Resonant IQ Help & Docs
Concepts QA Insights Open Resonant IQ
QA / Scores, flags & coaching
QA workspace

Scores, flags & coaching

The three actions you take on a scored conversation each open into a full workflow. This is where a correction, a dispute, or a coaching note actually goes — and how you close the loop.

7 min read
For QA managers & admins
QA Manager Admin Lead

In Working the Queue you met the three actions that matter — correct a score, raise a flag, add a coaching note. Each one writes its own audit trail and kicks off its own small process. Here's where each leads.

Correction
A human score that lives alongside the AI's, never erasing it.
Flag
A dispute that stays open until someone resolves it.
Coaching
An assignment that runs from “needs coaching” to “done.”

Score corrections

When the AI gets a score wrong, you correct it — per criterion or overall. Your number becomes the human score of record, but the AI's original stays visible. That preserved pair is what lets you measure how well the AI is calibrating over time.

Scored conversation · Correct score Reconstructed — not a live capture
Issue resolution ✏ corrected
AI score 4 / 10
Human score 7 / 10
Agent followed the refund policy correctly — the AI penalized a step that was actually required. Context was in the linked ticket.
Corrected by You · Apr 19, 2:14pm · original AI score retained

Flags & disputes

A flag is a held question on a score. Agents can flag their own scores when they think the AI got it wrong; managers and leads raise system flags on anyone's. A flag stays open — and surfaces on the Queue's Flagged tab — until someone resolves it.

Open Resolved Two states. No middle ground.
Reconstructed — not a live capture
Tone & empathy score disputed Open
Tom B. flagged his own score — says the transcript was truncated and the AI missed the resolution.
Raised by Tom B. (agent) · 2h ago
Resolved — re-scored Resolved
Confirmed the truncation, re-ran scoring on the full transcript. Tone corrected 5 → 8.
Resolved by QA Manager · 1h ago

The coaching workflow

A coaching note isn't a dead end — it becomes an assignment with a real lifecycle. You assign it to an agent with a due date, they acknowledge it, you coach, optionally follow up, and close it out. Every transition is tracked.

Open Acknowledged Coached Followed Up Completed (or dismissed)
QA · Coaching Reconstructed — not a live capture
MR
Coaching: Issue resolution Acknowledged
Maya R. · assigned by You · due Apr 25
Resolution scores dipped on billing tickets this week. Let's review the escalation path together.
Mark coached Schedule follow-up Complete Dismiss

Coaching history

The history view answers the question that makes coaching worth doing: did it work? Roll up by agent or by scorecard to see what you coached and whether scores moved afterward.

Assignments (30d)
12
9 completed
Avg score lift
+8 pts
after coaching
Follow-up rate
75%
of coached items
MR
Maya R.
4 assignments
64 78 +14
DP
Dev P.
3 assignments
71 80 +9
TB
Tom B.
2 assignments
69 72 +3

Staying in the loop

Coaching runs on notifications so nothing stalls. Each transition can ping the right person, in-app or by email, per their preferences.

coaching_assigned The agent is notified the moment you assign coaching to them.
coaching_due_soon A reminder fires before the due date so nothing slips.
coaching_completed You're notified when an assignment is closed out.

What happens next

Corrections, resolved flags, and completed coaching all feed back into the picture — sharper AI calibration, cleaner scores, and a team that's improving where it counts. To see how individual scores roll up into company-wide patterns, head to Reading the Insights hub.

← QA
Working the Queue
Next up
Building scorecards & rubrics →