Independent CX Ops Project

Support QA Scorecard & Coaching Framework

A repeatable standard for what a “good” support response looks like — and how to coach toward it.

Self-initiated project · built to demonstrate CX operations thinking

Most support teams can tell you a reply was “good” or “bad” but not why, which makes coaching inconsistent and quality unrepeatable. This is the framework I built to fix that: a weighted rubric, a worked example scored against real handled work, and a coaching template that turns a score into a conversation.

Honesty note: This is an independent project, not a deliverable from a past employer. The worked example below is scored against a real customer interaction I handled in a travel-support role, with identifying details removed.

1. The problem it solves

Without a shared definition of quality, three things break: new agents guess at the standard, feedback depends on whichever lead reviews the ticket, and “improvement” can’t be measured because there’s no baseline. A scorecard makes the standard explicit, portable, and trainable.

2. The QA rubric

Six weighted criteria, 100 points total. Weighting is deliberate: accuracy and resolution carry the most because a warm tone on a wrong answer still loses the customer.

Criterion	Weight	What a top score looks like
Accuracy	25	Information is correct and complete; policy applied correctly; no guessing.
Resolution completeness	25	The actual problem is solved end-to-end, not just acknowledged; no avoidable second contact.
Tone & empathy	15	Matches the customer’s emotional state; de-escalates rather than mirrors frustration.
Ownership language	15	Takes responsibility (“I’ll get this sorted”) instead of deflecting to policy or other teams.
Clarity & structure	10	Easy to follow; next steps and timelines are explicit; no jargon.
Proactivity / follow-up	10	Anticipates the next question; confirms resolution or sets a clear follow-up.

3. Worked example — scored against real work

Situation: a customer demanded to cancel a booking after a mispriced fare and a duplicate charge, opening the chat angry and threatening a chargeback. Below is that handled interaction scored against the rubric.

Accuracy

23/25

Correctly identified the duplicate charge and applicable refund policy on first pass; minor deduction for one detail confirmed a message later than ideal.

Resolution completeness

25/25

Refund processed, duplicate reversed, and the booking rebooked at the correct fare in one thread — no second contact needed.

Tone & empathy

14/15

Opened by acknowledging the frustration and the money at stake before explaining anything; the customer’s tone shifted within three messages.

Ownership language

15/15

“I’ll get both charges sorted for you now” — no deflection to policy or another department.

Clarity & structure

9/10

Laid out the three fixes as clear steps with what to expect and when.

Proactivity / follow-up

10/10

Confirmed the refund timeline unprompted and left the thread open for the reversal to post.

Total96 / 100

Outcome of the real interaction: the cancellation demand became a completed rebooking and a five-star review. The score isn’t the point — the point is that anyone on the team could now score the next angry-cancellation the same way, and coach against the same standard.

4. Review cadence

Weekly: each agent gets 3–5 tickets scored, sampled across channels and difficulty (not just the easy wins).
Monthly: roll scores into a per-criterion trend so coaching targets the specific weak criterion, not a vague “do better.”
Calibration: reviewers score the same ticket quarterly to keep scoring consistent between people.

5. Coaching template

A score with no conversation changes nothing. Each review is delivered in this four-part structure so feedback is specific and forward-looking:

Section	Prompt
Strength	The one criterion scored highest — name it and why, so it gets repeated.
Focus	The single lowest criterion — one, not five, so the next ticket has a clear target.
Example	The exact line from the ticket, with a suggested rewrite.
Next check	What “improved” will look like in next week’s sample.

Why this matters for the role

Answering tickets well is execution. Defining what “well” means, making it measurable, and building the loop that raises everyone to it is operations. This artifact is evidence of the second thing.