The article asks the question "So how long do workers stay with their employers nowadays?", but that question isn't answered by the cited data source. The data are based on 'how long have you been with your current employer?', which of course underestimates the time you will actually stay with that employer (it's unlikely today is your last day). Imagine a world in which every employee stays at each job for exactly 2…
Good point, but your question also has an issue: If my answer is "2", you have no idea if it's almost 1, or almost 3, which is a huge span. I think a better question would be "how long did you stay with your previous employer".
The potential huge span for an individual answer won't affect the usefulness of the median.
Your question is interesting but would pollute any summary statistic with too much old and irrelevant data. Imagine someone who has had two employers: the first for 2 years, and the second for 45 years.