Harbingers of Failure [pdf]
marketing.wharton.upenn.edu
marketing.wharton.upenn.edu
They don't care that they are buying a failed product because "failed" only means "didn't sell to a lot of people". The product is actually okay, and they bought it because it's new.
Might it be possible that the pattern of buying failed products is self-reinforcing? Failed products keep disappearing off the shelves. So the people who started buying them have to look for something else. This switches their brains into "look for new stuff" mode, so they find new stuff beyond merely replacing what disappeared.
Now, perhaps the problem is that most new stuff fails, so it is easy to get into this pattern.
Just because you are into buying new stuff and it keeps disappearing doesn't mean you're a predictor for failed products. Your purchase patter just isn't good predictor for what will become popular.
Another way to think about it is that what you really should care about is not the number of sales, but instead the the number of sales to people who have similar purchasing patterns to the typical consumer you hope to reach.
What's interesting is that the authors are arguing that sometimes having a lot of early sales is actually worse than having only a few. It's a problem if that large number of sales includes a disproportionately high percentage of "flop affinity" consumers. It would be better to have a small number of sales, but also with low percentage of high "flop affinity" consumers. The intuition behind this is perhaps that in order to reach so many of a relatively rare consumer type - the harbinger of failure consumer - you must have also reached a lot of the more typical consumers, and they decided your product isn't appealing to them. On the other hand if you have low sales, but only a small percentage of these sales is to harbinger consumers then presumably your product is appealing to mainstream consumers, and your sales problem can be fixed with relatively straightforward marketing and distribution tactics.
Maybe the tl;dr could basically just be: not all early sales are the same, so understand who your buyers are and how representative their purchasing habits actually are of the market you're trying to reach.
Pumps, valves, and hydraulic cylinders that don't leak are a post-WWII technology. Natural rubber doesn't tolerate oil. All the alternative to it (cloth packing, oiled leather, etc.) are worse, and hopeless at high pressures. This is an example of something that just doesn't work with natural materials.
It would be interesting to see a breakdown for many items that we take for granted to the key materials science breakthroughs that enable their functioning.