Skip to content
plugboard.appRequest a demo

Running the desk

Sentiment & staff experience

Two related readings, both about staff: the tone of a staff member’s ticket, and how the desk has been treating a staff member lately. Neither is ever read for a student, a parent or a sender who matches nobody.

Read from the requester’s own messages on a ticket: the subject, the report and their public replies. Staff replies are never read.

Each reading carries:

  • a tone: calm, concerned, frustrated, angry or distressed;
  • an urgency: low, normal or high;
  • signals, such as chased, reopened, waiting a long time, or blocked from class;
  • one line saying why.

A rules pass runs when a ticket arrives and after each requester reply. With a model connected, a separate typed decision is queued at the same moments and reads tone and urgency; it never holds up intake. The queue gets a Tone column and a Frustrated requesters filter, and Home lists frustrated requesters under Needs you.

Tone and the AI welfare read run only when the ticket’s requester is a known staff member. A ticket from a student, a parent or a sender who matches nobody gets no model call, no stored tone and no rules tone, in any mode, shadow included, and nobody can set a tone on it by hand. An unknown sender may be a student or a parent, so the rule is an allowlist: staff, and nobody else.

If a ticket’s requester is changed from a staff member to someone else, its tone is cleared the next time the requester writes.

A rules pass still gives a coarse level from facts the desk already has: chases, reopens, time waiting against the target, and the last satisfaction rating. Rules results are labelled Rules, never AI, so nobody mistakes one for the other. The rules never read a ticket as distressed.

Requester tone follows the same modes as every other AI feature, set in AI settings:

ModeWhat happens
ShadowThe reading is recorded; nobody sees it
SuggestShown on the ticket, and a one-step priority bump is proposed for accept, dismiss or undo
AutoRaises priority one step by itself, once per ticket, never to the top priority, and it can always be undone

A ticket is flagged to anyone who can read internal notes (ticket.internal), alongside the school’s configured safeguarding contact, in two cases:

  • A student’s ticket uses a word from the school’s safeguarding list. This is a word match, not AI. It runs whatever the mode, with AI off and with requester tone off, the same as it does in the portal.
  • On a staff member’s ticket, the AI reads a possible welfare concern. This is Welfare concern review (ticket.safeguarding) in AI settings. It flags in Suggest, records quietly in Shadow, is never automatic, and never runs on a student’s or parent’s ticket. It can add a flag; it never clears one.

Home shows open flags under Needs you as “A wellbeing concern to follow up”. The flag decides nothing on its own; a person follows the school’s process. Every flag is audited.

A nightly job builds a 90-day summary for each staff Person: their median wait for a first reply, reopen rate, chases per ticket, average satisfaction, and the share of their tickets that read as frustrated, against the previous 90 days. A 0–100 score is arithmetic on those facts. Where a score is low enough to be worth a conversation, the AI may add one sentence, written from the numbers alone, with no names and no ticket text sent.

It shows on the person page, on Home under Spotted (“may be having a rough time with IT”), and on the Experience report (/reports/experience) in the Knowledge & reports group, beside Reports. It is meant as a conversation starter for a manager and is not a performance measure; the product says so wherever it appears.

The report’s Frustrated, across the school figure is the share of staff tickets whose requester read as frustrated or worse. Its Staff, Campuses and Service desks views count frustration over staff tickets too.

WhatWhereNeeds
Tone chip, filter and sortTicket queueticket.view
Tone, reason and correctionTicket pageticket.view to see it, ticket.update to correct it
Safeguarding flagTicket page, Home (Needs you)ticket.internal
Frustrated requestersHome (Needs you)ticket.view
Staff experience cardPerson page, staff onlyreport.view and client.view
A rough time with ITHome (Spotted)report.view and client.view
Experience reportKnowledge & reports, Experiencereport.view and client.view

A requester never sees any of this. The portal, kiosk, parent portal and tracking page never select the sentiment fields.

No tone, no AI welfare read and no emotional profile of a student or parent is computed or stored. Nothing about their feelings is sent to a model, and the code refuses to write a tone for them. The only check on a student’s words is the safeguarding word list above, which is not AI.

  • Ticket sentiment lives on the ticket, so it follows the ticket’s own retention window and goes when the ticket does.
  • The readings kept for audit follow the AI activity window in what we process.
  • Staff experience is recomputed every night; a row not refreshed for 7 days is deleted, which covers somebody who has left or changed role.
  • Erasing a person clears the sentiment on tickets they raised, deletes the tone and welfare readings of those tickets, and deletes their staff experience row.
  • A staff member’s subject-access export includes their staff experience row.

Requester tone (ticket.sentiment) and Welfare concern review (ticket.safeguarding) are separate features in AI settings, under Look out for people. Turning off requester tone stops the AI reading, the rules reading, and the staff summary sentence. The safeguarding word match keeps running regardless.