Very few of the metrics that are assessed during interviews are quantified (making bias correction very hard).
It's highly unusual (except maybe at Google) to compare how well someone is doing in a job to how well they were thought they would do that job.
The idiosyncratic nuances of most jobs is hugely underestimated (in the sense that the more specific the prior experience the more it should be discounted).
My suspicion is that we'd all do a lot better if we took all the candidates that applied that we thought "would do" and then drew lots.
We'd probably do even better if we agreed "no fault" severance packages in advance that could be triggered by either party.
It's highly unusual (except maybe at Google) to compare how well someone is doing in a job to how well they were thought they would do that job.
The idiosyncratic nuances of most jobs is hugely underestimated (in the sense that the more specific the prior experience the more it should be discounted).
My suspicion is that we'd all do a lot better if we took all the candidates that applied that we thought "would do" and then drew lots.
We'd probably do even better if we agreed "no fault" severance packages in advance that could be triggered by either party.