QA evaluation sheets: how scorecards grade agent calls

A QA evaluation sheet is a structured scorecard a quality analyst uses to grade an agent’s call, chat, or email against fixed criteria like greeting, accuracy, empathy, resolution, and compliance. Each line item carries a weight, and the total score feeds coaching, calibration, and contact-centre performance reports.

Most BPO floors run one sheet per channel and one per programme. A voice form for a telco will look different from a chat form for an e-commerce account, but the bones are the same: a header (agent, date, call ID), a set of weighted criteria, a comments box, and a final percentage.

The form is the unit of measurement. Calibration sessions, coaching plans, bonus calculations, and even client QBRs all trace back to the score on the sheet. Get the sheet wrong and the rest of the operation drifts. ICMI’s contact-center resource library treats the scorecard as the single most important QA artefact.

How it works

A scorecard moves through a fixed loop, and most contact centres run roughly 4–8 evaluations per agent per month — enough to be statistically useful without crushing the QA team’s bandwidth. It also gives team leaders something concrete to walk through in weekly one-on-ones.

A typical voice scorecard splits into three buckets. Soft skills (greeting, tone, empathy, active listening) usually carries 25–35% of the weight. Process compliance (verification, disclosures, accurate data entry) takes another 30–40%. Resolution and accuracy (right answer, no rework, customer confirmed satisfied) closes out the rest.

Fatal errors — a missed compliance disclosure, a data-protection breach, or a rude exchange — usually auto-zero the score regardless of other line items.

The COPC Customer Experience Standard, now in Release 8.0, is the industry benchmark most enterprise BPOs design their forms against. It pushes operators to grade human agents, chatbots, and self-service against the same outcome metrics so the scorecard reflects real customer experience, not just script adherence.

Examples

Concentrix (Manila, ongoing). The world’s largest CX provider runs programme-specific scorecards across its Philippine sites for clients in banking, healthcare, and tech. Forms typically weigh resolution and compliance heavily for regulated accounts, and calibration sessions run weekly to keep analyst-to-analyst variance under 5%.

Teleperformance (global, 2024 deployment). TP’s TAP (Technology, Analytics, Process) framework layered speech analytics on top of manual QA sheets across multiple sites. Auto-scored items (silence, talk-over, sentiment) feed the same sheet humans grade, so the final form blends machine and analyst marks.

Foundever (formerly Sitel Group). Foundever publishes scorecard guidance that treats the form as a coaching artefact first and a compliance tool second. Comments and “what good looks like” examples sit beside every line item so agents can self-review before their next call, and team leaders can reference the same wording in coaching huddles.

A mid-sized Cebu BPO running outbound sales (2025). A 20-line voice scorecard with three fatals (no Do-Not-Call check, no recording disclosure, no payment-card masking). Analysts grade 6 calls per agent per month, and any fatal triggers an immediate refresher session before the agent goes back on queue.

Related terms

• Quality assurance: the broader function that owns the scorecard and the people running it.
• Call calibration: the session where analysts grade the same call and reconcile scores.
• Average handle time: a productivity metric that often sits alongside QA scores, not inside them.
• First call resolution: an outcome metric many scorecards weight heavily.
• Customer satisfaction score: the customer-side companion to the analyst-side QA score.
• Net promoter score: a loyalty metric clients often pair with QA scores in QBRs.
• Service level agreement: the contract that defines what “passing” QA actually means for a client.

FAQ

What is a QA evaluation sheet used for?

It’s used to grade an agent’s interaction against fixed, weighted criteria so you can coach consistently, calibrate across analysts, and report performance to clients. The form turns a subjective call into a comparable number.

How many calls should be evaluated per agent per month?

Most contact centres land between 4 and 8 calls per agent per month. Fewer than four, and the sample isn’t statistically useful; more than eight tends to overwhelm the QA team without changing the coaching picture.

Who fills out the QA evaluation sheet?

A dedicated QA analyst, not the agent’s direct team leader. Separation matters: the analyst owns scoring objectivity, and the team leader owns the coaching conversation that follows. Calibration sessions keep multiple analysts aligned on what each line item really means.

What’s the difference between QA scores and CSAT?

QA scores are internal: an analyst grading the agent against your standards. CSAT is external: the customer rating their own experience. The two correlate, but not perfectly, which is why mature operations track both and reconcile the gap each month.

What counts as a fatal error on a scorecard?

A fatal is anything that auto-zeros the score regardless of other line items. Typical fatals include missed compliance disclosures, data-protection breaches, abusive language, or providing incorrect financial or medical information. Fatals usually trigger an immediate retraining flag.

Can QA evaluation sheets work for chat and email too?

Yes. The format adapts: voice forms grade tone and active listening, chat forms grade typing accuracy and channel etiquette, and email forms grade structure and resolution completeness. The weighting changes, but the scorecard logic doesn’t.

If you’re standing up QA for an offshore team or rebuilding scorecards that have drifted, talk to Outsource Accelerator’s advisory team — we’ll match you with BPO partners whose QA frameworks already meet the standard you need, so the scorecard becomes a real coaching engine rather than a monthly compliance ritual.