How to calculate Top 3 factors affecting Pipeline and booking conversion?

RevSure uses Shapley value-based factors to pinpoint which factors attributes most significantly affect conversion probabilities. This guide outlines what Shapley values are, how they're calculated in a machine learning context, and how they relate to the insights you see in the product.

What Are Shapley Values?

Shapley values originate from cooperative game theory. In a game, the Shapley value tells us how much each player contributes to the total outcome, assuming all possible combinations of players are considered.

In RevSure, each "player" is a contributing factor (e.g., funnel lead source, industry, MQL age), and the "game" is the model's prediction for a lead or opportunity’s conversion.

The Shapley value for a feature answers:

“How much did this feature contribute to the predicted conversion probability compared to the average?”

How RevSure Applies Shapley Values

RevSure builds a predictive model  to forecast:

• Pipeline conversion probability

• Booking conversion probability

To explain a prediction for a given record (lead/opportunity), we use  Shapley values for ML models.

Steps:

1. Baseline Prediction: Average model prediction across the dataset (e.g., base conversion rate is 3%).

2. Individual Prediction: The model predicts, say, 12.7% for a specific lead.

3. Shapley Values Computation: For each feature:

   • Consider all permutations of features.

   • Measure how much the prediction changes when the feature is added to the subset.

   • Average that contribution across all permutations.

4. Result: The individual prediction is decomposed into a base value + sum of Shapley values:

   12.7% = 3.0% (base) + 6.0% (funnel source) + 2.0% (industry) + 1.7% (MQL duration) ...
   

Interpreting the Screenshots (Examples)

Let’s walk through the screenshots to understand the real-world application of Shapley values in RevSure.

Screenshot 1: Top 3 Factors Driving Pipeline Conversion

![Screenshot 1]

• Funnel Lead Source: +30,967% relative impact on conversion

• Industry = Consumer Services: +1018%

• MQL Duration (216–336 days): +412%

What this means: For leads converting in Q2 2025, these three features had the highest aggregate positive Shapley values across the cohort—i.e., they consistently lifted conversion probability above the average model baseline.

Screenshot 2: Individual SHAP Breakdown for Pipeline Conversion

![Screenshot 2]

• Lead’s conversion probability = 12.7%

• Funnel Lead Source contributed the majority of the score (large positive bar)

• Other features like industry and MQL duration also contributed positively

• Some features (e.g., low activities before MQL) had slight negative SHAP values (pink bars)

This breakdown is the Shapley value explanation for a single record. It allows users to:

• Understand why the score is 12.7% (vs. 3% baseline)

• Diagnose which features had the most influence

• Drive targeted actions (e.g., increase campaign touches)

Screenshot 3: SHAP Explanation for Booking Conversion

Similar view but for opportunity-to-booking predictions:

• Funnel Source, Industry, Campaign Touches, and MQL duration again dominate

• These features are consistent indicators of late-stage progression and deal closure

Aggregate vs. Individual Explanations

Example (Simplified)

Let’s say the model sees:

• Average conversion rate = 3%

• A lead has:

  • Source = Sales Prospecting (+5%)

  • Industry = Consumer Services (+3%)

  • MQL Duration = [216–336] days (+1.7%)

The model computes:

Base prediction: 3.0%
+ Source SHAP: +5.0%
+ Industry SHAP: +3.0%
+ MQL Duration SHAP: +1.7%
= Final predicted probability: 12.7%

Why Use Shapley-Based Insights in RevSure?

• Transparency: Know why the score is what it is.

• Targeting: Focus outreach on leads with favorable signals.

• Prioritization: Double down on high-lift campaigns and sources.

Summary