How to Build a Deal Scoring System in DenchClaw

Build a deal scoring system in DenchClaw that automatically ranks pipeline opportunities by close likelihood—so you always work the right deals first.

Mark Rachapoom

March 26, 2026·6 min read

Not all deals in your pipeline deserve equal attention. A score that quantifies close likelihood helps you prioritize where to spend time, surfaces at-risk deals before they slip, and removes the guesswork from pipeline reviews. DenchClaw lets you build a custom deal scoring model that runs automatically against your live pipeline data.

What Makes a Good Deal Scoring Model#

The best deal scores are built from signals that actually predict outcomes in your specific business. Generic scoring models borrowed from Salesforce or HubSpot playbooks often miss the signals that matter for your deal types.

Start by analyzing your won deals: what did they have in common? And your lost deals: what patterns predicted a loss? Ask DenchClaw:

"Analyze my closed-won deals from the last 6 months. 
What fields have the strongest correlation with winning? 
Compare to my closed-lost deals from the same period."

The agent will run statistical analysis against your DuckDB data and surface the signals with highest predictive value.

Step 1: Add a Deal Score Field#

"Add a Deal Score field (number) to my deals object. 
Also add a Score Breakdown field (richtext) where the agent can store 
the reasoning behind the score."

Step 2: Define Your Scoring Criteria#

Based on your analysis, define the scoring rules. Here's a starting framework you'll customize:

Positive signals (add points):

Signal	Points
Economic buyer identified	+15
Champion identified and engaged	+15
Budget confirmed	+20
Timeline specified (close date set)	+10
Company size > 200 employees	+10
Deal sourced from referral	+10
Active trial or POC underway	+15
Legal/procurement engaged	+10
Multiple stakeholders engaged	+10

Negative signals (subtract points):

Signal	Points
No activity in 14+ days	-15
Close date in past and not closed	-20
Budget not confirmed	-10
Competitor mentioned as preferred	-15
Key stakeholder is a blocker	-20
No champion identified	-10

Step 3: Implement Automatic Scoring#

"Every night at midnight, recalculate deal scores for all open deals using these rules:

Start with 50 points (baseline).

Add points:
+15 if Economic Buyer field is not empty
+15 if Champion field is not empty and engagement status is 'Engaged'
+20 if Budget Confirmed = Yes
+10 if Close Date is set and in the future
+10 if company size > 200 (look up from companies object)
+15 if Deal Stage = 'Active Trial' or 'POC'

Subtract points:
-15 if last activity date is more than 14 days ago
-20 if close date is in the past and stage is not Closed
-20 if any deal stakeholder has sentiment = 'Blocker'
-10 if Deal Value is not set

Update the Deal Score field and write a brief Score Breakdown explaining the main factors.
Alert me via Telegram if any deal's score drops by 20+ points from yesterday."

This runs as a nightly cron job. Your pipeline scores are always current as of the previous night.

Step 4: Create Score-Based Views#

"Create a view called 'Hot Deals' in my deals object: 
Deal Score >= 80, Stage not Closed, sorted by Deal Score descending."

"Create a view called 'At Risk' in my deals object: 
Deal Score < 40, Stage not Closed, Deal Value > 10000, 
sorted by Deal Score ascending."

The "At Risk" view is where you spend focused attention to understand what's wrong before deals die quietly.

Step 5: Add Score to Your Kanban Cards#

Configure your pipeline kanban to show deal scores on each card:

"Update my deals kanban view to show Deal Score as a badge on each card, 
color-coded: green for 70+, yellow for 40-69, red for below 40."

Now your kanban is not just a stage view—it's a health map. You can see immediately which deals need attention without opening each record.

Step 6: Use Scores for Pipeline Reviews#

Weekly pipeline review becomes much faster with scores:

"Run my weekly pipeline review. For each open deal, show: 
deal name, company, stage, score, score change from last week, 
and the main reason for any score drop. 
Flag any deal scoring below 40."

Instead of going through every deal one by one, you focus the conversation on the deals that moved significantly—up or down.

Step 7: Build Score-Weighted Revenue Forecast#

A score-weighted forecast is more realistic than either the pure expected-close-date forecast or the stage-probability forecast:

"Calculate my score-weighted pipeline forecast. 
For each open deal, multiply Deal Value by (Deal Score / 100). 
Sum by month based on Close Date. Show as a chart."

A $100,000 deal scoring 90 contributes $90,000 to your forecast. A $200,000 deal scoring 30 contributes $60,000—and probably deserves a hard look at what's wrong.

Step 8: Refine Your Model Over Time#

Every quarter, recalibrate your scoring model:

"Analyze deals that closed in Q1. For won deals, 
what was the average score 30 days before close? 60 days before? 
For lost deals, what was the score trend before they went cold? 
Suggest any adjustments to my scoring criteria."

Your model should evolve as you learn which signals actually predict outcomes in your pipeline. The first version won't be perfect—that's fine. The important thing is that you're making data-driven adjustments rather than relying on gut feel alone.