I think it would still help for a company like that to display stats. For example, if you’re a qualified candidate who is applying early, you’d see an abnormally high probability of callback. If you’re an unqualified candidate, you’d see a very low probability of callback. I mentioned that ML can help calculate these probabilities. In addition, I think it might help for companies to secretly encode “hard” job requirements. Most requirements are in fact soft, but the applicant never knows which is which. If they don’t meet a hard requirement, they should see a 0% probability of callback. Full stop. ML would do a good job of assigning importance weights to each of the requirements which aren’t encoded as “hard,” but it would start from weights provided by the employer when there isn’t enough data.