This data scientist’s perspective rings true to me
Read on timdellinger.substack.com/p/hey-wait-is-employee-performance
Tony – From the Outside
Hey, wait – is employee performance really Gaussian distributed??
This data scientist’s perspective rings true to me
Read on timdellinger.substack.com/p/hey-wait-is-employee-performance
Tony – From the Outside
From the outside 
Well, after almost 30 years I now know why my HR asked my team to see if we could construct a normal distribution for the performance of staff PS I thought it was a silly idea.
LikeLike
Hi Tony,
I remember this at Westpac – forced grading curves on the Gaussian distribution. I never interpreted it as being genuinely interested in getting things right or being fair. I always interpreted it as a tool to cull really bad performers and as a signal to push others to leave while instilling fear in the rest of the workforce. I don’t think much effort was applied to working out who were the bad performers. To me it was lets run this program for 3 years and directionally we will cull about the worst 30% of our workforce (broadly as we will make mistakes) and therefore our workforce after 3 years will have less shockingly bad performers and we can pause the program for 5 years before restarting it again.
As regards the folk at the top. We are Westpac not Goldman Sachs so if someone really is good and is more suited to GS they will leave anyway so the folk we are looking to retain (which is also largely the same as promote) are good performers but also those who fit the Westpac corporate culture (and by definition do not fit a culture outside the four majors).
Hence my conclusion. The article is likely correct and fits my experience as well, but whether the distribution is Gaussian or Pareto really makes no difference to Westpac’s objective in implementing it and whichever one they used would in practice yield the same result.
My person experience is that if you inherit a team of [8] people you have 2 good performers and 6 poor performers as per the Pareto principle. You then set about managing out the 6 poor performers and you will statistically get in 1.5 good performers and 4.5 poor ones (lets say you are good at hiring and get in 2 good and 4 bad), you then set about managing the four poor performers out and in the next round you may reduce poor performers to 2 or 3. The process is just a continuous one of getting rid of poor performers on an absolute rather than relative standard basis. I think if everyone is a good performer you want to get rid of no-one and if everyone is a poor performer you want to get rid of everyone.
An analogy here is a team promoted from the Championship to the Premier League in football. All teams ‘strengthen” their squads when this happens as they suddenly have access to a lot more cash and depending on how much risk they want to take splash different amounts of cash on new players, but also inherent in this is who within their squad is of premier league standard (or can be brought up to standard) and that is the biggest determinant of how many players get chopped
Best
Moray
Moray Vincent
Amicus Advisory
19 Bulga Road
Dover Heights, NSW 2030
Mob: 0416 266615
LikeLike
This post has hit a nerve I think. The other point often lost is that a statistical distribution which might have some science applied across 30000 employees is meaningless amongst a team of 10.
LikeLike