How to apply a data science approach to HR

null

Analyst Josh Bersin recently described people analytics as “the lynchpin of success for HR”. It’s not hard to see why. A recent survey from HR Magazine found that 47% of HR professionals feel like they cannot progress to CEO because the perception of their function among senior leaders would prevent them from getting there. There has been a longstanding concern within HR that they are failing to communicate in the language of the boardroom: in facts, figures, ROI and business metrics. 

Data analytics is helping HR to bridge this gap, furnishing teams with insight into the costs and success rate of everything from individual recruitment ads to engagement initiatives or training programmes. 

While the infancy of the market has prevented HR taking the sophisticated approach to data analytics seen in other departments, such as finance or strategy, in the coming years we’re due to see an explosion in programmes targeted at the function. These will measure people performance in all areas – from sales through to teamwork. Alongside data and insight from third parties, such as employee benefits providers, these tools will provide HR teams with unprecedented insight into their workforce. 

This data and insight will only have value if HR professionals have the capability to analyse and draw meaningful conclusions from it. To do this, they need to approach HR like a data scientist. So what steps can they take to achieve this?

1. Start at the end

A data scientist wouldn’t dive into a set of a data without first establishing what outcomes they are looking for. Likewise, HR teams need to understand what problems they are trying to solve, be they people issues or business challenges. Trying to do everything at once by exploring reams of data will get you nowhere. Instead, define your specific outcome: for example, I want to know the impact of benefits on retention. From here you can work out what data you need and what questions you need to ask of that data.

2. Set about exposing untruths

The next step is to articulate those questions, in the form of hypotheses. The generally agreed scientific method is that you can’t ever truly prove anything – just get closer to the truth by disproving things that are untrue. Continuing the theme of benefits and retention, our initial hypothesis would be that employee benefits have no impact on retention rates. Benefits professionals – myself included – know better. But that hypothesis is easily disproven. And the data that helps you disprove it, such as a correlation between certain benefits and retention rates, will point you in the direction of your next hypothesis, showing you where you should look in more detail. 

3. Challenge yourself

This brings me to my next point: don’t just use the data you have readily to hand. Top-level insights on benefits take-up and retention might help you disprove your first hypothesis, but it’s more likely to say that a relationship exists than explain why it exists. To really explore the trends and complex relationships that underpin your hypothesis, you need to closely examine whether the data you have is deep and rich enough for the task at hand. If it’s not, then work out where the gaps are, and how to fill them. 

4. Little and often

Most people tend to veer towards big projects: they are challenging and therefore rewarding when completed, and they also give you a chance to demonstrate your skills and abilities. But, when looking at people analytics, we should go the other way. Little and often is key. 

Run regular, small projects to disprove small hypotheses, and build up a steady body of evidence. This will also mean failures are less damning. Rather than waiting until you’re halfway through a big project before realising your hypothesis is wrong, you can take your small failures in your stride and move onto the next test.

5. Don’t be afraid to fail

This ties in well with my previous point: failure isn’t always a problem. It’s usually a critical step in the journey to success. As you analyse your data, you won’t always get the results you want. The important thing is to work out why something isn’t working and take that on board for your next test.

Sometimes, your hypotheses will seem unsolvable. When that happens, there’s no shame in changing it or moving to the next one – not everything will be easily discovered through data. If you return to it later, you might find that you’ve gained new skills – or better yet, valuable new data – that help you tackle it.

6. Build a data-based argument

Eventually, you’ll build up a body of evidence that you can use to support your arguments and suggestions. Now it’s time to present your findings to your decision-makers, drawing on the right data and facts to make an undisputable business case – and ultimately, effect positive change in your organisation.

With HR technology evolving at such a pace, it’s unrealistic to assume that all teams will have the skills or inclination to turn themselves into fully-fledged data scientists overnight. It takes time to build affinity with new technology, but this is time that HR professionals must invest. There’s no doubt we’re entering a new age where the potential of the function is being recognised. Bringing data-based arguments to the boardroom table, will be key to seeing this realised.    

Jack Curzon, Head of Scheme Design at Thomsons Online Benefits 

Image Credit: Monkey Business Images / Shutterstock