How Deal AI Insights Work

This article explains how the AI-generated insights on a deal record are calculated.

Deal Score

What it measures: Overall deal strength on a 0–100 scale, where 100 indicates excellent health and high probability of close, and 0 indicates high risk.

How it's filled:

The score is calculated using a weighted, multi-factor model that combines three factors:

• Engagement Quality & Momentum — reflects the quality of interaction and momentum, based on engagement with the best ICP contact or decision maker, average conversation sentiment, question-to-talk ratio, and days since last activity
• Customer Fit & Readiness — reflects the viability of the opportunity, based on the company's ICP fit score, fit scores of associated contacts, deal value, budget confirmed status, articulated pain points, and buying intent
• Process Adherence & Velocity — reflects rep discipline, based on timely follow-ups after interactions, pending emails or documents, and time in stage

The weighting of each factor changes based on the deal's current stage — for example, Customer Fit matters more during Qualification, while Process Adherence weighs more during Negotiation.

A risk penalty is applied so that one critically low factor cannot be masked by high scores in the others.

Behavior:

• The score recalculates every 24 hours and immediately on major events (call logged, stage change, key task completion)
• A trend indicator shows the change in score over the last 24 hours
• For any score change of 5 points or more, an AI-generated justification explains what drove the change (e.g., "Score dropped due to lack of Decision Maker involvement in the last 7 days")
• A newly created deal with no activity starts in the 40–55 range, pulled down by low engagement and process signals to prompt the rep to act

Deal Health

What it measures: How the deal is progressing relative to comparable successful deals.

How it's filled:

• Deal velocity — average days in the current stage compared against the average for successful deals, displayed as "Above average," "Below avg," or "On track"
• Engagement Score (Max.) — the highest engagement score among all contacts associated with the deal
• Strongest connection — the contact with the highest engagement score, shown with their name and engagement percentage (e.g., "Alex Albon — Engagement 53%")

Buying Committee Analysis

What it measures: The composition of the buying committee across the deal.

How it's filled:

Each person associated with the deal is classified along three dimensions based on their designation, title, and meeting talk summary.

Buyer type:

• Finance — controls budget, approves final decisions
• Technical — reviews technical fit, integration, and compliance
• Legal — responsible for security and legal compliance
• Decision Maker — person with the highest designation or hierarchy, or most engaged
• Influencer — provides recommendations, may not have final say
• Gatekeeper — controls access to decision-makers (e.g., EA, admin)

Decision-making power: High, Mid, Low, or NA — based on role.

Focus areas:

• Legal / Compliance — contracts, approvals, regulatory compliance
• Technical / IT — integrations, architecture, security, scalability
• Procurement / Finance — budget, cost justification, payment terms
• Operations / Process — efficiency, adoption, workflow impact
• Marketing / Sales Enablement — campaigns, analytics, enablement tools
• Product / Innovation — feature requirements, roadmap influence, innovation adoption
• Executive / Strategic — business strategy alignment, ROI, long-term planning

Competitor Mentions

What it measures: References to competitors detected in communications on the deal.

How it's filled:

The AI scans transcripts, emails, and calls for competitor names. Each mention is displayed with:

• Which competitor was mentioned
• When it was said and in what mode (email, meeting, or call)

Similar Deals

What it measures: Closed-won deals in the database that resemble this deal, useful for applying learnings from past outcomes.

How it's filled:

Displayed as "2 similar" or "NA" (number of matches found), each similar deal is shown alongside the reason for the match.

The AI matches deals based on:

• Industry
• Company size
• Customer problems — extracted from transcripts
• Pattern of deal progression — whether the deal was stuck in the same stage for a similar number of days, and how it eventually moved to closed-won (e.g., manager intervened, discount offered)

Each match includes recommendations for actions to take based on how the similar deal was progressed.