Research question and scope
This protocol measures how often a voicemail callback misses the window promised to the caller. The unit is one voicemail that entered the callback queue during the observation period. A miss means no successful contact occurred before the promised time. The measure describes the promise and the queue, not the effort of the specialist.
Set the scope to one support line, one published callback rule, and a fixed period. Record the staffed hours, the message types, and the time zone used for the promise. If the rule changes during the period, split the sample at the change date. A different window creates a different denominator.
Keep the measure separate from a performance rating. A miss may come from a late message, an unreachable caller, or a capacity shortfall. The purpose is to find where the promise breaks. A neutral framing makes the team more willing to report the real cause.
Population and unit of analysis
Define the population as every voicemail that reached the callback queue, including messages that were returned successfully. The unit is the voicemail, not the call attempt. One message may need three attempts, and each attempt should not become a separate case. Use the queue identifier to group them.
Include messages that were closed without contact when the caller could not be reached. Report them separately from messages that were never attempted. The two groups describe different failures. Excluding unreachable callers would overstate the desk's control.
Exclude calls that were answered live and never became a voicemail. Report the live answer rate beside the callback result if the line uses one queue. The two measures describe different parts of the service. Keep the boundary visible in the report.
Definitions that make the measure reproducible
Define the promise as the window published in the greeting or the client agreement, including the start time and the time zone. Define a successful contact as a conversation in which the caller receives the answer or a clear next step. A returned call that only leaves another message is not a success. Publish one example of each outcome.
Define a completed attempt as a dial with a recorded result such as answered, busy, or no answer. Define an unresolved message as one that reached the last allowed attempt without contact. Record the attempt count with the status. A status without the count cannot be checked.
Decide how to treat a message that the caller withdraws or resolves elsewhere. Mark it as withdrawn with the date and the source. Do not count it as a miss or a success. A clear rule keeps the same case from being classified differently by two reviewers.
Sampling and data collection
Draw the sample from the phone or queue system for the full period rather than from a list of missed callbacks. Record the export date, the filters, and the row count. Include every message type and every hour the line was staffed. A sample built from misses alone cannot produce a rate.
Capture the message identifier, the received time, the promised time, the caller time zone, the attempts, the outcome, and the owner. Add the queue record that holds the message. Store the export with the report. The timestamps and attempts are the core evidence.
Check a small overlap sample with a second reviewer. If the two reviewers disagree about whether a callback was on time, the definition of contact is too vague. Rewrite the rule and retest before the full sample. Agreement matters more than sample size for a measure that will be reused.
Record the time spent waiting for the caller to answer as part of the attempt. A number that rings for two minutes is different from one that connects at once. Keep the wait beside the outcome. The detail helps explain a miss that was not the desk's choice.
Measuring callback misses
Report the share of messages contacted within the window, the share contacted later, and the share never contacted. Show the counts behind each share. The three groups describe different levels of failure and should not be collapsed into one rate. A single rate hides whether the problem is timing or reachability.
Break the result down by message type, hour received, and owner group. A type with many misses may need a longer window or a different script. A cluster at the end of the shift may point to the cutoff. Compare only groups with similar volume and complexity.
Report the median and longest delay for late callbacks. A miss of ten minutes is different from a miss of two days. The delay distribution shows whether the problem sits at the boundary or earlier in the queue. Include the promised window in the report so the reader can interpret it.
Limitations and next action
The measure cannot show a caller who resolved the issue elsewhere before the callback or a message left on the wrong line. It also cannot show whether the answer was correct. Treat the result as a measure of the callback promise. Note the channels and hours that were not included.
Give the result to the desk owner and the client contact with one change to test. The change might be an earlier cutoff, a different greeting, or a protected callback block. Record the effective date and repeat the measure with the same definitions. A controlled change is easier to judge.
Review the boundary cases before changing the whole window. A few messages received in the last minutes may explain a large share of the misses. Fix the intake rule first if that is the pattern. The goal is a promise the desk can keep, not a lower number by definition.
