Harbingers of Failure: Meet The Customers You Don't Want

Being able to predict anything - especially commercial disaster - ought to be worth something to the consumers possessing those traits. Though whether they would be willing to be labelled in that way should be a fascinating experiment in economic theory. JL

Cathleen O'Grady reports in ars technica:

A group of customers whose support for a product is a “harbinger of failure,” a signal that the product will eventually flop. They're people who are more likely to buy the niche products that account for a low proportion of overall sales. Harbingers have preferences that are systematically different from other customers
A central premise in marketing seems so obvious that it doesn’t even bear scrutiny: if customers give you positive feedback on your product, that’s good. And if those people buy the product repeatedly, that's even better.
But what if certain customers just don’t have great taste? Or, more precisely, what if their tastes don’t match up with those of the rest of the population? Positive feedback and early sales from these customers might actually not be good news—they could be a sign that the product’s going to tank.
A recent paper in the Journal of Marketing Research has identified a group of customers whose support for a product is a “harbinger of failure,” a signal that the product will eventually flop. “Increased sales of a new product by some customers can actually be a strong signal of future failure,” researchers write. So who are these people?

Doom visible in the data

The researchers set out with twin goals: figure out how to identify people with quirky taste, then see if their purchasing data could be used to predict whether a product is likely to succeed for fail. To do this, they gathered data from a large retailer. They analyzed the customer loyalty card data of nearly 130,000 individual customers as well as the overall transaction data (including those purchases made without a loyalty card) from 111 stores. The data spanned nearly seven years and excluded seasonal products.
First, they needed to figure out which products had failed, so that they could see who had bought them. All of the products in the data set had made it through pilot stages and had been rolled out across the retail chain, but only 40 percent were still around three years later. Anything that disappeared within less than three years was labelled as having failed.
Then the researchers established which groups had frequently bought these products within the first year of their life. These are presumably the people who’d liked the product the most, and so their taste clearly differed from the mainstream, which had generally rejected the product.
Finally, they looked at what happened to other products that this group liked. Sure enough, products purchased by harbingers of failure don’t do so well in the long run: “If sales to these customers are high, the product is more likely to fail,” the researchers write.
The more harbingers of failure like the product, the worse this effect is. If they buy a product repeatedly, it’s even more likely to fail. And the logical opposites also hold true: if harbingers of failure avoid a product, it does pretty well, and popularity with a different group of consumers—the "harbingers of success"—spells good news.
The interesting thing here is that the researchers tweaked their definitions to see if the effect was robust, or whether it fell apart when they changed something small. They reran the tests, changing what constituted a failed product or what length of time they considered the "introductory phase," yet they kept finding the same thing: a certain group of customers was associated with a lot of failed products.
They say that their study can be used to help market research, because when data from harbingers of failure is incorporated into predictions of product success, the predictions get more accurate.

Niche tastes

So who are the harbingers of failure? The researchers don't have any especially insightful information on them, other than that they're people who are more likely to buy the niche products that account for a low proportion of overall sales. “Harbingers have preferences that are systematically different from other customers,” the researchers explain. If something appeals to them, it’s unlikely that it will appeal to the mainstream.
It’s a similar concept to a well-understood problem with tech products, where early adopters often have different preferences to the mainstream markets. This can result in really innovative technologies having initial success and then tanking, because they don’t have widespread appeal. Even though preferences for things like food and beauty products are often much simpler, similar mechanisms are at play.
It’s important to note that the researchers aren’t implying that certain people cause a product to fail. It’s not as though mainstream customers are steering clear of a product because of who else is buying it. But because harbingers of failure have reliably weird taste, their desire for a product can be used to predict the likelihood of the product's eventual fate.
The implications here for manufacturers and retailers are interesting, because they suddenly complicate how positive feedback should be understood. It could save companies a lot of money to identify failed products earlier, or avoid producing them in the first place. Discontinuing flops early in the game reduces the opportunity cost of selling more successful products instead.
It might be a pragmatic choice for retailers, but as anyone knows after experiencing a beloved product being swiped from the shelves, it’s hard luck on anyone with niche tastes