Companies like Google and Amazon have blazed a trail in their use of machine learning, data science and predictive analytics. Now the same technology is everywhere. This means that any organisation can use existing people (or other) data to make better decisions and inform strategies. Make no mistake, this is a game-changer for businesses. But how does it all work and where should you start? We’ll show you, and explain what we can do to help.
Rise of machine learning…
Machine learning certainly isn’t brand new but it has got fresh momentum. This has come from the technological revolution and a huge increase in mass market computing power. It is how Amazon knows which products to recommend to you based on prior purchases and how Netflix suggests other shows related to what you’ve just watched. These companies are continually ‘feeding’ data into algorithms to learn more about consumer trends and provide future recommendations.
This brings me to what is the essence of machine learning – namely, learning about patterns within existing data in order to predict future outcomes based on new data. Applying machine learning to data is often known as conducting ‘predictive analytics’.
Comparing old and new
The transformative power of predictive analytics
But while predictive analytics has massively impacted other fields; helping to predict stock market trends and detecting diseases, for HR it is still in its infancy.
A great example that’s close to home for us is what it can do for employee research. It enables us to easily process and make sense of multiple data points to reveal deeper insights to our clients.
This can be transformational for businesses. If you’re interested, below we’ve outlined how we’ll be using predictive analytics to gain a more detailed understanding of employee survey data alongside other key people metrics.
So, exactly what kind of predictions could you make? Well, we could apply machine learning algorithms to your people data and predict the future trends of your company’s KPI’s. This is done in broadly two stages, as follows:
1. Understand:What are the defining characteristics of a high or low-performing employee in your organisation, for example? These could be employee survey perceptions (as would be the case for a Key Driver Analysis). But machine learning provides detailed analysis of the data to provide a far more comprehensive picture – for example, by distinguishing between the most significant demographics and showing which keywords in the commentary supplement the model’s prediction.
2. Predict: Based on information learned about your organisation today, it can then make predictions based on new data. Attrition is a good example as, once the model has learned the defining characteristics of employees leaving your organisation, for the next survey it identifies and flags employees who match the criterion of somebody who has previously left the organisation. Naturally, there are data security and GDPR compliance considerations here but, fear not, there are ways to provide this data in a legitimate and useful way!
Consider this as an advanced and more comprehensive version of employee profiling/segmentation. You can use it as a ‘Predict’ analysis or as a standalone exercise to better understand employees by uncovering patterns in their survey responses.
It involves creating distinct employee profiles where, within each profile similar organisational perceptions are shared, but between each profile organisational perceptions differ most strongly. Once each profile is created (there’s usually positive, negative and neutral), we support you to understand:
- Who these employees are (age, gender)
- Where these employees are (hierarchically)
- What these employees are specifically saying (thoughts, feelings and commentary from free text)
- How to motivate each profile (e.g. what drives their engagement)
- Why you should focus on them (e.g. their connection to performance outcomes).
The ‘probe’ phase is where it gets really exciting! This involves deeper exploration of your people data. As an example, this could be extracting meaning and sentiment from free text comments in employee survey results, which has typically been an onerous, manual task. Once upon a time this would have involved each comment being reviewed and categorised according to their primary sentiment or theme. Manually. Very time-consuming.
But predictive analytics is opening many doors. While we’re not personally there just yet, there is scope to completely automate this through the application of machine learning algorithms that extract meaning from text data. Doing this means:
- Overall sentiment (positivity) can be detected in each comment, making it easy to explore qualitative insights throughout surveys
- Automating our standard comments (thematic) analysis, providing organisations with the richness of this information in a straightforward format
- Keywords can be extracted and used to support both prediction and profiling algorithms.
Ultimately at a time when every competitive advantage counts, predictive analytics is something you can ill afford to ignore.
So, if you’d like some expert guidance on how your organisation can learn about today in order to predict what’ll happen tomorrow, please get in touch!