How to Make Business Intelligence Data Smarter with Learning Analytics

What’s a BI tool, and how does it typically fit into a learning ecosystem?

A business intelligence tool (BI tool) processes and visualizes data across your business. Most medium-to-large organizations have at least one BI tool, such as Power BI, Tableau, or both.

Unlike learning analytics platforms (LAPs)—which are specialized to work with learning data stored as a stream of learner interactions (in a data format called xAPI)—BI tools are designed to work with tabular data from any department in the organization.

(NOTE: LAPs can also often ingest tabular data, converting this tabular data to an xAPI data structure to store it.)

The BI tool sits outside the learning ecosystem, is typically owned by IT or the data team, and often receives data from the learning ecosystem (much like the LAP). However, the BI tool isn’t generally set up to receive granular learning data like the LAP.

Because of the LAP’s functionality to aggregate and organize learning data, some organizations use the LAP to link the learning ecosystem and the BI tool.

This approach ensures that data coming into the BI tool is easier to work with because the LAP has already aggregated and processed the data—converting it from an xAPI format that it’s stored in, into the tabular data that the BI tool expects.

This processed data is imported into the BI tool as a CSV file and includes any xAPI data stored in the LAP, as this data can be transformed into tabular data for the BI tool to ingest.

Why should I use a learning analytics platform and a BI tool?

The BI tool is unlikely to be the source of any learning data, but there may be technical or commercial reasons why it’s easier to get learning data from your BI ecosystem rather than the source system.

The BI tool may also be the best data source for business metrics and job performance data that can be used for measuring the business impact of learning.

For instance, you need data relating to business KPIs. In this case, the best option is a CSV download, ideally automated via API. Some examples of potential business metrics are shown below alongside worker competency data.

If you already use an LAP for reporting on learning data, you may still need to export that data to a BI tool so learning reports can be presented alongside reports from other departments.

Those responsible for generating reports or the audiences for those reports (such as senior executives) may be accustomed to generating and viewing those reports via the BI tool—which is why learning data needs to be available and accessible.

As a general rule of thumb, the LAP is the ideal reporting platform for the L&D team and others responsible for development and skills (including line managers). But more senior managers looking at reports from multiple departments—not just L&D—may prefer to stick with the BI tool.

Some organizations have found that using the BI tool to filter and organize L&D reports based on organizational hierarchy is complex and time consuming. This is because the BI tool needs to be reconfigured to use that organizational data for each report generated.

By contrast, because LAPs are focused on learning data which is all about people, the ability to filter, sort and control access to data based on organizational hierarchy information is critically important.

(For this reason we’ve made sure that in Watershed, using organizational hierarchy data for reporting is as easy as possible and does not need to be reconfigured every time you set up a new report.)

Generic BI tools absolutely have their place, but domain-specific analytics tools—like an LAP in the domain of learning—are better placed to work with data in their areas of specialization, which means you also save time and gain deeper insights.

Many of Watershed’s biggest clients use Watershed alongside a BI tool, selecting the most appropriate tool to generate reports and dashboards for each audience.

In most cases, however, the BI tool’s reports are based on data that has been aggregated and processed in Watershed, rather than the BI tool aggregating the data from the various source systems themselves.

So why export data from a learning analytics platform to a BI tool instead of collecting data from the original sources? Here are four reasons:

1. The data is already aggregated in the LAP. Data from multiple sources has already been brought together and processed, so the BI tool only has to work with one data source.
   
2. The data has been integrated with the organization hierarchy. If the LAP has imported organizational hierarchy data, you can include it with data extracted to the BI tool—removing the need to merge that data after. In some cases, this is more than a convenience and may be a legal requirement.
   
   For example, a data team in one territory only has permission to process data from their territory. The LAP may provide a solution to prevent access to that restricted data so only people who have permission to access data about certain people are able to do so. The original source may not have the functionality to restrict data in that way.
   
3. The data has been checked and cleaned. The process of setting up the initial data connection with the LAP typically involves a certain amount of data cleansing and addressing issues. Further, because LAP reports use the data, it is constantly checked and reviewed by LAP users.
   
4. The data can be extracted in a tabular structure. Much learning data, particularly xAPI data, uses an interaction stream format that may be difficult to work with using a BI tool. The LAP can process and flatten that data into a tabular format, including just the data points you want to focus on in the BI tool.

BI tools and your learning ecosystem

Options for getting data out of the LAP and into the BI tool will depend on the specific functionality of the LAP. Watershed has three options for getting data out and into a BI tool, which I’ll share below as an example. Other LAPs may have similar functionality.

1. Data Export API. The Data Export API delivers data from the LAP in the format that it’s stored with no processing of the data. This option may look attractive—you’re getting the pure, original data pulled into the BI tool to process there.
   
   In practice, however, we’ve found that organizations that tried this approach ultimately found this raw data too difficult to work with. They eventually switched to using the Aggregation API instead.
   
2. Aggregation API. The Aggregation API delivers processed data similar to that which immediately underpins Watershed’s reports and visualizations. This option is helpful because you can filter, organize and aggregate the data using Watershed as a step before sending it to the BI tool.
   
   This aggregated data is generally easier to work with, and you have control over precisely what data you send to the BI tool and what data you want to ignore.
   
3. Connector. Watershed developed an integrated Web Data Connector for Tableau that creates a live, updatable connection between a Watershed report card and a Tableau data feed. From a data standpoint, this works similarly to the Aggregation API—except it’s only for Tableau, and it’s a little easier to set up. (NOTE: You can develop connectors for other BI tools, too.)

By aggregating data in the LAP, you may be able to smooth over any changes when replacing a platform in your ecosystem—minimizing the impact on data flowing to the BI tool.

The role of BI tools in skills and compliance

When senior management requires compliance reporting, it may be necessary to produce official compliance reports in the BI tool.

In this case, the LAP can still play a role in:

• aggregating the data for the BI tool, and
• providing real-time reports and dashboards to managers and the L&D team (which means they can monitor compliance progress ahead of the BI tool creating official reports).

Similarly, senior leaders may wish to receive a regular high-level report on skills created by the BI tool. Again, the LAP can supply the aggregated data to support this process and offer more granular reporting to managers and L&D.

Up Next: L&D data requirements for instructor-led training

BI tools and learning analytics platforms both hold data that is valuable in the other, and setting up a data integration between the two is very beneficial. One source of such learning data is instructor-led training (ILT), whether virtual (vILT) or in a classroom.

Next time, we explore the data requirements for VILT and ILT and consider how to get your learning data when the learning is happening in a physical space, rather than a digital space.