Interpret agent actions on leads

Completed

When the Sales Qualification Agent hands a lead to a seller, that seller needs to trust what they're looking at. They didn't do this research themselves; an AI agent did. Before they pick up the phone or send an email, they need to understand what the agent found, why it rated the lead the way it did, and what it's recommending they do next. This unit walks through the Lead research page, explains what sellers see in Research-only and Research and engage modes, and covers how to qualify or reassign those leads.

What the agent does to a lead record

After the Sales Qualification Agent processes a lead, the lead record in Dynamics 365 looks different from a manually researched lead. The agent adds a Lead research page to the lead record: a structured, comprehensive view of everything the agent discovered and decided.

This page includes:

• Lead rating: The agent calculates a Hot, Warm, or Cold rating based on Target Customer Profile (TCP) fit, engagement signals, and the quality of data available. The rating is explained with specific reasons so sellers understand how it was determined.
• Suggested action: A clear recommended next action (for example, "Draft email" in Research-only mode).
• Deeper insights: Structured cards covering company overview, financial health, recent news, competitor analysis, and stakeholder intelligence. If an administrator configured custom research topics (for example, pulling in Gong conversation data or other organization-specific intelligence), another custom insights card appears here alongside the standard cards.

The Key insights section is where the agent shows its work. Alongside the lead rating summary, the agent surfaces two transparency indicators. Signal strength tells the seller how complete and reliable the underlying data was. If the agent had limited information to work with, it flags that so sellers can factor it into their judgment. Steps interpretation shows the chain-of-thought reasoning the agent used to reach its conclusions, so sellers can see exactly why a lead was rated the way it was rather than just accepting a score at face value. This transparency is especially useful when a seller disagrees with the agent's assessment—rather than overriding it blindly, they can see where the reasoning diverged from their own.

Sellers don't need to do separate research; the relevant context is surfaced in one place on the lead record.

Research-only mode: the seller's workflow

In Research-only mode, the agent does the research and leaves the outreach decision to the seller.

When a seller receives a notification that a lead was handed over, they find the following on the lead record:

1. Research insights: Detailed cards showing company background, financial health, recent news, and competitor insights.
2. TCP evaluation: A clear assessment of how well the lead fits the target customer profile, with the specific reasons behind the rating.
3. Draft outreach email: An AI-generated initial email tailored to the lead, incorporating the research insights and your company's value proposition.

Let's take a look at the seller's workflow in Research-only mode:

1. Open the lead record and review the Lead research page.
2. Read the research insights to understand the lead's context and potential needs.
3. Review the draft outreach email. Edit it if needed to add a personal touch or adjust the tone.
4. Send the email directly from the lead record.
5. Make the final qualify or disqualify decision using the Qualify command on the command bar.

The agent did the heavy lifting—the seller applies judgment and takes action.

Research and engage mode: the seller's workflow

In Research and engage mode, the agent does far more before handing a lead to a seller. By the time a seller receives the lead, the agent has:

• Researched the company and contact
• Sent an initial outreach email
• Followed up based on the lead's responses
• Evaluated BANT criteria based on the email conversation
• Determined that the lead demonstrates positive purchase intent

The lead record the seller receives includes:

• Research insights: The same comprehensive company and contact research as Research-only mode.
• Full email thread: The complete conversation between the agent and the lead, so the seller can see exactly what was discussed.
• BANT evaluation: A structured assessment of the lead's Budget, Authority, Need, and Timeline based on their email responses.
• TCP and intent evaluation: A summary of why the agent handed over the lead, including the specific criteria that were met.
• Suggested next action: A recommended action (usually "Qualify") with context for why the agent determined the lead is ready for a seller conversation.

In Research and engage mode, the seller's role is to pick up the relationship where the agent left off and advance the lead to an opportunity. The qualification work is largely complete. The seller validates the agent's assessment, introduces themselves personally, and begins the opportunity conversation.

Tip

Before your first call with a handed-over lead, review the What happened to this lead? section in the Key insights area. This interaction summary shows the full engagement history between the agent and the lead, including any questions the lead raised during the email exchange.

Qualifying a handed-over lead

Regardless of mode, sellers qualify agent-processed leads the same way as any other lead in Dynamics 365 Sales:

1. Open the lead record.
2. Review the Lead research page and the agent's recommended action.
3. Select Qualify on the command bar.
4. Dynamics 365 creates the associated opportunity, contact, and account records.

The lead record remains linked to the opportunity for historical context, including all research insights and—in Research and engage mode—the full email thread. This gives sellers a complete picture of the lead's journey from first touch through qualification.

Working with disqualified leads

Leads that the agent disqualifies aren't automatically deleted or archived—they remain in the system and are visible in the Leads disqualified by AI agent view in the Leads section of Sales Hub.

Supervisors who review disqualified leads can select Reassign if they believe the agent made an incorrect assessment. The lead is then assigned to a seller based on the configured assignment rules.

This feedback loop is important for calibrating the agent over time. If you regularly see high-quality leads being disqualified, it's a signal that your TCP criteria or handoff thresholds need adjustment. The disqualification reason shown on each lead record helps you identify which specific criteria are causing the issue.