A very common challenge is that organisations are unable to monitor the effectiveness and efficiency of their internal processes in real-time. At Wimmy we have developed analytical workflows for several of our clients, as well as ourselves. Based on our experience, here are 6 recommendations for crafting data analytics that will delight and empower your leadership team:
Top-down, not bottom-up
Take a step back and put your CEO hat on. Leaders don’t want to dive into the details straightaway. They primarily want two things: a broad overview, and clear recommendations about where, what and how things need to be changed. So start with producing simple metrics and data visualisations that summarise the organisation’s productivity, efficiency and growth over time. Then gradually zoom in on the interesting, surprising or troubling parts of the picture that demand action.
Script, not point & click
If your analysis is on point, it won’t be useful just as a once off. It will need to be repeated over and over again. Depending on the metrics and the context, managers and CXOs will want to see daily, monthly or quarterly figures. So avoid having to retrace your steps in a manual fashion. Instead, develop scripts that can be rerun by a scheduler, so you can orchestrate the creation of reports at any frequency without incurring extra manual labour.
Expectation, variation, correlation
In that order. Start by calculating aggregate statistics that communicate the average or expected answer. Where that average falls short of the organisation’s targets, quantify the variation around the average. How big is the gap between the best and worst performing subunit, between the best and worst month this year, between the smallest and the largest contract? And lastly, where the gaps are large and the opportunity to improve things are the greatest, analyse the data in more detail to understand which factors correlate the strongest with outcome, so the organisation knows where to focus its efforts.
Stratify and adjust
People are not good at thinking in more than 2 dimensions. This means that confounding factors are often not taken into account. Stratifying and adjusting statistical models for additional variables is therefore essential to avoid invalid conclusions being drawn from the low-dimensional analyses. A simple example: imagine an organisation in which productivity decreased over the past few months, despite a concerted effort from everyone to be more productive. But over the same period, many new hires were made, and new hires are known to be notoriously unproductive in the first few months of their tenure. Without adjusting for the start date of employment, workers and managers may receive flack instead of praise for their efforts.
Let the machine learn
We may have some preconceptions and hypotheses about which factors are driving organisational efficiency but our intuition is likely biased and incomplete. By allowing machine learning models to learn from all of the available data, they can produce objective predictions with quantifiable accuracy of what the likely effect would be of changes that the organisation’s leadership is considering.
Communicate and iterate
As much as the data and its analyst can generate insights and recommendations, others in the organisation have additional knowledge. Therefore, by communicating the results of the analysis early and widely, everyone has a chance to ask questions and share hypotheses for why the data are what they are. Such two-way communication helps the analyst to improve and expand the value of the analytics by creating additional data tables, modifying statistical models and developing new models.
Too often we see analytics teams spin their wheels, working hard to achieve a somewhat mediocre Return On Investment. With lots of big data routinely available, it is easy to fall into the trap of producing an ever growing list of figures and tables, without any of these being acted upon by the leadership of the organisation. These tips should help in crafting the real-time, comprehensive, incisive, insightful and predictive analytics that leaders need to make the right decisions and improve the organisation’s operational efficiency and growth potential.
