Usage intelligence

Usage intelligence

CE.png

Licensing Information
The feature described in this documentation is optional and might not be part of the edition covered by your company's license.

Usage intelligence collects and analyzes usage patterns to detect issues proactively and identify unusual behaviors that may indicate potential problems or areas for improvement. The AI-generated descriptions of detected issues can then be used to notify the appropriate teams, along with suggested fixes to help prevent further problems.

Usage intelligence helps you get the most out of your usage data, transforming how you manage and interpret it. Instead of relying on fixed alerts, it continuously analyzes trends, detects early warning signs, and provides actionable insights.

For example, rather than simply notifying you when usage drops below a set threshold, Usage intelligence identifies patterns in real time and predicts potential issues before they occur: "Usage for Service X is trending lower than expected and is projected to fall below the threshold."

Benefits of Usage intelligence

  • Dynamic alerts – Alerts are based on trends rather than static thresholds, reducing the need for manual configuration.

  • Proactive issue detection – Potential problems are flagged before they fully develop, enabling preventive action.

  • Actionable insights – Notifications include possible root causes and recommended fixes, minimizing manual investigation and speeding up recovery.

Getting started with Usage intelligence

The first step is to determine the key metric you want to measure. This should be relevant to your industry, service, or customer behavior. Some examples are:

Movie streaming

  • Minutes watched

  • Number of movies watched

Ridesharing

  • Distance driven

  • Minutes driven

Printer manufacturing

  • Number of prints

  • Number of cartridges used

Next, ensure you can identify, access, and extract the required data to measure your chosen metric. Once your input data is ready, you can configure your streams to send it to Usage intelligence.

To train the machine learning (ML) models effectively, data must be processed in real-time or near real-time, with updates occurring at least once a day. The models are continuously trained, tested, and refined based on incoming data to improve prediction accuracy over time.

This ongoing process ensures that the models stay aligned with your data patterns and continue delivering relevant results.

You can access the Usage intelligence dashboard from the left navigation panel in Usage Engine by clicking the Usage intelligence iconScreenshot 2025-03-18 at 11.48.20.png. For more details, see Usage anomaly detection dashboard.