If two employees were doing the exact same task – I mean like-for-like work – would you expect to get the same results? The same productivity?
Your instinct probably tells you yes, but in reality, it’s not the case at all.
While it should be true that two employees doing the same work at the time do it the same way, they rarely do.
The numbers don’t lie
Recently, our team of data scientists used machine learning algorithms to analyse nine million employee interactions.
The results revealed a 2.5 variance in how long it took them to do the same thing.
This means that if it takes you 10 minutes to do a task, it might take the guy beside you an additional 25 minutes to complete the identical task.
What’s more, the exercise highlighted a daunting variance in quality too.
Delving deeper, we limited the artificial intelligence (AI) parameters to look at the variance between the 25th and 75th percentiles so that the outliers would not skew the findings.
We removed those who were ridiculously fast and those who were tortoise-like slow in order to improve the accuracy of the observation.
Even then, the alarm bells still rang loudly, with the variance in productivity stark.
It was a troubling discovery, especially when, in reality, there isn’t a good excuse for employee productivity being so different.
Employee variance can take on many forms, but we concentrate on productivity so that companies can expand their capacity in a given period of time.
With margins collapsing, the idea that labour is cheap is a complete myth, so we’re finding that many companies want to know how to get more from every hour worked.
This begs the simple question: how long does each and every task need to take for you to improve the overall speed and, ultimately, do more per hour worked?
Employee variance measures if you’re using your workforce in accordance with your expectations. If you think something should take 10 minutes, does it?
A.I. can be used to determine things like this. It is also useful for spotlighting areas where you’re wasting time and for predicting how you can use your workforce better.
Variance measures how spread out a data set is and enables you to measure how far each number in the set is from the mean.
This is calculated as the difference between the actual labour hours used to complete a task and the standard number of hours that should have been used.
For example, if the 25th percentile is 10 minutes, the 50th is 17 minutes and the 75th percentile is 35 minutes (2.5 times the 25th percentile), then the variance is 18 minutes.
Minutes may not seem like much, but those minutes add up to hours ‒ sometimes in the hundreds of thousands.
Variability in how long it takes to complete a task is regularly overlooked by leaders ‒ excused with explanations or, worse, accepted as normal.
Yet, it is actually costly to your bottom line so you need to act: it is important to reduce the variance to extract maximum productivity from your employees.
For the employees who unfavourably diverge from the mean ‒ i.e. when they take longer ‒ you need to improve the way they work.
Here, don’t focus on merely making them better than they were, they need to match your mean.
Over a century ago, we learned that there was an actual “best way” to work.
However, most employees work differently from the best way and instead argue that their way is better, even when it’s not.
This problem was addressed through the creation of the assembly line and other manufacturing era initiatives that introduced standardisation.
Despite this, variance lives on and is widely spread throughout modern work environments.
Measuring productivity and reducing employee variance are two of the practices that never made the transition out of manufacturing.
This begins to explain why two employees doing like-for-like work can produce results that are anything but.
Like-for-like work should be consistent in time and quality.