How to: Set up Azure App Insights for your code app (preview)

Azure Application Insights is a telemetry and monitoring service that helps you collect and analyze detailed telemetry from your applications. When you integrate it with Power Apps code apps, you can capture metrics such as session load performance and network request summaries. This integration gives you deeper visibility into how your app behaves in real‑world conditions. For more information, see the Application Insights OpenTelemetry observability overview.

Prerequisites

• An Azure subscription.
• An Application Insights resource created in the Azure portal.
• The connection string or instrumentation key from your App Insights resource.

Steps

Use the following steps to provision Application Insights, install the SDK, connect telemetry in your Power Apps code app, and verify data in Azure. This short guide walks you through end-to-end setup so you can start capturing performance and network metrics quickly.

1. Create an Application Insights resource

   • Sign in to the Azure portal.
   • Go to Application Insights and create a new resource.
   • Copy the Connection String or Instrumentation Key from the resource overview.

2. Install Application Insights SDK

   Add the npm package to your code app project:

   npm install @microsoft/applicationinsights-web 
   

3. Initialize Application Insights

   Add the logic to initialize Application Insights:

   
   import { ApplicationInsights } from '@microsoft/applicationinsights-web'; 
   
   const initializeAppInsights = () => { 
   const appInsights = new ApplicationInsights({ 
      config: { 
         connectionString: 'InstrumentationKey=<YOUR_KEY>;IngestionEndpoint=<YOUR_ENDPOINT>' 
      } 
   }); 
   appInsights.loadAppInsights(); 
   appInsights.trackPageView(); // Optional: Tracks page view 
   return appInsights; 
   }; 
   
   

   Note

   Environment variables aren't yet supported for code apps. To manage per‑environment instrumentation keys, store values in Dataverse (for example, a settings table) or use getContext() to detect the environment and select the appropriate connection string from app constants.

4. Configure the logger

   Provide a logger so the platform can forward session and network metrics to your monitoring system. The platform calls the logMetric function with metrics like sessionLoadSummary and networkLoadSummary. You should only call this function once.

   import { setConfig } from '@microsoft/power-apps/app' 
   import type { IConfig } from '@microsoft/power-apps/app' 
   
   setConfig({ 
   logger: {   
      logMetric: (value: Metric) => {   
         appInsights.trackEvent(   
         {   
            name: value.type, 
         }, 
            value.data   
         );   
      }   
   } 
   }); 
   

   Additional guidance

   • logMetric type: Define it as (value: Metric) => void for clarity.

   • Logger interface:

     interface ILogger { 
     logMetric?: (value: Metric) => void; 
     } 
     

   You can import ILogger from: @microsoft/power-apps/telemetry.

   import type { ILogger } from '@microsoft/power-apps/telemetry' 
   

   Note

   This is just an example. You must implement logMetric so that metrics are sent to your monitoring tool of choice, not necessarily App Insights. The platform simply delivers metric payloads. Your implementation determines how they are handled.

5. View logs in Azure portal

   • Open your Application Insights resource.
   • Go to Monitoring > Logs.
   • Query the customEvents table to see session summaries and network requests.

   The SDK currently provides two built‑in metric types:

   type SessionLoadSummaryMetricData = { 
   successfulAppLaunch: boolean; 
   appLoadResult: 'optimal' | 'other'; 
   appLoadNonOptimalReason: 'interactionRequired' | 'throttled' | 'screenNavigatedAway' | 'other'; 
   timeToAppInteractive: number; 
   } 
   
   type NetworkRequestMetricData = { 
   url: string; 
   method:  string; 
   duration: number; 
   statusCode: number; 
   responseSize: number; 
   } 
   

6. (Optional) Log custom events

   You can log more telemetry (beyond default metrics) by calling your monitoring tool's own APIs at the desired points in your app. For more advanced scenarios, refer to the official Application Insights documentation for guidance. You should avoid logging sensitive or unnecessary data and always follow your organization's compliance guidelines.

Sample queries

To help you get started with analyzing your telemetry data, the following sample queries use Azure Application Insights Analytics (Kusto Query Language).

App open performance

This query shows app open performance metrics by day. Use this query to evaluate performance trends over time or after making changes. Use the 75th percentile of the timeToAppInteractive field to avoid noise caused by outliers.

customEvents 
| where name == "sessionLoadSummary" 
| extend cd = parse_json(customDimensions) 
| extend cm = parse_json(customMeasurements)  
| extend timeToAppInteractive = todouble(cm["timeToAppInteractive"]) 
| extend successfulAppLaunch = tobool(cd.successfulAppLaunch) 
| where successfulAppLaunch == true 
| summarize percentile(timeToAppInteractive, 75)  
by bin(timestamp, 1d)  
| render timechart 

Network request performance by URL

This query analyzes custom networkRequest events in Application Insights. It extracts the URL and duration from the telemetry. It then summarizes daily counts and 75th percentile response times per URL. It visualizes the results as a time-series chart.

customEvents 
| where name == "networkRequest" 
| extend cd = parse_json(customDimensions) 
| extend url = tostring(cd.url) 
| extend cm = parse_json(customMeasurements) 
| extend duration = todouble(cm.duration) 
| summarize  
count(), percentile(duration, 75) by url, bin(timestamp, 1d) 
| render timechart