Retailers are using AI to get a jump on the next toy fad

By | 3 minute read | November 7, 2017

Traditionally, retailers have relied on a variety of knowledgeable individuals to spot trends. That's changing.

Fidget spinners were already a worldwide craze by the time big retailers figured out they needed to start selling them in the spring of 2017. At that point, demand was so high that retailers were left scrambling to get the toys from manufacturers to store shelves.

By June, the ball-bearing weights that allow the plastic spinners to move were in such short supply that a supplier for three major retailers was forced to bring bags of cash to Chinese factories to secure orders. Another supplier said he’d heard other factories demand 100-percent, up-front payment amidst the chaos. Some retailers were looking to spend extra to air freight spinners directly from factories to their warehouses because traditional sea shipping wasn’t fast enough for their immediate needs.

Fads are by nature fast-moving and therefore historically difficult for retailers to manage. But fidget spinners, and other modern fads with online origins, have the power to move faster than ever before, causing a mad dash to seize supplies and sales before interest wanes. The first spinner video made it to YouTube’s trending video section on February 12. Online interest in spinners exploded in the US in March and reached its peak in May. By June, interest was in decline.

“When these things break out they break out very fast. Demand surges and there’s a frenzy. They’re hard to predict,” Richard Gottlieb of Global Toy Experts told IBM.

Trends are certainly not impossible to predict though.

Fidget spinners, which were originally designed as attention aids for children, had flown under the radar for years without attracting much attention. But when a crop of YouTube videos popped up last year that featured teenagers doing tricks with spinners, interest began to spike online. Chinese factories, many of which employ workers to monitor social media and Google Trends, noticed and started producing the propeller toys.

Last October, Ryan Weaver, a former oil rig worker, noticed that the toys were popular on YouTube, and he decided to start selling some independently on Amazon. At the time there were only a handful of other sellers on the site. This May, he was selling hundreds of spinners a day for $17.95 a piece. Other early adopters with their eye on online buzz were rewarded similarly.

“Social media has this weird ability to create white hot demand over a very short period of time. The hard thing for businesses in that space is the demand will likely disappear just as quickly as it came,” Eric Johnson, dean of the Owen Graduate School of Management at Vanderbilt University, told IBM. “A few weeks lead time can be a big difference when it comes to getting the maximal revenue out of a fad.”

Traditionally, retailers have relied on a variety of knowledgeable individuals to spot trends. But that approach is inconsistent, said Danica Konetski, an expert with IBM’s Retail and Consumer Products Center of Competency.

“Everybody in an organization has an opinion about trends,” Konetski said. “But if everybody has an opinion, how do you know what’s actually going to be a trend and when it will happen?”

More companies are turning to cognitive computing to solve that puzzle. By monitoring unstructured social media data, companies can get smart, quantifiable insights on consumer trends and get a leg up on the competition. They can also use those insights to better understand a fad throughout its lifespan.

“Retailers need to make once they see a trend to hop on quickly enough but hop off quick enough that they’re not burdened with a lot of extra merchandise,” Toy Trends CEO Reyne Rice told IBM.

Insights from cognitive social media monitoring can help businesses manage ripples affecting their supply chains and adapt to a variety of disruptions, whether it’s a sudden fad or bad weather or a local event.

“Before we had cognitive analytics you really couldn’t use social media for anything other than qualitative information and you couldn’t use it to drive business decisions. Now with cognitive you certainly can,” said Michael T. Dobbs, an expert in consumer packaged goods and retail in IBM’s Global Center of Competence.