Predictive analytics is often seen as the Holy Grail of talent analytics. Bersin argues that a journey to a mature talent analytics function is a continuing investment that moves HR from operational reporting to forecasting future talent outcomes. Similarly many HR practitioners are obsessed with the idea that predictive analytics is a solution to a better future- one that create better talent outcomes and efficiency gains.
I want to challenge that view and reframe the debate. First of all, ‘better talent outcomes’ shouldn’t be the ultimate goal of HR’s effort. HR goal’s should be tied and aligned to the higher organisational goals. The terms HR analytics, talent analytics and people analytics have often pigeonholed the work that HR is doing into ‘HR for HR only’, instead of ‘HR for the whole organisation’. Let’s talk about the five key mindsets that HR can change to win the heads and hearts of stakeholders across the business.
1. Organisational analytics not HR analytics
For HR to become a strategic force in the business, it needs to broaden the scope of its analytics agenda to organisational design and start blending people data with business data. It needs to ask the question, how good is my workforce in executing the business strategy? This fundamentally lies beyond just looking at people data in your payroll or HR Management Information Systems (HRIS), but blending it with business data. By pulling these datasets together, HR will be able to give insights beyond simple HR metrics but to answer questions that will be valuable throughout the organisation. For example, they can start to analyse: relationships between employee absence and productivity; which competencies of the sales force best predict sales productivity; what the impact of various training programs are on various outcomes, e.g., CSAT, engagement, retention, productivity and so on.
2. Action not prediction
HR shouldn’t be seduced by the allure of predictive analytics. It’s not that prediction isn’t important, but in the world of HR analytics, the importance is overplayed. For example, Bersin’s model positions predictive analytics as the final stage of the analytics maturity journey. In reality, predictive analytics is not that difficult to do and can be done along with operational reporting and analytics. There are already tools in the market that can help people to do predictive analytics—performing scenario modelling or forecasting future headcounts against sales target. Having prediction is desirable, but the real value lies in taking action. Can HR use the foresight they get to take preventive actions? Is there a feedback loop that drives insight to action to result that in turn brings new or additional insights? Is the iteration cycle fast enough to be valuable?
3. Guide your analysis plan with hypothesis
Organisational data contains a goldmine of information (download chapter), but where do you start? How do you ensure the insights you look for are valuable to the business? Use a top down hypothesis-led approach. Start by defining the key objectives that stakeholders across the business want to prioritise. For example, if the business is trying to boost productivity through centralisation, build a series of hypothetical questions such as: are some departments more productive than others? Does training affect productivity? Where do our top performers come from? Once you build a rigorous hypothesis tree, you can then guide your analysis plan by focusing on the right metrics. Watch out for my next blog to understand the hypothesis-led approach in greater depth.
4. Data management platform is not one size fits all
Different types of analysis will require different data models and therefore different technology platforms. Managing relational data (e.g. organisational data) vs streaming data vs fast-moving, semi-structured data will require different platforms. For example, while a data warehouse is fit for storing streaming data, it is not suitable for scenario modelling. There is Hadoop for handling fast-moving, semi-structured Big Data and OrgVue for collecting, cleansing, and modelling organisational data on the fly. Other tools such as Alteryx is great for transforming and blending data, R for performing advanced statistics and Tableau for visualising data. A common mistake is to think there is a ‘one-size fits all’ platform that can solve all of your data purposes. Most of the time, in your analytics project, you will need a portfolio of tech platforms and a mechanism to move data across them – configuring which one is the master and the slave. Implementing a robust data management plan is fundamental for success.
5. Gamify and crowdsource
When you pull data together from various sources, there will be data quality issues. How can you make this process easy and valuable for all? Gamify and crowdsource the process. In other words, make it fun, interactive and visual for people to provide and alter data. Agree upfront on the rules on data structure, collection process and storage with your data owners and clean it centrally. There are tools in the market (including OrgVue – watch features) with fast-feedback, surveying and visualisation capabilities which can instantly update a person’s data record once completed. The key is to make the data upload process as quick and mobile as possible. For example, sending out a short webform that can be accessed through smartphones and tablets, will make it easy for data owners to provide input, review charts and add insights.
When HR gets these five mindsets right, HR will be able to make the most of organisational data and analytics. It will be an invaluable resource to the business that can drive leaders to think through what their people do now, what they need to do in the future and what the impacts will be.