Despite the popularity of private labeling as a topic in the marketplaces market, few companies have been able to scale it to the $100 million a year level and beyond. There are many reasons for this, access to capital being one of them, but the key challenge is being able to find new ideas. Many have found one or two great products, but few have been able to build a repeatable process to keep launching new products.

In the 2013 story in the Fast Company magazine by Jason Feifer titled “The Amazon Whisperer”, he describes a company using the web as the source of ideas for their own products. They have assembled a team scouring the web, including Amazon, learning about all the features people wish a product had, and then hiring a manufacturer, often in China, to make the desired version.

“He has an entire team of people who read reviews on Amazon, looking for moments when people say, “I wish this speaker were rechargeable.” Pikarski then makes a rechargeable version. Hipe exists, in essence, because enough people think like me. It’s a profitable trick: C&A Marketing does “in the nine figures” in sales every year, Pikarski says, and grows at about 30% annually.”

Hipe is one of their product lines. The brilliance in this model is that they make new products with proven demand, instead of postulating on what customers might want. And even if the product is a flop, the risk to reward ratio is big enough to be worth it. There is always new product ideas to try.

Yet their approach can be taken even further - by automating the discovery of product flaws they currently do by hand. We think the most interesting sellers right now are the ones investing heavily into artificial intelligence. By building advanced data analytics and forecasting tools, these companies are able to process millions of possible ideas. Often the data source is the massive Amazon’s catalog, providing billions of reviews with ratings, and sales rank information.

Amazon, of course, also has this data, even more than anyone else, and using it to build Amazon Basics. But we think that among the millions of sellers there is going to be a few to do it better than Amazon itself.

For the last couple of years the leading marketplace sellers were using software to scale retail arbitrage. Pharmapacks for example calls their system the “Master Brain.” It is somewhat know how to automate retail arbitrage since it is a self-contained supply and demand problem, but the future is going to be the ones to figure out how to build a master brain for new products. Which also includes managing products once they are launched, including ads.

Stitch Fix, a San Franscisco-based e-commerce company, which sends customers boxes of preselected outfits, is leveraging artificial intelligence and staff stylists to learn what elements of style are popular. The proprietary algorithms developed by Stitch Fix allows them to analyze the data collected from their shoppers to learn which cuts, colors, and styles are the most attractive. The software then combines those elements into brand new designs. Guaranteed to be in demand.

Though Eric Colson, chief algorithms officer at Stitch Fix, said in an interview:

“Our business is getting relevant things into the hands of our customers. We couldn’t do this with machines alone. We couldn’t do this with humans alone. We’re just trying to get them to combine their powers.”

He has also said that those automated designs are “not even 1%” of their business, but they are expanding.

What’s interesting about this approach to building products is that it it’s not how companies built products for centuries before. Gone is the concept of a products line, a single product is enough. Gone sometimes is even the notion of a brand, customers buy based on features. Gone is guessing what people want, Amazon and other marketplaces are testbeds for ideas. This is a completely different approach to product creation, and thus existing brands are unlikely to be able to adopt, creating a huge opportunity in the market.