Let’s face it; HR analytics can sometimes seem like a dark science, requiring lots of data know-how and lots (and lots) of algorithms.
But one notable analytics success I’ve seen was in a mid-sized company investigating diversity – and it barely involved anything more than some simple number crunching on the back of an envelope.
The question it sought to understand was this: ‘based on current practices, how many women leaders would there be in five years’ time?’
The basics of this predictive modeling are fairly easy to grasp. This (or any) organization would need to look at the percentage of women hired at various levels; it would need data about female attrition at each level; and finally, it would need to delve into rates of promotion for women – this last factor being affected by turnover at the senior levels and organizational growth.
So far, so good. But how should the analysis of all this actually be tackled? Well, if you’re an analytics pro, you might imagine building a model in a spreadsheet. If you are a data scientist, you might lean towards a Monte Carlo simulation (this is a risk management technique typically used to conduct a quantitative analysis of risks).
However, if you’re an HR business partner, with a limited analytics skill set but a keen understanding of the business issue you’re trying to solve, then you’ll probably do it on the back of an envelope.
Back of the envelope analysis
So how do you do this kind of modeling on the back of an envelope? Well, you just use the best estimates you have for hiring, attrition, and promotion and figure out roughly what the number of women will be at each level next year. You iterate four more times, and you have your five-year prediction.
Of course, since we’re doing the calculations by hand, they won’t be as precise as any mathematical model. Nor will it be practical to analyze many different scenarios.
However, in the organization I’m thinking of the HR business partner was trying to solve something very specific. The business question was this: ‘Given we’ve succeeded in bringing entry-level hiring up to 50% women, are we more or less on track in terms of moving toward our goal of having more women at the senior leadership level?’
The answer was no.
The reason the answer was no was simply due to the slow rate of turnover at the most senior levels and the length of time it takes for people to rise through the organization. Both things combined, it meant there would not be many more women in senior leadership in the next five years. Sure, in 30 years there would be a big change. But that’s a long way off.
So the business had one clear action to take: stop promising they would soon see many more women at the senior leadership level. Secondly, if it wanted to see any significant change in the next five years, current HR practices would not be sufficient. In short, the business would need much more aggressive, affirmative action.
There are two lessons to take away from this. One is that an HR pro who can’t do sophisticated predictive analytics shouldn’t shy away from numbers. Even estimates worked out by hand, and calculated on the back of an envelope, may be able to give the business what it needs.
The other lesson is that if you do have the capability to do a Monte Carlo analysis, you still have to ask if this is the best use of your time – especially if the business only needs a ballpark answer.
What I’m really saying is this: analytics should just be driven by the business issue, and not by the mathematical tools. A clear understanding of the issue is what drives real analytics success.