Automate custom reports with Application Insights data

Periodical reports help keep a team informed on how their business critical services are doing. Developers, DevOps/SRE teams, and their managers can be productive with automated reports reliably delivering insights without requiring everyone to sign in the portal. Such reports can also help identify gradual increases in latencies, load or failure rates that may not trigger any alert rules.

Each enterprise has its unique reporting needs, such as:

  • Specific percentile aggregations of metrics, or custom metrics in a report.
  • Have different reports for daily, weekly, and monthly roll-ups of data for different audiences.
  • Segmentation by custom attributes like region, or environment.
  • Group some AI resources together in a single report, even if they may be in different subscriptions or resource groups etc.
  • Separate reports containing sensitive metrics sent to selective audience.
  • Reports to stakeholders who may not have access to the portal resources.


The weekly Application Insights digest email did not allow any customization, and will be discontinued in favor of the custom options listed below. The last weekly digest email will be sent on June 11, 2018. Please configure one of the following options to get similar custom reports (use the query suggested below).

To automate custom report emails

You can programmatically query Application Insights data to generate custom reports on a schedule. The following options can help you get started quickly:

Sample query for a weekly digest email

The following query shows joining across multiple datasets for a weekly digest email like report. Customize it as required and use it with any of the options listed above to automate a weekly report.

let period=7d;
| where timestamp > ago(period)
| summarize Row = 1, TotalRequests = sum(itemCount), FailedRequests = sum(toint(success == 'False')),
    RequestsDuration = iff(isnan(avg(duration)), '------', tostring(toint(avg(duration) * 100) / 100.0))
| join (
| where timestamp > ago(period)
| summarize Row = 1, TotalDependencies = sum(itemCount), FailedDependencies = sum(success == 'False'),
    DependenciesDuration = iff(isnan(avg(duration)), '------', tostring(toint(avg(duration) * 100) / 100.0))
) on Row | join (
| where timestamp > ago(period)
| summarize Row = 1, TotalViews = sum(itemCount)
) on Row | join (
| where timestamp > ago(period)
| summarize Row = 1, TotalExceptions = sum(itemCount)
) on Row | join (
| where timestamp > ago(period)
| summarize Row = 1, OverallAvailability = iff(isnan(avg(toint(success))), '------', tostring(toint(avg(toint(success)) * 10000) / 100.0)),
    AvailabilityDuration = iff(isnan(avg(duration)), '------', tostring(toint(avg(duration) * 100) / 100.0))
) on Row
| project TotalRequests, FailedRequests, RequestsDuration, TotalDependencies, FailedDependencies, DependenciesDuration, TotalViews, TotalExceptions, OverallAvailability, AvailabilityDuration

Application Insights scheduled digest report

  1. Create an Azure Function App.(Application Insights On is required only if you want to monitor your new Function App with Application Insights)

    Visit the Azure Functions documentation to learn how to create a function app

  2. Once your new Function App has completed deployment, select Go to resource.

  3. Select New function.

    Create a new Function screenshot

  4. Select the Application Insights scheduled digest template.


    By default, function apps are created with runtime version 3.x. You must target Azure Functions runtime version 1.x to use the Application Insights scheduled digest template. Go to Configuration > Function Runtime settings to change the runtime version. runtime screenshot

    New Function Application Insights Template screenshot

  5. Enter an appropriate recipient e-mail address for your report and select Create.

    Function Settings screenshot

  6. Select your Function App > Platform features > Configuration.

    Azure Function Application settings screenshot

  7. Create three new application settings with appropriate corresponding values AI_APP_ID, AI_APP_KEY, and SendGridAPI. Select Save.

    Function integration interface screenshot

    (The AI_ values can be found under API Access for the Application Insights Resource you want to report on. If you don't have an Application Insights API Key, there is the option to Create API Key.)

    • AI_APP_ID = Application ID

    • AI_APP_KEY = API Key

    • SendGridAPI =SendGrid API Key


      If you don't have a SendGrid account you can create one. SendGrid's documentation for Azure Functions is here. If just want a minimal explanation of how to setup SendGrid and generate an API key one is provided at the end of this article.

  8. Select Integrate and under Outputs click SendGrid ($return).

    Output screenshot

  9. Under the SendGridAPI Key App Setting, select your newly created App Setting for SendGridAPI.

    Run Function App screenshot

  10. Run and test your Function App.

    Test Screenshot

  11. Check your e-mail to confirm that the message was sent/received successfully.

    E-mail subject line screenshot

SendGrid with Azure

These steps only apply if you don't already have a SendGrid account configured.

  1. From the Azure portal, select Create a resource > search for SendGrid Email Delivery > click Create > fill out the SendGrid specific create instructions.

    Create SendGrid Resource Screenshot

  2. Once created under SendGrid Accounts select Manage.

    Settings API Key Screenshot

  3. This will launch SendGrid's site. Select Settings > API Keys.

    Create and View API Key App Screenshot

  4. Create an API Key > choose Create & View. (Review SendGrid's documentation on restricted access to determine what level of permissions is appropriate for your API Key. Full Access is selected here for example purposes only.)

    Full access screenshot

  5. Copy the entire key, this value is what you need in your Function App settings as the value for SendGridAPI

    Copy API key screenshot

Next steps