Performance Analytics

Best Arrow in the L&D Quiver

This is part II of our learning analytics blog series. Read Part I here

Once you’ve gathered learning, talent and performance data, you still must have the skills and experience to analyze the data and extract valuable insights. Gartner’s analytics maturity model helps describes the way we use data from simple reporting (descriptive analytics), to diagnostics, predictive and prescriptive analytics.

Gartner’s analytics maturity model applied to learning

Descriptive analytics describes what happened. It includes standard LMS reports, but this exploratory data analysis should also include talent and performance data such as a distribution analysis of last year’s policy violations.

Diagnostics analytics combines one or more data points to explain why something happened. The traditional learning impact evaluation study explores the correlation between the completion of learning paths and performance metrics.

In most cases, the latest version of Microsoft Excel provides sufficient functionality to conduct descriptive and diagnostics studies. In addition, Tableau, Power BI, or other online platforms provide excellent data visualization capabilities with relatively short learning cycles.

Predictive analytics is about what will happen, and it requires more advanced software such as R, Python or SAS. If these skills are not available in your L&D departments, a) you can hire new talent, b) check if IT has these skills and would support some of the analysis, c) partner with external resources such as RPS’ Performance Consulting team, or d) grow new skills within your team. An online Python boot camp will get you started for less than $10.

In the following are some examples of predictive analysis:

  • How can we predict attrition from individual learning, talent and performance data?
  • How can we naturally group learners into “clusters” sharing similar characteristics? For each of these clusters, we would then be able to implement targeted interventions and communication plans to drive their performance to the next level. We could even create cohorts of similar learners from within each cluster.

Clustering learners into logical groups with similar characteristics

Prescriptive analytics builds on the algorithm model to define how we can make it happen. The prescriptive analytics studies model decisions during what-if analysis. For example, what will be the impact on performance if all employees are certified up to a minimum level.

Predictive helps define the right individual and organizational interventions that maximize impact on performance. In the figure below, we see a minimum certification level has no impact on performance until 60%. From there onward, performance is modeled to increase. In the bottom half of the figure, we see the impact of a minimum experience level on performance — 5 to 10 years is the sweet spot with the best outcome.

What-if analysis allows to define the right targets and training standards

In summary, learning analytics requires that we gather new types of data and learn to analyze them with new tools and skill sets. Significant progresses in software and computing power turn what was once specialized skills into the hands of generalists. This turns learning analytics into a unique opportunity to address the skill challenges voiced by our CEOs, and without doubt, one of the best arrows in our quiver to strengthen our understanding of performance gaps and define data-driven interventions that deliver predictable, measurable outcomes.