Monitoring and observability for Dynamics 365

A Dynamics 365 environment without monitoring is opaque — slow pages, failed integrations, and broken flows go unnoticed until users complain. Modern Dynamics 365 emits substantial telemetry; collecting and acting on it is one of the highest-leverage operations investments a customer can make.

The telemetry surfaces.

• Application Insights — the primary telemetry destination for Business Central, Power Platform, F&O, custom plug-ins, and custom code. Configurable per environment.
• Microsoft 365 Service Health — Microsoft's own service-status feed showing outages and degradations across M365 / Dynamics 365 services.
• Power Platform admin centre analytics — Dataverse capacity, API call usage, flow runs, app usage.
• Azure Monitor — for Azure resources behind Dynamics 365 (Service Bus queues, Functions, Logic Apps in the integration stack).
• Microsoft Purview — audit and compliance retention.

Each captures different signals; a complete observability programme aggregates from all of them.

Application Insights setup.

For Business Central: configure the Application Insights connection string on the tenant in the BC admin centre. All BC telemetry (page loads, API calls, errors, slow queries, lock timeouts, environment lifecycle events, extension lifecycle events) flows to the configured workspace.

For Power Platform / Dataverse: Application Insights integration captures Dataverse operations, plug-in execution, Power Automate flow execution, and Power Apps telemetry. Configure per environment.

For F&O: telemetry flows through LCS and increasingly through dedicated Application Insights workspaces.

What to monitor.

• Performance — page load times, API response times, slow queries, slow flow runs.
• Errors — exception rates, error events, plug-in failures, flow failures.
• Capacity — Dataverse storage trends, API call volumes, AI Builder credits.
• Integrations — webhook delivery success, Service Bus queue depths, dead-letter queue arrivals.
• Security — sign-in failures, conditional-access denials, privileged-role usage.
• Custom telemetry — your own AL / X++ / plug-in code emitting business-meaningful events.

KQL — the query language

Application Insights uses Kusto Query Language (KQL) for querying. Examples:

// Slow page loads in the last 24h
customEvents
| where timestamp > ago(24h)
| where name == "PageLoaded"
| where customMeasurements["duration"] > 5000
| summarize count() by tostring(customDimensions.pageName)
| order by count_ desc

Microsoft publishes the BCTech GitHub repo with canned queries for common Business Central questions; equivalent libraries exist for other Dynamics 365 products.

Workbooks and dashboards

Application Insights workbooks turn KQL queries into shareable dashboards. Microsoft publishes pre-built workbooks for each Dynamics 365 product:

• Business Central health workbook — performance, errors, telemetry insights.
• Power Platform admin workbook — environment health, flow performance.
• F&O telemetry workbook — service-update events, batch performance.

Customers customise these for their specific needs.

Alerts

Application Insights alerts fire on configurable thresholds:

• Error rate above N% in 5-minute window.
• API call volume approaching tenant quota.
• Specific exception strings.
• Dataverse storage approaching cap.
• Failed flows over a threshold.

Alerts route to Action Groups — email, Teams, SMS, webhook to PagerDuty / ServiceNow. Configure them before you need them.

Service Health alerts

Subscribe to Microsoft 365 Service Health notifications for the customer's services. Tenant admins get alerted when Microsoft reports an issue.

The operational dashboard

A useful tenant has at least one daily-glance dashboard covering:

• Total errors and trend.
• Slow operations.
• Flow / integration failures.
• Capacity headroom.
• Service Health open incidents.

10 minutes of dashboard review per morning catches most issues before users complain.

Operational reality

Monitoring isn't an after-go-live add-on. Configure telemetry on day one of any environment provisioning; build dashboards before users start; iterate based on what surprises you.