Signal — QA Manager · Quality & Voice of Customer

Prepared for the CX Leadership Review100% Auto-QA Coverage

Lead Indicator

89%

▲ 1.4 pts MoM Internal Quality Score · 12-week trend Industry benchmark band ≈ 88%

IQS (this team) Benchmark band

✦

Auto-QA · 100% coverage AI graded all 41,902 conversations this week — not a 5% sample. The +1.4-pt lift is driven almost entirely by Tone & Empathy on cancellations, the behavior most correlated with CSAT. Two reviewers show grading drift; calibration assist has flagged 9 conversations for re-grade.

CSAT

94%

▲ 0.8pp · 4,118 ratings

NPS

+62

▲ 3 pts · promoters 71%

CES

5.4/7

▲ 0.2 · low-effort

Auto-QA Coverage

100%

+ 38% manual sampled

Reviews / wk

1,612

manual scored · 6 reviewers

Coaching

91%

▲ 4pp · actioned

The Breakdown

What's moving the score

Weighted scorecard · driver correlation · click any row to drill in

Weighted Scorecard

Compliance is a gate, not a score

Accuracy 40%92

Tone & Empathy 20%95

Resolution 20%88

Process Adherence 10%90

Documentation 10%87

! Compliance gate · auto-fail 1.8% — failed conversations score 0 regardless of category points. Top reason: missing identity verification.

Agent QA Leaderboard

12-week IQS · 4 coaching active

Agent	IQS	CSAT	Rev	Trend	Coaching
1Priya N.	94	96	41	▲	on track
2Dan M.	91	93	38	▲	active
3Lena K.	90	92	35	▬	active
4Marcus T.	89	90	37	▲	—
5Sofia R.	87	89	33	▼	active
6Jamal W.	86	88	30	▬	active

CSAT × QA Drivers

Which behaviors actually move satisfaction

Tone & Empathy+0.41

Strongest lever — a 5-pt tone gain ≈ +2.0pp CSAT

Resolution+0.33

One-touch resolutions rate 8pp higher on CSAT

Accuracy+0.22

Necessary, but plateaus once above 90

Process Adherence+0.08

Weak correlation — don't over-index coaching here

✦

Driver insightCoaching the bottom quartile on Tone over Process is projected to lift floor CSAT +1.6pp next quarter.

Root-Cause Clusters

AI-named low-CSAT themes · click to open

Refund delay

Billing Ops · −6pp CSAT drag on affected tickets

312tickets

▲ $18K

Tone on cancellations

Retention queue · coachable behavior, trending down

188tickets

▼ $9K

Shipping ETA confusion

Logistics · KB article gap suspected

141tickets

▬ $7K

Account access loops

Auth · repeat-contact heavy

97tickets

▲ $5K

Program Health

Is the QA program itself trustworthy?

Calibration · coaching follow-through · auto-fail watch

Calibration Variance

Reviewer agreement spread is within ±5% tolerance across 6 reviewers. AI calibration assist flagged 9 conversations where human and model scores diverge by >8 pts.

Coaching Follow-Through

Share of flagged sessions actioned into a documented coaching step

142 of 156 flagged actioned91% · ▲ 4pp

1.8%

Auto-fail rate · ▼ 0.3pp · top reason: missing verification (0.7%)

In Their Words

The Verbatim Wall

AI-selected, sentiment-tagged customer comments · most representative of each cluster

”

Agent fixed it in one chat — honestly amazing. Didn't have to repeat myself once.

Order tracking · 2-touchPositive

”

The refund took way too long. Polite people, but I waited eight days for my money.

Refund delay clusterNegative

”

Friendly and patient, but couldn't actually solve my address change. Got bounced.

Shipping ETA clusterMixed

”

She caught the duplicate charge before I even noticed it. That's service.

Billing · proactivePositive

”

Fastest support I've had from any subscription I pay for. Five stars.

Account · 1-touchPositive

”

Third time contacting about the same login loop. Nice tone doesn't fix the bug.

Account access clusterNegative

Field Guide

Reading this report like a QA lead

How to use this view

What decisions this report drives in a CX leadership review

1
Watch the hero curve first. IQS vs the benchmark band tells you if quality is genuinely improving or just bouncing inside normal noise. A point above the band's top edge is a real, defensible gain.
2
Let the driver chart pick your coaching focus. Coach the behavior with the highest CSAT correlation, not the lowest raw score. Here, Tone (+0.41) beats Process (+0.08) — spend coaching hours where they move the customer.
3
Treat compliance as a gate, never a number. A 92 Accuracy score means nothing if the conversation auto-failed verification. Read the gate banner before the bars.
4
Trust the score because of calibration. 100% Auto-QA is only credible if variance stays in tolerance. Check ±variance before you present any leaderboard.
5
Close the loop with follow-through. Flagging a session means nothing until it becomes a coaching step. 91% actioned is the metric that turns insight into behavior change.
★
For your own org: if you only graded 5% of conversations, which of these clusters would you have missed entirely — and what did that blind spot cost you last quarter?

In context

Sample feed Illustrative — wire to your VoC & benchmark feed

How this team reads against published industry benchmarks

Internal Quality Scorevs SaaS support median

89%/ 88%+1.0pp

CSATvs SaaS support median

94%/ 90%+4.0pp

Net Promoter Scorevs software industry avg

+62/ +45+17

Customer Effort (CES)vs cross-industry avg

5.4/ 5.1+0.3

Auto-fail / compliance ratevs regulated-vertical avg

1.8%/ 2.6%−0.8pp

VoC · 41m

Sentiment up: "one-touch" mentions +14% week-over-week across chat verbatims.

Cluster · 1h

Emerging theme: AI grouped 38 new tickets under "promo code wouldn't apply."

Calib · 3h

Reviewer drift: grader R-4 trending +6 pts strict on Tone — re-grade proposed.

Watch the walkthrough

Four AI agents walk this dashboard — what to watch, what to act on

Four AI agents walk this dashboard.

The Quality Review