A QA score should show how well a customer interaction met a defined set of expectations. When those expectations carry different levels of importance, a weighted score gives each parameter its agreed share of the final result.
This guide explains the calculation using a simple example, shows how a critical error can override the result, and highlights the checks to complete before using a score for coaching. All scores and weights below are illustrative.
What is a weighted QA score?
A weighted QA score combines individual parameter scores according to their assigned importance. For example, a scorecard might give resolution 50% of the total, communication 30%, and compliance 20%. These are example weights, not a recommended standard for every operation.
Keep two values separate: the weight is the parameter’s share of the scorecard; the score is the performance achieved on that parameter. A parameter can have a 20% weight and a 100% performance score.
The weighted QA score formula
Weighted QA score (%) = sum of (parameter score (%) × parameter weight (%) ÷ 100)
This formula assumes that weights total 100% and every parameter score uses a 0–100 percentage scale. If your scorecard uses points, convert each score first: earned points ÷ available points × 100. For example, 9 out of 10 becomes 90%.
A contribution is measured in percentage points. A score of 90% on a parameter with a 30% weight contributes 27 percentage points to the overall score.
Worked example: a final score of 93%
| Parameter | Weight | Score | Contribution |
|---|---|---|---|
| Communication | 30% | 90% | 27 points |
| Resolution | 50% | 92% | 46 points |
| Compliance | 20% | 100% | 20 points |
| Total | 100% | — | 93 points |
The calculation is (90 × 30 ÷ 100) + (92 × 50 ÷ 100) + (100 × 20 ÷ 100) = 93%.
Simply averaging 90, 92, and 100 would give 94%. That would treat all three parameters as equally important and ignore the agreed weights. Use a simple average only when equal weighting is intended.
How fatal or critical errors affect the score
A critical error rule is a separate decision applied to the calculated score. Where an organisation’s scorecard requires an overall zero after a qualifying failure, the final score becomes 0% even if other parameters performed well. Definitions and consequences must be documented in that organisation’s QA policy.
Consider a second example with communication at 90%, resolution at 92%, and compliance at 0%. The raw weighted score is 27 + 46 + 0 = 73%. If the compliance failure also meets the documented fatal-error definition, an overall-zero rule changes the final result to 0%.
A zero parameter score does not automatically mean a fatal error occurred. A normal scoring deduction and a fatal-error override are separate decisions. Similarly, labelling a category “critical” does not prove that an error occurred in the interaction being reviewed.
Retain the raw score, final score, error reason, and supporting evidence. This lets a reviewer understand the override while still seeing which other behaviours need coaching.
Use the CX Metric Works calculator
Open the QA Scorecard Calculator. Add your parameter names, enter each weight, and enter the percentage score achieved. Check that the weights total exactly 100%.
For the first example, enter Communication: weight 30, score 90; Resolution: weight 50, score 92; and Compliance: weight 20, score 100. Calculate to obtain 93%.
Select Fatal / critical error occurred only when the reviewed interaction contains a qualifying error under your policy. The calculator applies the overall-zero rule when a parameter is flagged and the inputs are valid. It also displays the raw weighted score.
You can enter an optional target for comparison. Choose this from your own agreed expectations; the example score of 93% is not a universal pass mark. After changing inputs, calculate again so the result reflects the updated values.
What should you do with “Not applicable”?
An inapplicable parameter is different from a failed parameter. Entering zero for an item that could not be assessed would reduce the score as if the employee failed it.
Follow your documented N/A method. If that method permits proportional redistribution, removing a 20% parameter leaves 80% of the original weights. Original weights of 30% and 50% then become 37.5% and 62.5% respectively: each remaining weight ÷ 80 × 100.
The CX Metric Works calculator does not automatically redistribute N/A weights. Make only adjustments allowed by your scoring policy, and keep the revised total at 100%.
Check the scoring rules before reviewing interactions
Define observable evidence. Replace vague criteria such as “good communication” with behaviours reviewers can identify consistently. Record what was expected, what happened, and the evidence supporting the score.
Document critical failures. Specify which errors trigger an override, how exceptions are handled, and who resolves disputed findings. Do not introduce a new fatal-error rule after an interaction has already been scored.
Calibrate with the same examples. Have reviewers independently assess the same interaction, compare their decisions, and clarify differences in interpretation. A correct formula cannot compensate for inconsistent scoring.
Turn the result into a coaching action. Use the parameter breakdown to identify an observable behaviour to improve. The percentage is a useful summary, but the explanation tells the employee what to do differently.
Try the calculation yourself
Use the free QA Scorecard Calculator to create a weighted scorecard, test the worked example, and see how a confirmed critical error changes the final result.
Further reading: Zendesk’s guide to creating a customer service QA program discusses rating categories, weights, critical categories, and calibration. The worked calculations in this article are illustrative examples created for CX Metric Works.