Gong + Momentum forecast augmentation (L4)

The steps

1. 01

   Pipe Gong + Momentum signals into the warehouse

   Tool: Snowflake

   Use Fivetran/Airbyte to land both Gong (call sentiment, topics, talk-ratio, competitor mentions) and Momentum (MEDDPICC field history, deal updates per call) into Snowflake/BigQuery alongside Salesforce opportunity data. The history matters, you want a snapshot table, not a current-state overwrite. Owner: data eng. Pitfall: only landing current state, you can't train a forecast model on "right now" data. DoD: weekly snapshot table of (deal_id, week, all_signals) running for >12 weeks of history.

2. 02

   Build the augmented score in dbt

   Tool: dbt

   Write a dbt model: base score = SFDC stage + amount + close date. Augmentations: Gong sentiment trend (last 3 calls), MEDDPICC completeness %, days since last contact decision-maker, # of distinct contacts engaged, competitor mention frequency. Output: a 0,1 "win probability" per open deal. Validate against the last 4 quarters of closed-won/closed-lost data. Owner: analytics eng. Time: 2 weeks for v1. Pitfall: shipping v1 without holdout backtest, you'll surface a worse forecast than the manager's gut. DoD: model beats baseline (SFDC stage alone) on holdout by >15% on F1 score.

3. 03

   Surface the AI forecast in the rep's SFDC view

   Tool: Salesforce

   Add a custom field on Opportunity: "AI Win Probability" + "Top 2 reasons." Reps see it but can override with a written reason. Track override rates by rep, reps who override a lot become the focus of coaching. Owner: SFDC admin. Pitfall: hiding the AI forecast in a separate tool, reps won't look at it. DoD: every open deal has AI score + reasons visible in the standard opportunity layout.

4. 04

   Run forecast as "AI-first, rep adjusts"

   Tool: Manual

   New Friday ritual: the AI forecast is the starting number. Reps spend 20 min adjusting (not authoring), documenting each adjustment with a reason. Manager reviews adjustments, not the underlying number. Quarter close: compare AI forecast at week 1, week 6, week 13 vs actuals. Owner: VP Sales. DoD: by quarter 2 of running this, AI forecast at week 6 is within 8% of actual; rep-overridden numbers are tracked and graded.

Tools in this playbook

• Snowflake
• dbt
• Salesforce
• Manual

Next playbooks

• Voice-of-Customer → Product & Enablement Loop (L2)
• Zapier Agents for sales ops chores (L2)
• Agentic Nightly CRM Hygiene + Ownership Sync (L3)
• AI-driven price-quote acceleration with CPQ (L3)