Data and automation have been at the forefront of predicting demand for existing products in certain regions of the world.
Now IBM has begun testing the use of data, analytics and artificial intelligence to forecast the demand of future products that have not yet been developed or designed, with what Steve Laughlin, vice president and general manager of IBM’s Global Consumer Industry, calls "a reasonable amount of accuracy."
The data comes from hyperlocal trends and patterns, as well as the key attributes that identify why a certain feature on a blouse, for example, helps it sell in a specific region of the world. Predicting whether or not certain items will sell in the future season using hyperlocal data and AI also helps marketers design campaigns that communicate with consumers about those products prior to each being developed.
When asked to cite the accuracy of this approach, Laughlin said it is “a lot more accurate than the human guess.”
Laughlin provided this example from his days working at Sears during the start of his career about 30 years ago. He ran hardware, paint and household goods, and had the freedom to buy and resell many items. One unique item was a wired clear-cased phone that lit up when it rang.
“I blew them out of the water, selling between 400 and 500 of them,” he said. “Everyone thought I was genius. Then I got in a wench set. They sat on the stockroom shelf for about a year and a half. When I was right, I was really right. But I was wrong more than I was right.”
The math works, he said. The technology, analytics and artificial intelligence has caught up to the business challenge, but it doesn’t take away from the creativity.
The idea to use data and AI came a couple of years after a dinner in Dubai with an unnamed iconic American fashion designer. Laughlin asked the designer whether he would use AI when designing clothing. He said yes, if it would help inspire creativity, and it would help him make fewer mistakes.
Data has been at the forefront of change, from product design to targeting advertisements. A few years ago, companies began to use hyperlocal data to forecast the impact of products. Now IBM uses the data around attributes to determine what will sell and how to sell it.
“This is a data crunching exercise, he said. “It’s about understanding what sold, why, where and to whom.”
One IBM client -- an apparel manufacturer and retailer based in Amsterdam with a business unit in the Far East -- now uses artificial intelligence to populate their content management system with descriptions of products. When the company introduces a new item, such as a women’s blouse with a low-cut neckline, the content is described by the AI system, rather than by a human.
It means when consumers search on the internet for an item from this retail manufacture there’s consistency in the company’s products. When products sell, or don’t sell, the AI aids marketers, merchandisers and designers at the company understand why certain items, based on color, style, location, store and other attributes sold or didn’t.
Data consistency and insight supports the analysis of trends. “We can look at reviews and comments, and use natural language processing to analyze commentary. We also can look at how competitors’ items are selling.”