Glossary
The FeedbackGraph glossary
Plain-language definitions of the concepts behind FeedbackGraph — from the revenue-backed Evidence Graph to the read-first MCP server. Each entry stands on its own.
Evidence Graph
- FeedbackGraph’s revenue-backed graph that links every feedback signal to its account and the ARR behind it, so themes can be ranked by the revenue they affect rather than by raw vote count.
Revenue-backed opportunity
- A scored theme — a cluster of related signals — ranked by the accounts and ARR it touches, so a team can build what moves revenue instead of just what is loudest.
Closed-loop
- Automatically telling the original reporter when their issue changes status or ships. It closes the gap where feedback usually disappears into a backlog and the reporter never hears back.
MCP server
- A Model Context Protocol endpoint that lets AI agents (Claude Code, Cursor, VS Code, Windsurf, Slack) read FeedbackGraph reports, debugging evidence and revenue-ranked opportunities directly — with scoped tokens and per-tenant isolation.
AI triage with evidence
- FeedbackGraph’s AI writes each report’s title, summary, type and severity, and shows that verdict beside the raw evidence it came from (console errors, failed requests, JS exceptions, on-screen text) so engineers can verify it rather than trust a black box.
Semantic dedupe
- Using vector embeddings to merge near-identical reports into a single issue with an occurrence count, instead of relying on exact text matching — so the inbox shows one issue reported 200 times, not 200 issues.
Two-way status sync
- Keeping a report’s status and its routed tracker issue (Jira, Linear, GitHub) in agreement in both directions, so a change made in either place is reflected in the other.
Opportunity dossier
- The detail view of an opportunity: its score, the ARR influenced, the affected accounts (with their ARR), the backing signals, and a suggested next step.
PII redaction
- Stripping personally identifiable information from captured evidence before it is processed by AI or shown in the dashboard, so sensitive data never leaves the boundary it should stay in.
See these ideas in the product
Explore the features, the MCP server, or how FeedbackGraph compares to Jam.dev, Canny and Enterpret.