Forecast augmentation with Gong + ML (L4)

The steps

1. 01

   Enable Gong AI Forecasting

   Start by activating Gong’s AI-driven forecasting capabilities, which analyze deal health based on interaction signals (email velocity, participant seniority, and sentiment). In Gong, navigate to 'Settings' > 'Forecast' > 'Models.' Select the 'AI Forecast' or 'Deal Likelihood' model. Ensure your CRM integration (Salesforce or HubSpot) is fully synced so Gong can pull historical outcome data to train its local ML instance. • Role: Gong Admin or RevOps • Time Estimate: 1 hour (plus 24-48 hours for data processing) • Setup: Go to 'Forecast' > 'Settings' > 'Board Configuration' and ensure the 'Gong Forecast' column is toggled on. • Pitfall: If your reps don't use Gong for all calls or if email integration is broken, the AI will be blind and provide a 'low confidence' score. • Definition of Done: You can see a dollar-value forecast generated by Gong next to your CRM pipeline stages.

2. 02

   Build Reality-Check Dashboards

   Extract both the human-submitted forecast and the Gong AI forecast into a centralized environment like Snowflake or BigQuery, then visualize them in a BI tool (Tableau or Looker). You need a SQL table that joins your CRM Opportunity ID with Gong’s 'Forecast' API output. If using Tableau, create a calculated field named 'Forecast Variance %' using the formula: ABS([Gong Forecast] - [Rep Forecast]) / [Rep Forecast]. • Role: Data Analyst or RevOps • Time Estimate: 4-6 hours • SQL Example: SELECT opp_id, rep_submission, gong_ai_prediction, (gong_ai_prediction - rep_submission) AS gap FROM forecast_data. • Pitfall: Ensure you are comparing 'apples to apples',both sources must be looking at the same time period (e.g., 'Current Month') and the same deal categories (e.g., 'Commit'). • Definition of Done: A dashboard showing a side-by-side bar chart of Rep vs. AI forecast, sortable by Manager.

3. 03

   Establish Coaching Triggers

   Configure an automated alert to notify Sales Managers when the discrepancy between a rep's manual forecast and the Gong AI prediction exceeds 20%. Use a tool like Zapier or Slack Workflow Builder to monitor your BI tool’s data feed. When 'Variance %' > 0.20, send a Slack message to the Manager: 'Alert: High Forecast Variance on [Rep Name]’s rollup. Rep says $100k, Gong says $75k. Review requested.' • Role: RevOps or Sales Ops • Time Estimate: 2 hours • Prompt/Config: Set the trigger frequency to weekly on Monday mornings before 1:1 sessions. • Pitfall: Avoid alerting on small deals; set a minimum threshold (e.g., only alert if the deal value is >$10k) to prevent notification fatigue. • Definition of Done: Managers receive a summarized Slack or Email report highlighting specific reps with 'at-risk' accuracy scores.

4. 04

   Execute Gap-based Coaching

   Embed the 'Gong vs. Rep' gap into the weekly 1:1 cadence. Managers should not simply tell reps they are wrong; they should use Gong’s 'Deal Boards' to inspect why the AI is pessimistic. Click into the Gong Deal Board, look at 'Interaction Signals' (e.g., 'No response in 10 days' or 'No power user involved'), and ask the rep to explain the gap. This moves the conversation from 'gut feel' to 'evidence-based' coaching. • Role: Sales Manager • Time Estimate: 30 minutes per rep/week • Process: Open the BI dashboard, identify the gap, then open the specific deal in Gong to see the 'Risk factors' identified by the ML. • Pitfall: Using the AI as a 'policing' tool rather than a coaching tool. If reps feel punished by the AI, they will stop entering honest CRM data. • Definition of Done: 1:1 meeting notes reflect specific action items based on AI-identified deal risks.

5. 05

   Audit and Adjust Weights

   At the end of each quarter, perform a 'Winner Analysis' to see which signal,the Rep or the AI,was closer to the actual closed-won revenue. Calculate the 'Mean Absolute Percentage Error' (MAPE) for both. Use a spreadsheet to list: [Rep Name], [Final Submission], [Gong Prediction], [Actual Result]. Use this data to adjust your weighting for next quarter's 'Executive Forecast.' • Role: RevOps Leader • Time Estimate: 3-5 hours • Analysis: If the Rep is consistently 95% accurate on 'Commits' but the AI is 80%, trust the Rep. If the AI is consistently more accurate on 'Best Case' deals, use the AI number for the board report. • Pitfall: Ignoring the 'Human Element',sometimes reps have offline info (a verbal 'yes' at a golf game) that the AI cannot see. • Definition of Done: A QBR slide showing 'Forecast Reliability' trends and a decision on which forecast source to prioritize for next quarter.

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)