Outlier Detection for SAP (Configuration)

⚙️ Configure Outlier Detection for SAP

📖 Overview

GermainUX identifies unusual behavior in monitored SAP applications by comparing current activity with historical patterns or configured thresholds.

Outlier detection can be applied to telemetry collected from:

Source
SAP Fiori, SAPUI5, Web GUI, and compatible web applications
SAP Commerce Cloud storefronts
User sessions and business processes
APIs and integrations
Application logs
Hosts, services, processes, and databases
Synthetic transactions

GermainUX analyzes data collected through configured monitoring components and integrations. It does not natively profile ABAP code.

✅ Prerequisites

Before configuring outlier detection:

Prerequisite
Deploy the required GermainUX monitoring components.
Verify that the relevant telemetry is being collected.
Configure the SAP application, business process, or infrastructure KPIs.
Collect enough representative data to establish normal behavior.
Exclude maintenance, testing, and other nonrepresentative periods.

Depending on the use case, telemetry can be collected through RUM JS, JS Profiler, GermainUX Engine, RPA Bot Recorder, Code Profiler for supported Java services, or custom integrations.

🔍 Common Outliers

GermainUX can detect unusual changes in:

Area	Examples
User experience	Page load time, user-action duration, network latency, JavaScript errors
SAP Commerce Cloud	Search performance, cart errors, checkout duration, abandonment, conversion
Business processes	Execution time, failure rate, completion rate, step duration
Applications	Error volume, HTTP failures, response time, throughput
Integrations	API latency, timeouts, retries, failed requests
Infrastructure	CPU, memory, disk, process availability, database performance
Synthetic monitoring	Transaction duration, failures, availability

🔧 Configure Outlier Detection

Open the GermainUX configuration for the relevant KPI, application, or business process.
Select the measurement to analyze.
Define the application, environment, transaction, or other dimensions used to group the data.
Select the evaluation and comparison periods.
Configure the sensitivity and minimum data volume.
Exclude maintenance, testing, and known nonproduction activity.
Save and enable the configuration.

The available fields depend on the GermainUX version and the type of telemetry being analyzed.

💡 Configuration Guidelines

🧩 Use meaningful dimensions

Create separate baselines for workloads that behave differently, such as:

Dimension
Production and test environments
SAP applications
SAP Commerce Cloud sites
Business processes
Pages, routes, APIs, or transactions
Regions and device types

Do not combine unrelated transactions into the same baseline.

📊 Require sufficient volume

Configure a minimum number of requests, sessions, errors, or process executions before evaluating an outlier. This prevents low-volume fluctuations from generating misleading results.

🔁 Account for recurring patterns

SAP activity may vary by hour, day, business schedule, or batch-processing window. Compare similar operating periods when the workload has predictable patterns.

🔗 Combine outliers with thresholds

Use outlier detection to identify unexpected changes and fixed thresholds to enforce known service objectives.

For example, alert when checkout time is significantly above its normal range or exceeds the maximum acceptable duration.

🛒 SAP Commerce Cloud

For SAP Commerce Cloud, GermainUX can identify unusual changes in:

Area
Storefront performance
Product searches and no-result searches
Product and cart activity
Checkout duration and failures
Cart and checkout abandonment
Conversion rates
JavaScript and network errors
API and third-party service performance
Synthetic browse, login, cart, and checkout transactions

Use authoritative order data for confirmed orders and revenue. Browser events provide insight into the customer journey but should not replace server-side transaction records.

🔄 Business-Process Outliers

GermainUX can analyze configured SAP business processes when their steps are observable through browser events, logs, APIs, custom events, or synthetic transactions.

Examples include:

Process
Login
Order creation
Approval workflows
Purchase-to-pay activities
Customer-service workflows
Browse-to-purchase journeys
Cart and checkout

GermainUX can identify abnormal duration, failure, abandonment, repeated steps, or unusual transitions.

🕵️ Investigation

When an outlier is detected, correlate it with available evidence such as:

Evidence
Affected users and sessions
Session Replay
User actions
Browser and JavaScript errors
Network requests
Application logs
API and integration performance
Host, process, and database metrics
Synthetic transaction results
Business-process steps
Recent deployments

Cross-system correlation requires compatible timestamps and shared identifiers where available. GermainUX does not automatically create browser-to-ABAP execution traces.

🔔 Alerts

Configure alerts for sustained or significant outliers. Include:

Alert Item
Application and environment
Affected transaction or process
Current and expected values
Number of affected users or transactions
Duration and severity
Related errors
Link to the relevant GermainUX analysis

Prioritize alerts based on business impact, affected-user count, duration, and severity.

✅ Validation

After enabling outlier detection:

Confirm that the selected KPI receives data.
Verify the filters and grouping dimensions.
Check that the comparison period represents normal activity.
Confirm that low-volume periods do not create false positives.
Validate maintenance and test exclusions.
Review detected outliers and adjust sensitivity as needed.
Verify that alerts reach the correct recipients.

🔧 Troubleshooting

❓ No outliers are detected

Verify that:

Check
The KPI contains sufficient data.
The rule is enabled.
Filters are not excluding all telemetry.
Minimum-volume requirements can be met.
Sensitivity is appropriate.

❗ Too many outliers are detected

Verify that:

Check
Different applications and transactions have separate baselines.
Recurring daily or weekly patterns are considered.
Maintenance and test activity are excluded.
Minimum-volume requirements are configured.
Sensitivity is not too high.
Alerts require the condition to persist.

🛡️ Privacy

Do not collect passwords, authentication tokens, payment information, or unnecessary personal data. Apply GermainUX masking, exclusion, access-control, and retention settings to all monitored SAP telemetry.

ℹ️ Get More Information

GermainUX can help determine which monitoring, analytics and automation capabilities are appropriate for your SAP environment.

Contact GermainUX Support.

Component: Engine, JS Profiler, Mobile App, RPA Bot Recorder, RUM JS

Feature Availability: 2022.1 or later