Application Insights Overview

What Application Insights collects automatically

When you add the Application Insights SDK (or enable auto-instrumentation on Azure App Service), a range of telemetry flows automatically without writing a single monitoring-specific line of code:

Creating an Application Insights resource

Application Insights resources are created as workspace-based components, linked to a Log Analytics workspace.

# Create a Log Analytics workspace first
az monitor log-analytics workspace create \
  --resource-group myRG \
  --workspace-name myWorkspace \
  --location eastus

# Create the Application Insights component linked to the workspace
az monitor app-insights component create \
  --app myapp-insights \
  --location eastus \
  --resource-group myRG \
  --application-type web \
  --workspace /subscriptions/<sub-id>/resourceGroups/myRG/providers/Microsoft.OperationalInsights/workspaces/myWorkspace

# Retrieve the connection string (used in your application)
az monitor app-insights component show \
  --app myapp-insights \
  --resource-group myRG \
  --query connectionString \
  --output tsv

# Enable Application Insights auto-instrumentation on an App Service
az webapp config appsettings set \
  --resource-group myRG \
  --name myapp \
  --settings \
    APPLICATIONINSIGHTS_CONNECTION_STRING="<connection-string>" \
    ApplicationInsightsAgent_EXTENSION_VERSION="~3" \
    XDT_MicrosoftApplicationInsights_Mode="Recommended"

Custom events: tracking what matters to your business

Auto-collected telemetry tells you about HTTP requests and database calls. Custom events tell you about business outcomes: how many users completed onboarding, how many orders were abandoned at the cart stage, which features are being used. Custom events are the bridge between technical monitoring and business intelligence.

Custom events appear in the AppEvents table in Log Analytics and can be queried with KQL:

// How many orders were placed vs abandoned in the last 7 days?
AppEvents
| where TimeGenerated > ago(7d)
| where Name in ("OrderPlaced", "CartAbandoned", "CheckoutStarted")
| summarize EventCount = count() by Name, bin(TimeGenerated, 1d)
| render columnchart

// Funnel: what percentage of checkout starts result in a placed order?
let CheckoutStarts = toscalar(
    AppEvents
    | where TimeGenerated > ago(7d)
    | where Name == "CheckoutStarted"
    | count
);
let OrdersPlaced = toscalar(
    AppEvents
    | where TimeGenerated > ago(7d)
    | where Name == "OrderPlaced"
    | count
);
print
    CheckoutStarts = CheckoutStarts,
    OrdersPlaced = OrdersPlaced,
    ConversionRate = round(100.0 * OrdersPlaced / CheckoutStarts, 1)

// Custom event with properties: which payment methods are used?
AppEvents
| where TimeGenerated > ago(30d)
| where Name == "OrderPlaced"
| extend PaymentMethod = tostring(customDimensions["paymentMethod"])
| summarize Orders = count() by PaymentMethod
| order by Orders desc

Live Metrics Stream

Live Metrics Stream is one of Application Insights’ most practically useful features during an incident or deployment. It shows a real-time, sub-second stream of incoming requests, failure rates, response times, and server performance — updated every second without any query delay.

Use Live Metrics Stream when:

• You are deploying a change and want to see immediately if error rates increase.
• You are responding to an active incident and need real-time confirmation that a fix is working.
• You are load testing and want instant feedback on how the application is responding.
• You are debugging a specific user scenario and want to watch telemetry appear in real time.

Access it from Application Insights → Live Metrics in the portal, or via the CLI:

# Open the Live Metrics Stream URL for a specific component
# (navigate in browser — it requires WebSocket connection)
az monitor app-insights component show \
  --app myapp-insights \
  --resource-group myRG \
  --query id \
  --output tsv
# Then go to: https://portal.azure.com/#@<tenant>/resource/<id>/quickpulse

Smart Detection: automated anomaly alerts

Smart Detection uses machine learning to monitor your application’s telemetry and automatically detect anomalies — without you defining alert rules. It watches for:

• Abnormal rise in failure rate (based on historical baseline)
• Abnormal rise in exception volume
• Degradation in server response time
• Degradation in page load time
• Memory leak detection
• Slow database calls
• Security detection (potential security issues in dependency calls)

Smart Detection takes 7 days to establish a baseline. After that, it sends proactive notifications via email or action group when it detects something unusual.

Application Map: visualizing your service topology

Application Map is a visual representation of your application’s service topology, automatically built from the dependency telemetry Application Insights collects. It shows each service or component as a node, with lines connecting them for each type of dependency call. Nodes are colored by health: green for healthy, red for failing, yellow for degraded.

Application Map is invaluable for new team members who need to understand the system architecture, and during incidents for quickly identifying which node is the source of cascading failures across the topology.