• Request a Demo
  • Resources

Trendspotting: How to Predict Fashion Trends Before They Make it Big


In the 1930’s James Laver, an English fashion historian and author, developed a simple timeline to map the evolution of fashion trends. According to Laver, when a trend is in fashion it’s deemed ‘smart,’ a year before it’s a trend it’s ‘daring,’ 20 years later that same trend is ‘ridiculous’ and after 100 years the trend is ‘romantic’, a nod to the past perhaps.

Devil-Wears-PradaWhile Laver’s timeline neatly maps how our feelings towards trends change over time, social media (especially Instagram) and fast fashion chains like Zara, H&M and Topshop have vastly shortened the lifetime of a trend and given way to micro trends that emerge quickly and fall out of fashion even quicker.

While cycles have shortened greatly since Laver published his timeline, trends today simply don’t emerge out of thin air, a point made clear by Meryl Streep’s character in the Devil Wears Prada.

At the same time though, trend forecasting has become less predictable thanks to the advent of fashion bloggers like Chiara Ferragni of The Blond Salad. If Ferragni wears acme brand backpack on Monday, it’s likely that same backpack will be flying off shelves by the end of the week. While this type of ‘black swan’ trend is impossible to predict unless you’re the one donning the backpack, certain trends can be inferred from hard data.

Using Choice Modeling to Predict Trends

For large retailers that plan inventory assortments a year in advance, capitalizing on fleeting trends can be challenging. But, all hope is not lost. Social media, ironically the same channel that has made trend forecasting so unpredictable, can be analyzed along with omni-channel data like browse behavior, transaction information and inventory data to infer and predict trends among a retailer’s customer base even before they fully form.

While it’s easy for any casual fashion observer to spot a trend once it’s already emerged, from an analytical standpoint trends are fundamentally ill-defined - precisely why predicting them can be so challenging. This is only further complicated by the fact that trends are determined by individual choices and that people choose from the options they observe.

But, again there is a silver lining. Trends are generated by people and people are fundamentally simple in the choices they make (though not necessarily in how they arrive at those choices). So, while there are hundreds or even thousands of fashion trends at any given moment, there are really only a few different ways in which these trends arise.

For instance, take trending topics on Twitter. These trending topics arise because either a celebrity does something noteworthy, a natural disaster occurs or a popular event like the Oscars or Super Bowl takes place. Because there are only a few ways for trends to arise on Twitter, predicting a trending topic or hashtag before it emerges is possible.

For another example, consider predicting the price of Bitcoin, an instrument truly driven by human perception and speculation. Again, for similar reasons, the prices of Bitcoin are predictable, making it feasible to amass a sizable profit through trading.

Now, let’s go back to fashion trends. While people have many choices, they make these choices in only a few ways, therefore the resulting fashion trends emerge only from a few different patterns, which makes predicting fashion trends...predictable. These few choice patterns can be captured by observing the modes by which people choose - also known as choice modeling. And choice modeling, even using sparse data, is capable of identifying choice patterns. So, for retailers that harness the power of choice modeling, predicting and capitalizing on trends even seasons in advance is a real possibility.

Final Thoughts

On the surface, trends are hard to forecast, in fact they’ve become so unpredictable that many trend forecasting firms no longer predict trends, but instead use global networks of individuals to pinpoint trends that have already emerged. But when you dive deeper, trends are not as fickle as they might seem. Fashion by nature is evolutionary and cyclical. And, while those cycles may be shorter than ever, with the right data, a little intuition and a lot of science, forecasting trends is within the grasp of every retailer.

Free Download - Building Optimized and Hyperlocal Product Assortments with Omnichannel Data

Ready for more?

Request a Demo