> Now, it's understandable why they're using an inappropriate metric. It's the only thing they can even remotely reliably derive from their data set of scraped

> Now, it's understandable why they're using an inappropriate metric. It's the only thing they can even remotely reliably derive from their data set of scraped Linkedin profiles. I don't think even that is true. They could do a better metric like "Among the employees that were employed by company X ten years ago, what percentage are still employed there?" This also solves the "average tenure at a growing company" problem.