Mango Signal

Eleven apps ingest signals. Eight of them drive the weighted engagement score, theme clustering, and risk detection: Surveys, Shift Feedback, Leader Rounds, Recognitions, Performance, Communications, Ideas, and Ballots. The other three — Time & Attendance, Leave Management, and Frontline Execution — ingest attendance, burnout, churn-risk and execution-risk signals that are visible in the signal browse, export, and API but are not part of the weighted score. Mango Signal works with whatever subset you have enabled, and you can backfill history from a retired feedback tool via CSV import.

Mango Signal averages the last 30 days of normalized signals per type and blends them with fixed weights — sentiment 40%, recognition 20%, performance 15%, passive engagement 15%, and themes 10% — into a single 0..1 score. The window is admin-configurable (1–365 days), and scores are computed for the company and for every location and segment.

No. The dashboards, analytics, and AI assistant always aggregate to the company, a location, or a segment; named individual sentiment is never surfaced. The one exception is the raw-signal browse/export used for backfill and BI, which is gated to administrators by default — an admin can add named groups such as People Ops.

The close-the-loop flow fires (while it is switched on): a baseline Impact row captures the engagement score for the action's scope over the 30 days before completion; a "you said, we did" update goes to the people the action is about — a company Broadcast for a company-wide action, the Communications feed for an unscoped action or for the people at a location, and for a segment action only if the completer opts in to posting it to everyone — while the contributors behind the action get a direct email; and a follow-up job re-measures the score 30 days later to compute the delta.

When a location or segment drops below the risk threshold (default 0.4), the group you choose in Settings — or, with none chosen, your designated Mango Signal admins, or your business administrators if you haven't designated any — get one deduplicated email naming every segment that slipped and the signals behind it. On the dashboard and in Ask AI, Mango Signal recommends one to three concrete next actions per risk segment — AI-drafted via Claude, with a stable heuristic fallback — which a manager can assign with an owner, due date, and linked signals.

The assistant is gated by the per-tenant agent setting. When enabled it appears in the Ask AI sidebar with seven read tools and two write tools (create action, update action status). Every write requires explicit confirmation before it persists, and manager-tier data stays gated to managers and admins even inside the agent.

Yes, with the same admin gate the raw-signal browse carries: raw signals export to CSV or XLSX (with a spreadsheet formula-injection guard) across 7/30/90/180-day windows, restricted to administrators by default because those rows are individually attributable — an admin can widen that to named people or groups in the app's settings. Aggregate dashboards and analytics stay open to managers. There's also a scoped read/write REST API under /api/v1/mango_signal for native mobile and external integrators, tenant-scoped and role-gated to match the web surface.

Mango Signal is a premium add-on that requires a license and is opt-in per tenant. It ships disabled by default until an administrator enables it from the Apps Marketplace. Contact your OneMango administrator about turning it on for your organization.