Quality and Call QA: Disputes
Objective. Open Disputes and read a dispute as a status that changes a score's meaning and a fairness signal for the programme - not as blame on the reviewer or the agent.
Disputes give agents a fair path to challenge a score. This screen answers 'which scores are being contested, and how fairly are challenges resolved?' - a trust control, not a scoreboard of who was wrong.
See it in the demo
What this screen answers
The Disputes screen tracks scores an agent has challenged and how they are being resolved. It answers 'are scores defensible when questioned, and is the challenge path working?' Sales uses it to disarm the 'we don't trust QA scoring' objection by showing agents have a fair route; a CSM uses it to validate that the path works and that disputes resolve rather than fester.
The task
Scope to the team or period and read disputes by status: raised, under review, resolved, and the outcome. Open one and trace it - the original score, the agent's challenge, the evidence re-examined, and the resolution. The task is to confirm the path functions and resolves fairly, and to treat dispute rate as a signal to investigate, not a verdict on anyone.
How to read it
A dispute changes a score's meaning and status - an open dispute means the score is provisional, not final, and should not be rolled into an aggregate as settled. The classic misread is reading a dispute, or a high dispute rate, as proof the reviewer was wrong or the agent is difficult. It is neither. A high dispute rate is most often a rubric-ambiguity or calibration signal - reviewers and agents are reading 'good' differently. Read disputes as a health check on trust and rubric clarity, and remember a resolved dispute is the point where a score becomes defensible, not where someone is blamed.
Honest limits
Dispute data does not prove the overall scoring is accurate, only that contested scores got a second look. A low dispute rate can mean fair scores or disengaged agents who have stopped bothering. Confirm with product how the dispute workflow, roles and resolution SLAs are configured and entitled; the demo shows the path on seeded data.
Interpretation check
An open dispute makes a score provisional, and a high dispute rate is usually a rubric-clarity or calibration signal - not proof the reviewer was wrong or the agent is difficult.
In the field
OVO (contact centre): in a voice operation where agent consequences ride on scores, a working dispute path is the difference between QA agents trust and QA they resist - and the dispute rate is an early warning on rubric clarity.
Practical task
Open the Disputes screen, pick one dispute, and write it up: original score, the agent's challenge, the evidence re-examined, the resolution - then state what a rising dispute rate would tell you to investigate first and why.
Self-check
An interaction has an open dispute. Can its score go into this month's aggregate as final?
No. An open dispute makes the score provisional; treating it as settled overstates certainty. It counts once resolved.
Dispute rate jumped this quarter. What is the most useful hypothesis?
Rubric ambiguity or a calibration gap - reviewers and agents are reading 'good' differently. Investigate the rubric and calibration before blaming reviewers or agents.
A team has almost no disputes. Is that unambiguously good?
No. It can mean fair, trusted scores, or agents who have disengaged and stopped challenging. Read it with adoption signals, not alone.
How does the dispute path help a Sales conversation about AI distrust?
It shows a human-in-the-loop fairness route - any score with consequences can be challenged and re-examined against evidence, so a human stands behind it.
Verified vs demo. Screen is live in the demo at /#/review/disputes; exact controls and figures are SME-validation-pending.