Use the insights feature to uncover problems and inefficiencies in your process. You can save your own findings as insights as well as get automated insights.

Insights have a snapshot of the data and a description of the discovery. The data snapshot visualizes the data at the time of capture.

Manual insights

When you discover bottlenecks or peak loads in your process data, you can save them as insights. Read how to add insights on an investigation and on a widget in section Capture manual insights.

Automated insights

Automated insights are generated by a built-in analytical engine. This engine automatically provides advanced statistics and data points of interest.

You can generate insights for a process or individual investigations.

Insights generation is based on the following:

  • Algorithms

    For example, algorithms detect correlations or anomalies in the data set. These algorithms are built in and can't be customized.

  • Metrics

    All metrics that were added to your process are used to generate insights. The more metrics there are, the more insights can you gain. Read how to add metrics in section Add metrics to a process.

You can save automated insights which are then added to the Insights management page. You can also turn them into widgets which are then added to an investigation. Read more in section Automated insights.

Anomaly detection algorithm

Anomaly detection is based on the identification of outliers. For that, each value is grouped based on the selected columns. The values that fall outside the range of the mean +/- twice the weighted standard deviation (95%) are identified and displayed in a bar chart.

Examples insights:

  • The net promoter score (NPS) is 30% below the average for a given country of interest.

  • The cycle time is 30% above the average for a given currency.

The following limits apply:

  • By default, no more than 10 insights on outliers are displayed. Users can load further insights.

  • 1 insight is created per metric and column pair.

  • A column must have 2–500 distinct values to be considered for insights generation. The number of rows isn't limited.

Correlation algorithm

Correlation determination is based on the Pearson Correlation Coefficient.

The algorithm calculates the coefficient for all pairs of case level attributes with numeric values. Custom attributes are also considered. Insights are provided when the calculation produces one of the following results:

  • The coefficient is equal to or greater than 0.5.

  • The coefficient is equal to or less than -0.5.

Insights are displayed for the 5 strongest correlations.

Read more on custom attributes in section Custom attributes for event-level analysis.

Who can work with insights

All users who have access to your process in SAP Signavio Process Intelligence can work with insights as follows:

  Manager Analyst Consumer
Manually create insights yes yes yes
Generate automated insights yes yes yes
Edit insights yes yes yes
Comment on insights yes yes yes
Save generated insights as widgets yes yes -
Delete insights yes yes yes

Effects on insights

Filters of investigations or widgets

Filters are automatically applied to the following operations:

  • Capture insights manually

  • Generate automated insights

  • Save insights as widgets

If filters exist for an investigation or a widget, the filters are stored in all created insights. To view the filters applied to an insight, you need to open the insight.

Changes to filters aren't applied to existing insights. When you change a filter, you need to create new insights.

Read how to create filters in section Filter process data.

Data views

Data views control the data for which users can create insights. Also, users can only access insights for data views to which they have access.

If insights exist for a data view, changes to the data view aren't applied to the existing insights. So, the data snapshot is still visible to anyone with access to the data view. If you want to restrict a data view, we recommend to create a new data view and re-assign the users.

Read how to generate insights for a specific data view in section Automated insights.

Read more in data views in section Define access to process data with data views.