You receive dozens of customer emails each day, and quickly gauging the writer’s tone is critical for prioritizing replies. Outlook’s built-in sentiment detection helps you identify positive, negative, or neutral language without reading every line. This feature uses Copilot to analyze the email body and highlight emotional cues such as frustration, satisfaction, or urgency. In this article, you will learn how to enable sentiment analysis in Outlook, run the feature on a customer email, and interpret the results to respond more effectively.
Key Takeaways: Detecting Sentiment in Customer Email with Copilot
- Copilot pane > Summarize > Sentiment: Analyzes the email body and returns a sentiment label: positive, negative, neutral, or mixed.
- Copilot pane > Ask Copilot > “What is the tone of this email?”: Returns a detailed breakdown of emotional cues and suggested response strategies.
- Microsoft 365 admin center > Copilot > Data sources: Must include Exchange Online mailboxes for Copilot to access message content for analysis.
How Copilot Sentiment Detection Works in Outlook
Copilot sentiment detection uses a large language model trained on conversational text. When you open a customer email, Copilot scans the body for keywords, sentence structure, punctuation, and capitalization patterns that indicate emotion. Words like “frustrated,” “disappointed,” or “unacceptable” signal negative sentiment. Phrases such as “thank you,” “great service,” or “pleased” indicate positive sentiment. The model also detects mixed sentiment when the email contains both praise and complaint.
The feature requires an active Microsoft 365 subscription with Copilot for Microsoft 365. Your organization must have Copilot enabled in the Microsoft 365 admin center under Settings > Copilot > Data sources. Exchange Online must be toggled on so Copilot can read message content. No additional add-ins or plugins are needed. The sentiment analysis runs locally within the Outlook client and respects existing data governance policies.
Sentiment detection is not available in Outlook on the web or Outlook for macOS in some regions. It works in the Outlook desktop client for Windows and the new Outlook for Windows preview. Microsoft plans to expand availability to mobile and web versions in future updates.
Steps to Detect Sentiment in a Customer Email
Follow these steps to analyze a single customer email in Outlook. You must have the email open in its own window or in the reading pane.
- Open the customer email
Double-click the email in your inbox to open it in a separate window. Alternatively, select the email so it appears in the reading pane on the right side of the Outlook window. - Open the Copilot pane
Click the Copilot icon in the top-right corner of the Outlook ribbon. The icon looks like a purple diamond. If you do not see it, go to View > Copilot to enable the pane. - Select the Summarize option
In the Copilot pane, click the Summarize button. Copilot generates a brief summary of the email content. Below the summary, you will see a Sentiment label: Positive, Negative, Neutral, or Mixed. The label appears in bold text with a small color indicator: green for positive, red for negative, gray for neutral, and orange for mixed. - Ask Copilot for a detailed tone analysis
If you need more than a label, type the following prompt in the Copilot chat box at the bottom of the pane: “What is the tone of this email?” Press Enter. Copilot returns a paragraph describing the emotional cues it found, such as “The customer uses all-caps in two sentences, which indicates frustration” or “The closing phrase ‘Looking forward to your reply’ suggests a cooperative tone.” - Use the sentiment to draft a reply
After you understand the sentiment, ask Copilot to draft a response. Type: “Draft a reply that addresses the customer’s frustration and offers a solution.” Copilot generates a draft that matches the detected sentiment. You can edit the draft before sending.
If Copilot Does Not Show a Sentiment Label
Several issues can prevent the sentiment label from appearing. Check each item below.
Copilot pane is empty or shows “No summary available”
This usually happens when the email is too short or contains only images or attachments. Copilot needs at least 20 words of text to produce a sentiment analysis. If the email is very short, manually type a prompt such as “What is the tone of this email?” to force analysis.
Sentiment label shows “Neutral” for an emotional email
Copilot may label an email as Neutral when the language is polite but the content is negative. For example, a customer writing “I am not entirely satisfied with the service” uses indirect language. In this case, use the detailed prompt method. Ask Copilot: “Does this email contain any negative sentiment?” Copilot will identify the phrase “not entirely satisfied” as a negative cue.
Copilot does not respond at all
Verify your Microsoft 365 license includes Copilot for Microsoft 365. Go to File > Account > Subscriptions and confirm Copilot appears in your list. If it does not, contact your IT administrator. Also check that your Outlook client is updated to version 2402 or newer. Go to File > Office Account > Update Options > Update Now.
Copilot Sentiment Detection vs Manual Reading: Key Differences
| Item | Copilot Sentiment Detection | Manual Reading |
|---|---|---|
| Speed | Instant analysis in seconds | Requires 30 seconds to 2 minutes per email |
| Consistency | Applies the same model to every email | Varies based on reader fatigue and bias |
| Detail | Returns a label or a paragraph of cues | Depends on reader’s attention to nuance |
| Language support | English only at launch | Any language the reader knows |
| Privacy | Data stays in Microsoft 365 compliance boundary | No automated data processing |
Copilot sentiment detection is not a replacement for human judgment. It provides a first pass to help you prioritize emails. For high-stakes customer interactions, always read the full email yourself before deciding on a response.
You can now use Copilot in Outlook to quickly detect customer sentiment and draft appropriate replies. Start by opening a recent customer email and running the Summarize feature to see the sentiment label. For deeper analysis, use the Ask Copilot prompt with specific questions about tone. As a next step, try combining sentiment detection with Copilot’s priority inbox rules to automatically flag negative emails for faster response.