The paper is looking at these types of consumers in aggregate, effectively at the proportion of a product's sales that come from each group. These "harbinger"

The paper is looking at these types of consumers in aggregate, effectively at the proportion of a product's sales that come from each group. These "harbinger" consumers are only harbingers if, collectively, they make up a disproportionately large percentage of sales. 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.