A team of six operators handles 900 calls a week. The quality lead samples five calls and reports that quality is good. The sample is enough to complete a form and not enough to show anything.
Quality sample sizes are a practical decision for a small team. This method note describes how to choose a size, how to draw the sample, and what a small sample cannot prove.
Start from the purpose
A sample can support coaching, process review, or a client report. Each purpose needs a different size and a different rule.
Write the purpose down. A sample drawn for coaching should include the calls the operator needs to discuss, not only a random selection.
Choose a rule
A common rule is a fixed number per operator per week, plus every complaint and every escalated call. The fixed number gives a baseline; the exceptions catch the cases that matter most.
Record the rule. A rule that changes every week produces a sample that cannot be compared.
What a small sample can show
A sample of five calls can show specific behaviours: a missed verification step, an unclear greeting, or a note that omits the callback number.
It can support a coaching conversation. The reviewer can point to the call and discuss the choice without claiming a team-wide pattern.
What a small sample cannot show
A sample of five cannot show a rate. A single defect in five calls is not a 20 percent failure rate for the team; it is one observed defect.
It cannot show whether the defect is common. That needs a larger sample or a longer period, and the report should say so.
It also cannot prove the absence of a problem. A clean sample is evidence that these calls were fine, not that the process is sound.
Methodology and limits
This note is an operating design informed by public guidance on time measurement, privacy, and quality review. It does not report an empirical study or prescribe a statistical standard.
The approach was tested against common review needs: a weekly coaching sample, a complaint follow-up, a client quality report, and an audit preparation. The design keeps the purpose visible.
Draw the sample by rule
Use a fixed interval, such as every twentieth call, plus all complaints and escalations. The rule prevents the reviewer from choosing the easy calls.
Record the interval and the exceptions. A sample that always includes the same operator or the same hour is not representative.
Review the sample consistently
Use the same checklist for every call and record the result. A changing checklist makes the comparison meaningless.
Where a call fails, note the specific behaviour and the coaching action. A general comment such as "needs improvement" is not reviewable.
Report the count and the finding
Report the number of calls sampled, the defects observed, and the actions taken. Do not convert a small sample into a percentage.
Where a client asks for a rate, explain the sample size and the uncertainty. A short explanation is better than a number that will be challenged.
Calibration
Two reviewers should agree on what a defect is. Run a short calibration every quarter with the same three calls and compare the findings.
Record the disagreements and resolve them. A checklist that two people read differently will produce a sample that cannot be compared.
Reviewer time
A review takes longer than the call. Budget fifteen minutes per call for listening, noting, and coaching, and choose the sample to fit the time available.
Record the time spent. A sample that consumes the whole week leaves no time for the coaching it is meant to support.
Client reporting
If the client receives a quality report, show the number of calls sampled and the findings. A percentage without the count invites a false confidence.
State the limits in one line. The client should know whether the sample can support a claim about the team.
Data quality
Check that the recordings are complete and the notes match the call. A sample drawn from missing recordings is not a sample of the work.
Record the excluded calls and the reason. A high exclusion rate is a finding about the recording system.
A worked example
A team samples 30 calls in a month: five per operator per week plus all complaints. The review finds four missed verification steps, all in the same week.
The finding is a training need for one shift, not a team-wide failure. The sample was large enough to locate the pattern and small enough to review.
Review the sample size quarterly
Ask whether the sample is finding problems and whether the reviewers can keep up. A sample that is never completed is worse than a smaller one that is.
Adjust the rule and record the change. The size should follow the purpose and the capacity.
Limits and appropriate use
This note is a method design, not an empirical study. It does not establish a statistical confidence level. Results depend on the consistency of the review and the honesty of the reviewer.
Small teams should read the counts directly and avoid rates from tiny samples. The sample is a tool for learning, not a scorecard.
Conclusion
A quality sample size is a decision about purpose and capacity. A small sample can support coaching and specific findings; it cannot support a team-wide rate. Used with a stable rule and a consistent checklist, it improves the work without overclaiming.
