Einstein Conversation Insights
Einstein Conversation Insights is a Sales Cloud feature that ingests recorded sales calls, transcribes them, identifies key moments (mentions of competitors, pricing, objections, action items), and surfaces the highlights in Salesforce on the related Account, Opportunity, or Contact record.
Definition
Einstein Conversation Insights is a Sales Cloud feature that ingests recorded sales calls, transcribes them, identifies key moments (mentions of competitors, pricing, objections, action items), and surfaces the highlights in Salesforce on the related Account, Opportunity, or Contact record. The product is sometimes called ECI in internal docs. It connects to call recording sources (Zoom, Microsoft Teams, Sales Dialer, and several partners), pulls the audio, runs speech-to-text and entity detection, and writes the resulting Insight records linked to the relevant CRM record.
Conversation Insights sits at the intersection of revenue intelligence and AI-assisted coaching. Sales managers use it to spot patterns across the team's calls (which reps handle pricing objections well, which calls mention a competitor we should learn about). Reps use it to find a specific moment in a past call without re-listening to the full hour. The product is competitive with Gong and Chorus and shares much of the same value proposition: turn the hidden data inside sales calls into searchable, reportable, coachable signal.
How Conversation Insights turns sales calls into searchable signal
The recording-to-insight pipeline
Calls arrive from a connected recording source through an integration the admin configures. The platform stores the audio, runs speech-to-text to produce a transcript, runs entity and topic detection over the transcript to flag moments, and writes Insight records (Mentions, Action Items, Topic Trends) linked to the originating call and any related CRM records. The full pipeline runs asynchronously; a one-hour call usually has insights available within minutes of upload completion. The transcript itself is browsable on the Voice Call record.
Mentions: competitor, pricing, product, custom
Mentions are the moments where the AI flagged a topic of interest. Out of the box, the feature identifies competitor names (from a maintained list), pricing references, and product mentions. Admins can configure custom mention categories (specific feature names, regulatory terms, internal initiatives) by uploading keyword and phrase lists. The Mention surfaces with the speaker, the timestamp, and the surrounding snippet, so a manager scanning a call can jump directly to the moment without watching the whole recording.
Action items and the follow-up loop
The model also detects action items from the conversation: phrases like I will send you the pricing or we will get back to you next week. Detected action items surface on the related Opportunity or Account and can be promoted to a Task with one click. This is one of the highest-ROI parts of the feature because it closes the loop between what the rep said on the call and what gets tracked in the CRM, which is the longstanding weak point of sales activity capture.
Voice Call as a record
Conversation Insights uses the Voice Call object as the parent record for everything. Each ingested call becomes a Voice Call record with the transcript, the recording, the speakers, the related Account or Opportunity, and the detected insights. The object is reportable and shareable like any standard object. Reports on Voice Call let managers see call volumes, average duration, and Mention frequency across the team. Sharing rules apply, so reps cannot necessarily see each other's calls unless the rules are set up to allow it.
Integrations with call sources
The integrations that bring recordings in vary by source. Zoom and Microsoft Teams connect via OAuth and pull recordings after the meeting ends. Sales Dialer ingests directly because it lives inside Salesforce. Partner integrations (Aircall, RingCentral, Dialpad, others) connect through their respective platforms. Setting up the integration is usually the longest part of a Conversation Insights deployment because it touches the org's telephony stack, not just Salesforce setup.
Coaching workflows and the manager view
Sales managers get a dedicated view in Conversation Insights that surfaces calls flagged for coaching. The manager can listen at higher speed, add comments at specific timestamps, share with the rep, and track whether the rep reviewed the feedback. This coaching loop is what most teams cite as the highest ongoing value of the product. The AI surfaces calls worth reviewing (long talk time by rep, missed competitor mentions, unclosed objections). The manager spends time on the calls that need attention rather than sampling randomly.
Privacy, consent, and the recording boundary
Recording calls invokes a stack of privacy and consent rules that vary by jurisdiction. Salesforce ships consent prompts in the supported integrations (Zoom auto-disclaimer, Microsoft Teams compliance recording), but the responsibility for legal compliance lives with the customer. Two-party consent states require explicit notice; some jurisdictions require participant opt-in. The product respects whatever the upstream recording captured; if a call was not recorded, no insights exist. Failure to plan consent before turning the product on is the single most common adoption blocker.
How to roll out Einstein Conversation Insights
The bulk of the work in a Conversation Insights rollout is upstream of the Salesforce setup: confirming licensing, choosing recording sources, and addressing consent. The Salesforce-side configuration is straightforward once those are settled.
- Confirm licensing and edition
Conversation Insights requires its own license and is included in some Sales Cloud editions. Confirm before scoping the rollout to avoid mid-project surprises.
- Decide the recording source
Pick the source the team will record from (Zoom, Microsoft Teams, Sales Dialer, partner). The choice constrains which calls flow into the product and what consent prompts already exist.
- Address consent and privacy obligations
Work with Legal to confirm jurisdiction-specific recording obligations. Configure the consent prompts in the recording source. Document the compliance posture before turning recording on.
- Connect the source to Salesforce
Setup, Einstein Conversation Insights, configure the integration. Test with a sample call. Confirm transcript, insights, and the link to the related Account or Opportunity all populate.
- Train managers on the coaching view
The product's value comes from the coaching loop. Run a workshop with sales managers on how to find calls worth reviewing, leave coaching comments, and track rep follow-through.
Zoom, Microsoft Teams, Sales Dialer, and several telephony partners. Each has its own connection setup.
Built-in: competitor, pricing, product. Custom categories supported via keyword and phrase lists.
Built-in. Promotes detected items to Tasks with one click; configurable on the Voice Call page layout.
Manager-facing surface for finding calls worth reviewing and tracking coaching follow-through.
Standard sharing rules on the Voice Call object. Often configured to give managers visibility into their team's calls.
- Consent rules vary by jurisdiction and recording medium. Failure to address consent up front is the most common rollout blocker.
- The integration setup is often the longest step. Telephony plumbing, OAuth approvals, and recording retention policies live outside Salesforce.
- Action item detection is good but not perfect. Some rep promises get missed; some non-commitments get flagged. Treat detected items as drafts, not certainties.
- Competitor mention coverage depends on the maintained competitor list. Keep the list current; new competitors are invisible until added.
- Storage costs grow with recording volume. Long-term retention policies should be set deliberately, not by accident.
Trust & references
Cross-checked against the following references.
- Einstein Conversation InsightsSalesforce Help
- Conversation Insights recording sourcesSalesforce Help
Straight from the source - Salesforce's reference material on Einstein Conversation Insights.
- Einstein Conversation Insights overviewSalesforce Help
- Einstein Conversation Insights setupSalesforce Help
About the Author
Dipojjal Chakrabarti is a B2C Solution Architect with 29 Salesforce certifications and over 13 years in the Salesforce ecosystem. He runs salesforcedictionary.com to help admins, developers, architects, and cert/interview candidates sharpen their fundamentals. More about Dipojjal.
Test your knowledge
Q1. What does Einstein Conversation Insights analyze?
Q2. Why is ECI valuable for sales coaching?
Q3. What kind of moments can ECI surface?
Discussion
Loading discussion…