Terrible products that fail miserably delight us not only because of the time-tested humor of a spectacular pratfall, but because it’s satisfying to feel now and then that we’re not just a pack of Pavlovian dogs prepared to lap up whatever is fed us, especially if it’s a Colgate Ready Meal and a Crystal Pepsi.

In a really smart Financial Times column, Tim Harford takes a counterintuitive look at how companies can avoid launching surefire duds. The usual manner has been to find out which products representative people want, but he writes of an alternative strategy: Discover what consumers of horrible taste embrace and then bury those products deep in a New Mexico desert alongside Atari’s E.T. video games. Of course, it does say something that companies can’t just identify what’s awful. Why do almost all businesses become echo chambers?

An excerpt:

If savvy influential consumers can help predict a product’s success, might it not be that there are consumers whose clammy embrace spells death for a product? It’s a counter-intuitive idea at first but, on further reflection, there’s a touch of genius about it.

Let’s say that some chap — let’s call him “Herb Inger” — simply adored Clairol’s Touch of Yogurt shampoo. He couldn’t get enough of Frito-Lay’s lemonade (nothing says “thirst-quenching” like salty potato chips, after all). He snapped up Bic’s range of disposable underpants. Knowing this, you get hold of Herb and you let him try out your new product, a zesty Cayenne Pepper eyewash. He loves it. Now you know all you need to know. The product is doomed, and you can quietly kill it while it is still small enough to drown in your bathtub.

A cute idea in theory — does it work in practice? Apparently so. Management professors Eric Anderson, Song Lin, Duncan Simester and Catherine Tucker have studied people, such as Herb, whom they call “Harbingers of Failure.” (Their paper by that name is forthcoming in the Journal of Marketing Research.) They used a data set from a chain of more than 100 convenience stores. The data covered more than 100,000 customers with loyalty cards, more than 10 million transactions and nearly 10,000 new products. Forty per cent of those products were no longer stocked after three years, and were defined as “flops.”•