – collecting real-time information from different sources, and synthesising recommendations from those data that they then present as being better alternatives to human insight.

To some artificial intelligence (AI) forecasters like Heuritech and T-Fashion, social media is the ‘right’ data source. With the digital identities of consumers being so rich yet so fragmented between platforms, services like these take advantage of social media’s ability to provide comprehensive real-time insights that adapt to consumer-driven content, and try to scrape relevant information from them all.

This approach seems to make great sense.  According to T-Fashion, ‘80% of consumers decide whether to buy a product or service on Instagram’ a statistic likely achieved through the app’s global accessibility, e-commerce integration, and focus on visual content. However, when there are 400 million users of Instagram stories per day, it is clear to see the challenges presented by infinite scrolls. What does it mean to sort the signal from the noise?

There is also an ever-growing number of social media platforms built upon new talking points, communities, activism and entertainment. All of which simultaneously generate a mammoth amount of data in different formats, from text to image to video. So whether you are making decisions a mass-market retailer with a global consumer base, or a small independent brand who needs to find their audience in an oversaturated home market, extracting meaningful information from volumes of information that large is beyond human capability.

AI has long been earmarked as the solution to the inefficiencies of forecasting – not least because of its ability to address huge, ever-changing datasets like those generated by social media. By independently recognising patterns in past and present, micro and macro data at scale, AI and machine learning can autonomously refine and readapt its predictions as the future market context shifts. Image recognition and computer vision (both subsets of machine learning) only sophisticates this process by giving a visual language to these forecasts. 

Now, as designers begin to acknowledge just how much control consumers have over trend creation, their relationship with foresight is ripe for change – mounting pressure on trend services to utilise AI more readily. It was forecasters who predicted the customer-driven digital transformation of the industry after all, but the ability to move from macro-level conclusions like these two ever-shifting, seasonless insights is probably going to be contingent on forecaster’s ability to blend their creative abilities with the discipline of data science.

Fashion, of course, has already footed heavy investment in AI, although it has been focused mainly on front end solutions. These often centre around quantitative results that demonstrate improved retail performance, such as automated eCommerce A/B testing, smart product recommendations . But the back end integration of predictive technology into creative roles is still at an early stage, and it will mean a seismic shift in how retailers and brands develop their products.

As the technology continuously maps cultural changes on global, national, and regional scales, for buying and merchandising teams, there is the opportunity to unpack geographical market insights to distribute more precise product quantities and design variations.

For design teams, smart forecasting could take the form of the delivery of more frequent and smaller collections that engage with immediate future trends, whilst providing contingency for products with longer lead times to market. This planning of product positioning in both the short and long-term future will help creatives to tether their ideas to commercial viability. 

And if AI were to be implemented across the entire retail value chain, we could see less linear and more iterative design processes that interact with customer needs. Subsequently, elevating the efficiency of offerings such as customisation, tailoring and personalisation with predictive capabilities.

In short, there’s a lot than AI can offer when it comes to providing the raw, unfiltered insights that could underpin a whole new kind of prediction, planning, and forecasting.  But brands will also still need to carve out an aesthetic identity within these predictive forecasts and contextualise AI-generated insights with their own data, to underpin decisions that still reflect their DNA. If not, they run the risk of too heavy a reliance on intelligence, and the dilution of the essence of their brand.

This time last year, ACNE Studios collaborated with AI artist Robbie Barrat on their A/W 2020 Menswear Collection. Using AI to directly assist in generating design ideas by filtering in handpicked data from archival collections, they were able to weave results into their design development process and still emerge with garments that were unmistakable theirs. Although this takes AI a step-beyond a tool for forecasting, it does highlight the possibilities in interrogating predictive models with more specific data-sets, and in using machine learning generated insights as a launchpad for creativity, rather than a strict framework. By tailoring forecasts and reacting to them in their own way, brands and trend services could more granularly scope the performance of designs and their finer details, and also respond to both small-scale and sweeping changes without sacrificing what makes them unique.

When technology advances, there are always concerns about human roles being cast into obscurity. But it is essential not to overestimate AI. Planning for the future is never going to be a complete science; it should be the tool for bolstering intuition rather than replacing it. And as more trend services and brands begin to experiment with predictive technology to develop unique working methodologies, we will see the emergence of a new breed of futurists and designers who use data but are not ruled by it. 

Harnessing AI stands to strengthen the agency of these emerging creatives. By combining innovative new methods with and tried and tested techniques, they can safely explore the realistic possibilities of their own instincts and find clarity within the increasingly ambiguous fashion system. 

So, to any industry insiders on the fence charged with predictively planning for what this year will bring, consider this: AI’s fundamental strength is its ability to capture the unexpected and shine a light on the unknown. Both are valuable assets for forecasting now we operate in the legacy of the lessons the world learned in 2020.