Feedback, without the front-end
Pulse is a headless, API-first feedback engine for NPS and CSAT. Deploy it behind your own chat, CRM or product interface and let every interaction become a signal you can act on. There is no dashboard to log into, because the score belongs where the work already happens.
One MCP endpoint. Two tools. No credit card to begin.
One call, whole workflow
Everything below happens through execute. There is no second system to open, no CSV to reconcile,
and no dashboard anybody has to remember to check.
// An agent, mid-conversation, writes the score down. { "operation": "feedback.record", "params": { "email": "[email protected]", "survey": "q3-nps", "score": 9, "comment": "Support fixed it in an hour." } } // Back comes the segment, already computed, // and everything that followed from it. { "recorded": true, "response": { "score": 9, "segment": "promoter", "period": "2026-08" }, "escalated": false, "wrote_back": [{ "source": "crm", "ok": true }] }
A score goes in raw and the segment comes out computed — promoter, passive, detractor, or the CSAT and CES equivalents. You cannot write a segment here, which is the only reason the numbers can be trusted a year later.
A detractor is escalated the moment it lands: the owner is emailed, a ticket is raised in your helpdesk, and it stays on the worklist until somebody records that they went back to them.
The score is logged on the contact's own timeline in headless-crm, so the account owner reads it where they already work, without ever opening this.
The idea
Customers come from your CRM and employees from your HRMS. Pulse keeps a thin index — enough to address an email and group a report — and the source system stays authoritative for who anybody is. Nothing is copied twice, and nothing goes stale in two places at once.
Mirror your contacts once. Their company, owner, plan and tags come with them, and the answers go back onto their timeline there.
The same instruments, asked about working here. Employee surveys are anonymous by default, and the anonymity is structural.
A pasted list, a post-call NPS from an address you have never seen. Pass an email and the pointer is created on the fly.
Standard instruments, scored properly
Pick a type and the question, the scale and the segment bands come with it. Move a threshold if you must — but the arithmetic is always the engine's, never the caller's.
| Instrument | Scale | Segments | Reported as |
|---|---|---|---|
| NPS Net Promoter Score |
0–10 | promoter 9+ passive 7–8 detractor 0–6 | % promoters − % detractors |
| CSAT Customer Satisfaction |
1–5 | satisfied 4+ neutral 3 dissatisfied 1–2 | Top-two-box, as a percentage |
| CES Customer Effort Score |
1–7 | effortless 5+ neutral 4 strained 1–3 | Mean effort, with % effortless |
| Pulse A scale of your own |
Yours | satisfied neutral dissatisfied | Mean score |
Every score comes back with the count and the margin beside it. Thirty answers out of forty is a reading; thirty out of four thousand is a self-selected complaint, and a report that does not say which is worse than none.
What you get
Standardised instruments with automatic segment classification and threshold scoring — and a stamped period, so a year of trend is an index scan.
Bind satisfaction to a ticket, an order or a call. context travels with the answer, so you can read the score by agent, by product, by the exact moment it happened.
Unhappy answers route to whoever owns the relationship the instant they arrive, and open a ticket in your helpdesk.
A single unified grid across every survey, response and respondent — filters, sorting, search, saved columns, CSV. No admin sprawl.
An anonymous answer has its link to the person destroyed, not hidden — and a team too small to be anonymous is suppressed from every breakdown.
Every ask, answer, escalation and sync on one timeline. Soft deletes, restore, and CSV or JSON out whenever you want it.
Your Organization is your isolated tenant — every survey, response and score lives under it. No credit card required to begin.