Prioritize leads with predictive scoring
Contoso's sales team manages dozens of open leads at any time. Reps invest hours on leads that don't convert while missing others with strong buying intent. Predictive lead scoring uses AI to assign each lead a score from 1 to 100, giving reps an immediate, objective signal about which leads to prioritize. Here, you learn how to configure and use predictive lead scoring to help sellers focus on the leads most likely to convert.
| Score Grade | Range | Action |
|---|---|---|
| A | 75–100 | High priority—contact immediately |
| B | 50–74 | Strong potential—schedule follow-up soon |
| C | 25–49 | Monitor—review periodically or automate nurturing |
| D | 1–24 | Low priority—consider disqualifying or routing to automated nurturing |
How predictive lead scoring works
Predictive lead scoring is an AI-powered feature in Dynamics 365 Sales that trains on your historical lead data to predict conversion likelihood. The system analyzes patterns from leads that were qualified versus those that were disqualified, then applies that learning to score your current leads.
The model evaluates lead attributes such as lead source, industry, company size, email engagement, and website activity signals. It identifies which fields most strongly predict conversion for your specific business—these appear as "top influencing factors" on each lead record. Unlike static rules-based scoring, the AI model adapts to your unique sales patterns and continuously learns as new data arrives.
Each lead receives a numerical score from 1 to 100 and a corresponding grade (A, B, C, or D) for quick visual reference. The score refreshes automatically as lead behavior changes—when a lead opens emails, visits pricing pages, or updates their profile, the score adjusts to reflect the new information.
Configure the scoring model
Predictive lead scoring is available with a Dynamics 365 Sales Enterprise license, which includes up to 1,500 scored records per month. Organizations that need to score more than 1,500 records per month can upgrade to Sales Premium for unlimited scoring. Before you begin, ensure your organization has at least 40 qualified leads and 40 disqualified leads from the past two years. The model needs this historical data to identify meaningful patterns.
To configure predictive lead scoring, go to App Settings > Lead + opportunity scoring > Quick setup, then select Create and publish in the Predictive lead scores section. The system trains the initial model on your historical lead data, which takes approximately two hours.
After training completes, review the model accuracy score and the top influencing factors the AI identified. The accuracy score indicates how well the model predicts outcomes based on your data—a higher score means more reliable predictions. The top influencing factors show which fields have the strongest correlation with lead conversion in your organization.
Publish the model to make scores available on lead records. Sellers now see the predictive score widget on each lead, showing the score, grade, and the specific factors influencing that lead's score.
Tip
Add custom fields to the model to improve accuracy. If your sales process relies on specific data points—such as engagement level or product interest—include those fields when configuring the model.
Organizations selling different product lines with distinct qualification patterns can create multiple scoring models. For example, Contoso might create one model for enterprise sales leads and another for small business leads, each trained on the relevant historical data for that segment.
Interpret and use predictive scores
The predictive score appears on the lead record and in lead views and grids, giving sellers visibility across their entire pipeline. Each score displays the numerical value, the grade, and lists of positive and negative influencing factors.
Positive factors show what's driving the score up—for example, the lead works in a target industry or has engaged with multiple marketing emails. Negative factors reveal what's pulling the score down, such as a lead source that historically converts poorly or a company size outside your typical customer profile.
Sellers use the grade to triage their queue efficiently. Grade A leads demand immediate attention—these represent the highest conversion probability and should be the first contacts each day. Grade B leads show strong potential and warrant scheduled follow-up soon. Grade C leads benefit from periodic review or automated nurturing campaigns. Grade D leads often indicate poor fit and may be candidates for disqualification or reassignment to marketing for longer-term nurturing.
The score is informational, not prescriptive. Sellers retain full judgment and can apply their experience alongside the AI recommendation. A rep who has built rapport with a Grade C lead can continue that relationship. Conversely, contextual factors might lead a seller to deprioritize a Grade A lead that doesn't align with current territory focus.
As leads progress through the pipeline, their scores update automatically. A lead that initially scored low might jump to Grade A after attending a webinar and downloading a white paper. This dynamic scoring keeps sellers focused on leads with current buying signals rather than static data.
Note
Predictive lead scoring requires sufficient historical data for accurate predictions. Organizations with limited lead history should focus on capturing quality data as they build toward the minimum thresholds.
Predictive scoring provides the initial triage signal that helps sellers decide which leads warrant immediate personal attention. In the next unit, you explore how Copilot in Dynamics 365 Sales generates email content and meeting summaries to accelerate outreach to those prioritized leads.