In the early 1980s I invested in one of the earliest online retailers, Comp-U-Card. Comp-U-Card succeeded and went public by understanding that customers were only willing to buy hard goods, such as autos, TVs, air conditioners, etc, and would not buy fashion items and accessories. In the early 1990s shoppers grew comfortable with buying fashion online. Today we buy everything online, in part thanks to the success of Amazon.
Many have published stories recently about Amazon pulling its diaper product, including Venture Beat here. Venture Beat also lists other product or merchandising failures, such as a payment wallet and a phone. Everyone talks about these failures as if Amazon was a retailer. I think that is the wrong way to think about Amazon. Amazon is an exchange operator that facilitates the widest range of suppliers and customers entering into transactions. Amazon's expertise is in the infrastructure to support the exchange, which perhaps explains why they have been so successful providing a non-retail service like cloud services.
When Amazon starts to think like a retailer and introduce private label products, such as diapers, they move far away fom their expertise in IT infrastructure and fail. This logic would also explain why wallets and phones were not successful. Retailers succeed in part by selecting merchandise for their target customers. Exchanges succeed in part because their infrastructure supports the needs of the exchange users. While some expertise is needed at a product level, many exchanges tend not to be discriminating. This is the case with Amazon's strategy to sell everything, from consumer products to boat parts to building supplies. Think of Amazon as an exchange and not a retailer.