Business process observability: Alerting on process KPIs

Business process observability tracks the health and performance of the critical business processes that automate your business and serve as the foundation for digital transformation. By leveraging business data gathered at each step, business and IT teams can share real-time insights into process KPIs, throughput, and exceptions.

Dynatrace Business Flow simplifies business process observability, connecting top-level process KPIs with detailed flow analytics. The app tracks the progress of every process instance, reporting individual and aggregated process performance, throughput, exceptions, and business outcomes.

With this update, Davis® AI can track and alert on KPI threshold violations to assure end-to-end process efficiency and reliability.

Anomalies are inevitable

Despite the deep IT observability you might have deployed, many of your business processes will likely encounter failures, intermittent delays, or ongoing inefficiencies. These issues can impact a wide range of business outcomes, from a direct reduction in revenue to a decrease in Net Promoter Score (NPS), from idle production lines to departmental budget overruns. Each business unit relies on a collection of processes, and each process has metrics and KPIs that can be affected by delays, exceptions, or failures. Examples include:

• A stalled connection between two services delays the process of validating food orders, resulting in dissatisfied customers, idle delivery drivers, and canceled orders.
• A failure in a third-party API breaks a loan validation process, reducing the number of applications approved and directly impacting the company’s revenue.
• A delay in warehouse order processing results in late deliveries, impacting customer experience.

Business process observability

Real-time business process observability can minimize the impact of these and other incidents by ensuring anomalies are detected quickly and include the IT context required for rapid—even automated—remediation. Business process observability is a new approach to ensuring process efficiency and reliability, quickly detecting and remediating anomalies, and identifying optimization opportunities. Are detected business exceptions specific to one host in a cluster? Which services does the credit check API depend on? Is there a fund transfer bottleneck at the source or destination bank?

Business process observability becomes increasingly important as new regulatory frameworks like the Digital Operational Resilience Act (DORA) impose strict operational resiliency requirements on financial and other institutions.

Dynatrace Business Flow

Business Flow is a Dynatrace® App that delivers on the promise of business process observability, connecting top-level process KPIs with supporting step- and instance-level analytics. Some of the questions Business Flow answers include:

• What is the average end-to-end process delay? How does this change over time?
• Are the business goals achieved? (For example, average loan amount, service response time, or close rate.)
• What is the business exception rate? Is there a trend?
• Which process steps are most problematic? Which steps take longer than normal?
• Were optimizations successful in their goal of improving performance or reducing exceptions?
• How does one process path differ from another regarding exceptions and throughput?
• What is the current status of a specific customer order?

Business Flow: KPI alerts

The latest version of Business Flow introduces a significant enhancement to support anomaly detection and alerting. App users can configure alerts for any of the four reported business process KPIs, leveraging Davis AI Anomaly Detection for continuous KPI analysis.

Alerting can be configured for each of the app’s four business process KPIs:

• Conversions or fulfillments. How many process instances—claims, deliveries, payments, etc—have successfully completed the final step?
• Errors or business exceptions. Distinct from IT errors, these are process-specific exceptions, usually defined in the application and reported within the application payload.
• Average duration. This is the cycle time of process instances, the elapsed time from start to finish.
• Custom business KPI. This can be any user-defined metric extracted from process instances. Examples include revenue, order quantity, loan amount, and approval rate.

Configure anomaly detection

Use the time series charts from the KPIs over time view to configure Davis AI Anomaly Detection for any of the four key business process KPIs. Each KPI graph includes a Create alert button.

To create an alert for a KPI, select Create alert to open the Davis anomaly detection configuration page. Choose the analyzer type: auto-adaptive, seasonal, or static threshold.

The configuration window is from the Davis AI Anomaly Detection app, adapted for use with Business Flow KPIs. Since Business Flow automatically defines the DQL for the KPIs, the anomaly detection configuration is quite simple.

There are three Davis AI Analyzers to choose from:

• Static thresholds are fixed and don’t change over time
• Auto-adaptative thresholds are calculated by Dynatrace automatically and adapt dynamically to your data’s behavior
• Seasonal baseline thresholds adapt to observed seasonal baselines and are charted with a confidence band

Alerts can be triggered for any of the three Davis Analyzers whenever the metric falls above, below, or outside the defined thresholds. Dynatrace Documentation provides configuration guidance for Davis AI Anomaly Detection.

The actor—or simulated Dynatrace end-user operated by the Davis AI Anomaly Detector in automated workflows—can be a service user or an interactive user with privileges to run the anomaly detector and execute the DQL queries for the time series chart. For more details, see service users.

Once configured, the new Davis AI Anomaly Detector evaluates a certain metric each minute, creating problem events when alert conditions are met. The time series charts add alert annotations to facilitate further analytics in such cases.

What’s next

For more information about alerting on business process KPIs, please visit Dynatrace Documentation.

You can also try out the new Business Flow alerting feature in the Dynatrace Playground.

Note that anomaly detection currently supports process flows of less than 60 minutes. Support for longer-running processes will be added in a future release.