> If your best hire is somebody you had less confidence in, then it's a sign your confidence function is not fully tuned to what's causal to success.
I think it's unreasonable to think that you can guess so well as to how performance will be over 6 months-1 year+ periods in several hours of measurement.
> If your average tenure is < 2 years, there's something wrong
My average tenure was much higher than that. But again, about 1 in 6 didn't make it a year. Turnover rate past that year was very low.
I'm not sure why you think having 15-20% of people not working out in the first year (after which turnover drops) translates to an average tenure of less than 2 years. It implies one greater than 5 years.
> If your best hire is somebody you had less confidence in, then it's a sign your confidence function is not fully tuned to what's causal to success.
If I had a fully tuned confidence function, it would output 1.0 for candidates that would work out and 0.0 for candidates that wouldn't. This isn't realistic.
Nor is it realistic for it to monotonically increase with candidate quality (in which case it could be transformed to the above function with a threshold).
> Finally, it's pretty bad for company culture and morale to have a high fire rate.
I don't think getting up to 1 in 6 (first year) freaks people out too bad. Rates across a broad variety of industries and employers are more than 1 in 3 quitting or being fired in the first 18 mos. I think one can do significantly better than this, but I also think 1 in 10 isn't attainable and question your numbers.
I think it's unreasonable to think that you can guess so well as to how performance will be over 6 months-1 year+ periods in several hours of measurement.
> If your average tenure is < 2 years, there's something wrong
My average tenure was much higher than that. But again, about 1 in 6 didn't make it a year. Turnover rate past that year was very low.
I'm not sure why you think having 15-20% of people not working out in the first year (after which turnover drops) translates to an average tenure of less than 2 years. It implies one greater than 5 years.
> If your best hire is somebody you had less confidence in, then it's a sign your confidence function is not fully tuned to what's causal to success.
If I had a fully tuned confidence function, it would output 1.0 for candidates that would work out and 0.0 for candidates that wouldn't. This isn't realistic.
Nor is it realistic for it to monotonically increase with candidate quality (in which case it could be transformed to the above function with a threshold).
> Finally, it's pretty bad for company culture and morale to have a high fire rate.
I don't think getting up to 1 in 6 (first year) freaks people out too bad. Rates across a broad variety of industries and employers are more than 1 in 3 quitting or being fired in the first 18 mos. I think one can do significantly better than this, but I also think 1 in 10 isn't attainable and question your numbers.