Key Concepts

Compass Score

A universal quality score from 1 to 5 assigned to every customer interaction. The score reflects how well the interaction achieved its objective, whether that's resolving a support issue, converting a lead, or providing information.

Signals

Business-meaningful insights detected in each interaction. Signals surface things like churn risk, customer dissatisfaction, successful objection handling, and follow-up requirements. Each signal includes a confidence level and strength indicator.

Coaching Summaries

Automated narrative summaries that highlight what went well and what could improve, with specific quotes from the conversation. Managers can use these directly in one-on-ones.

Interactions

Any customer conversation — phone call, chat, email, or SMS — that Compass evaluates. Each interaction gets a Compass Score, signals, and coaching insights.

Projects

Organizational units in Compass. A project typically represents a team, department, or client. Each project can have its own scoring rubrics, agent roster, and configuration.

MCP (Model Context Protocol)

Compass exposes its data as an MCP server, letting AI-powered tools query your interactions, signals, and coaching insights directly. If you use Claude Desktop, Cursor, or any MCP-aware client, you can search calls, compare evaluations, and explore performance data through natural conversation. See the MCP Server docs to get started.