I agree with your first point partially—a broad funnel provides orgs with optionality. They can find good candidates even when the hiring processes are not optimized—e.g. job descriptions are not well written, etc.
That said, I do feel that the advantages of a broad funnel fall off when the number of applications/​position scales moves from 10-100x to >100x. This is more an intuitive reasoning—I do not have a good rationale for it. The closest is that every application pool is a biased sampling of the population distribution. Good sampling would provide distributions with means closest to the hiring needs. Just broadening the funnel would provide sampling distributions closer to the population distribution. Practically, this would mean a lot of manual filtering work to sub-select candidates for next stages of applications. In these scenarios, incentives of hiring committees and candidates might be more aligned.
You mentioned that feedback of the 95th or 10th percentile is useful, but not the 50th percentile. But I do think the feedback is actionable—it tells that the applicant is scoring average and they do not have any counterfactual impact for that position. I agree, it doesn’t tell the applicant how to improve, but it does provide some information on how they fare in the applicant pool.
However, I completely agree that real life is too noisy/​messy and applicants need more information on why they were scored in a particular way. Saulie’s comment shows a nice way this can be done—providing some information about scores with respect to the key requirements of the job.
My optimistic hope is that once orgs start doing a simple percentile rank-based feedback, they can be pushed towards more feedback with respect to the key application requirements. This is slowly moving the bottom line, from no feedback to something useful. And an automated, easy-to-setup percentile rank system might provide just a low enough barrier to get the ball rolling...
Great feedback, thanks for sharing, Vinoy!
I agree with your first point partially—a broad funnel provides orgs with optionality. They can find good candidates even when the hiring processes are not optimized—e.g. job descriptions are not well written, etc.
That said, I do feel that the advantages of a broad funnel fall off when the number of applications/​position scales moves from 10-100x to >100x. This is more an intuitive reasoning—I do not have a good rationale for it. The closest is that every application pool is a biased sampling of the population distribution. Good sampling would provide distributions with means closest to the hiring needs. Just broadening the funnel would provide sampling distributions closer to the population distribution. Practically, this would mean a lot of manual filtering work to sub-select candidates for next stages of applications. In these scenarios, incentives of hiring committees and candidates might be more aligned.
You mentioned that feedback of the 95th or 10th percentile is useful, but not the 50th percentile. But I do think the feedback is actionable—it tells that the applicant is scoring average and they do not have any counterfactual impact for that position. I agree, it doesn’t tell the applicant how to improve, but it does provide some information on how they fare in the applicant pool.
However, I completely agree that real life is too noisy/​messy and applicants need more information on why they were scored in a particular way. Saulie’s comment shows a nice way this can be done—providing some information about scores with respect to the key requirements of the job.
My optimistic hope is that once orgs start doing a simple percentile rank-based feedback, they can be pushed towards more feedback with respect to the key application requirements. This is slowly moving the bottom line, from no feedback to something useful. And an automated, easy-to-setup percentile rank system might provide just a low enough barrier to get the ball rolling...