Having recently completed an integration of Read AI meeting data into our revenue operations stack, I found the native reporting to be somewhat limited for advanced pipeline analysis. To address this, I constructed a custom dashboard in Google Data Studio (now Looker Studio) that correlates meeting intelligence with Salesforce opportunity data. The goal was to move beyond simple engagement metrics and build a model for forecasting the impact of meeting quality on deal velocity and win rates.
The core of the configuration hinges on a blended data source. The primary components are:
* **Read AI Data Export:** Accessed via the daily CSV export to a dedicated Google Cloud Storage bucket. Key fields include meeting participant attendance duration, conversational analytics scores (e.g., talk ratio balance), and the generated "Key Points" and "Action Items" which I parse for specific keywords related to our sales stages.
* **Salesforce Opportunity Object:** Extracted via the native Data Studio connector, focusing on fields for Stage, Amount, Close Date, and a custom field for the associated Meeting Record ID.
* **A Unifying Key:** A calculated field in Data Studio that matches the Read AI meeting ID to the Salesforce field. This allows for a true join at the opportunity level, not just the contact or account level.
The dashboard itself is structured across three interconnected pages:
1. **Meeting Health & Engagement:** This page visualizes the relationship between conversational analytics and subsequent deal progression. A scatter plot compares a composite "Meeting Engagement Score" (derived from attendance duration and participant talk balance) against the number of days an opportunity remains in a stage post-meeting. The initial data suggests a moderate inverse correlation—higher engagement scores correlate with shorter stage duration, particularly in the discovery and evaluation phases.
2. **Action Item Fulfillment Tracking:** This is a more qualitative analysis. Using REGEXP_EXTRACT functions on the Read AI "Action Items" text, I categorize promised next steps (e.g., "send proposal," "schedule demo with engineering," "provide case study"). A time-series chart then tracks the percentage of these action items logged as completed in Salesforce Tasks within a 3-day window against the eventual win rate for those deals. This has been illuminating for coaching reps on the tangible impact of post-meeting follow-through.
3. **Forecasting Integration:** The most complex component is a table that blends the meeting-derived signals with our standard Salesforce forecast. It applies a weighted adjustment factor to the Salesforce probability based on the meeting engagement score and action item completion rate for the most recent customer-facing meeting. This creates a parallel "Adjusted Forecast" view that our sales leadership is beginning to use as a secondary lens.
The main challenges were data hygiene—ensuring consistent meeting naming conventions so the joins work reliably—and the latency of the batch export. Real-time analysis isn't possible with this configuration, but for weekly forecast and pipeline reviews, it provides a significantly deeper layer of insight than we had previously. I'm interested to hear if others have attempted similar models with Read AI or other conversation intelligence platforms, and what dimensions you found most predictive.
--JK
measure what matters