The new capability evaluates incoming emails before case creation, filtering noise and reducing unnecessary SLA pressure
Microsoft has announced that email classification is now generally available in Dynamics 365 Customer Service. The feature uses AI to evaluate every incoming email before a case is created. It assigns each message to a category that administrators define, such as Billing, Technical Support, or Marketing.
This addresses a long-standing pain point for service teams. Until now, the system created a case for every email sent to a monitored mailbox. Each case started an SLA clock and pulled in a service representative. Marketing blasts, notifications, and other non-actionable messages consumed the same resources as real customer issues. Email classification changes that by adding an intelligent decision point at the front of the intake process.
How email classification works
Email classification analyzes the content of each incoming message as it arrives. The AI then assigns the email to one of the categories the organization has configured. That category becomes an attribute on the email record. Administrators can use it to control what happens next.
The configuration model is straightforward. Administrators define between two and nine custom categories. Each category needs a name and a description, which help the AI classify messages accurately. Scope conditions then control which emails get evaluated. For example, classification can apply only to specific mailboxes or departments. This matters because the feature consumes AI resources, so organizations can limit evaluation to where it adds value.
One design detail deserves attention. The system always assigns one of the configured categories, even when an email doesn’t clearly match any of them. If an organization sets up only Billing and Marketing, every email lands in one of those two buckets. Microsoft recommends adding a catch-all category, such as Uncategorized, to handle ambiguous messages. Depending on configuration, service representatives can also update a category manually.
Pairing classification with automatic record creation
The real value emerges when classification feeds into Automatic Record Creation (ARC) rules. Administrators can use the assigned category as a condition in ARC. As a result, cases are created only for emails that genuinely need one. Categories like Marketing or Notifications can be excluded from case creation entirely.
This prevents non-actionable emails from starting SLA timers or landing in representative queues. Service teams see cleaner queues and spend less time closing cases that never should have existed. Beyond ARC, categories can also drive routing rules and other case creation logic. An email tagged as Billing, for instance, can route directly to the billing queue.
Setup and prerequisites
Configuration happens in the Customer Service admin center under email settings. The process involves four steps: define categories, set scope conditions, enable classification, and connect categories to downstream processes like ARC or routing.
Two prerequisites apply. The environment must have pay-as-you-go billing enabled, which means classification carries consumption-based costs. Additionally, setup requires the system administrator or CSR manager role.
Before going live, administrators can use simulation mode to test classification against historical emails. This allows teams to validate categories and refine descriptions before the feature affects live case creation. Starting in simulation mode is a sensible first step for any organization adopting the feature.
Part of a broader AI push in Customer Service
Email classification arrives amid a wave of AI investments in the email channel. Recently, Microsoft launched a Governance feature for Dynamics 365 Customer Service in public preview, which checks outbound emails against admin-defined policies. Microsoft also extended its Quality Evaluation Agent to cover email interactions in public preview. Together, these features show Microsoft building AI controls around both inbound and outbound email in the service workflow.
For customers, the direction is clear. Microsoft wants AI handling the repetitive triage and compliance work so representatives can focus on resolving genuine issues. However, organizations should note the pay-as-you-go requirement across many of these features. AI capabilities in Customer Service increasingly carry consumption costs on top of user licensing.
Conclusion: what this means for service organizations
Email classification offers a practical fix for a common operational problem. Teams drowning in noisy shared mailboxes can now filter irrelevant messages before they trigger cases, SLAs, and assignments. The full details are available in Microsoft’s announcement on the Dynamics 365 blog.
As a next step, administrators should review their incoming email patterns and define categories that reflect them. Enabling simulation mode against historical emails will reveal how well those categories perform before anything goes live. Organizations that want help configuring classification, ARC rules, and routing can work with a Dynamics 365 Customer Service partner to design an intake process that fits their service model.
