Measuring the Share of Callback Promises Kept Within the Promised Day

Research question and evidence scope

When a caller promises to call back on a specific day, how often does the documented attempt actually happen inside that day? The scope is one outsourced calling queue during a defined period, not an industry benchmark. The share matters because a promise that slips by one day can turn a warm conversation into a complaint. The measure tracks the promise and the attempt together rather than counting calls in isolation.

This review does not decide whether the callback resolved the customer's question or whether the caller was polite. Those questions need the recording and a quality review. The narrow purpose is to make the promise-keeping rate visible and repeatable across shifts. Record the definition, the sample, and the observation window beside the result so another reviewer can check the same thing.

The unit is one callback promise linked to one customer record. A promise made twice for the same issue should be kept as two events with the same reference. An attempt that reached voicemail is still an attempt, but it is not a completed conversation. Keep the outcome and the next action linked so the share is not confused with a resolution rate.

Operating scenario and evidence

Start with a dated export of promises made during the review window. Record the customer reference, the promise date, the promised day, the time zone, the owner, and the campaign. Include promises that were later rescheduled so the change is visible. Exclude internal reminders and test records, and state those exclusions in the note. Keep the raw export unchanged for the comparison.

A promise is kept when a documented attempt reaches the customer on the promised day. A promise is late when the first attempt happens after that day, even if the call eventually succeeds. A promise is missed when the review window closes with no attempt at all. Report each state separately so a late call is not hidden inside a broad success count.

Method and review criteria

Compute the share by dividing kept promises by all promises that were due in the window. Do not include promises whose due day falls after the window closes, because they are still open. Report the count of open promises beside the rate. That step prevents a team from improving its rate by moving hard cases into the next period. State the denominator in the note.

Split the share by campaign, by shift, and by owner type. A renewal reminder may carry a different expectation from an event invitation, and a promise taken near shift end may have less coverage. Compare the observed share with the written rule for each campaign. A mismatch is a process finding rather than proof that one caller failed. Report the count in each split so a small group is not overweighted.

Check whether the promised day was clear to the customer. A note that says next week is not the same as a promise for a named date. Record the phrase used on the call and the date the caller wrote down. If the two disagree, mark the record as ambiguous rather than counting it as kept or missed. Keep the customer's phrase in the note.

Authority and practical implications

Include the channel and the number used for the attempt. A callback that dialed the wrong number may look like a missed promise even though the owner tried on time. Record the number, the channel, and the outcome for each attempt. That detail separates a coverage failure from a data-entry error. Keep the original attempt record when a later note corrects it.

Use a census for a small queue and a documented sample for a larger one. Oversample promises made in the last hour of a shift and promises that crossed a weekend. Freeze the sample before reading the outcomes and report later additions separately. Have a second reviewer check a subsample and resolve disagreements with the written definition. The same rule should apply in the next review.

Report late promises separately from missed promises. A late call shows that coverage or priority failed, while a missing call may show that the item was lost. Name the owner of each open or late item and the date it was discovered. A queue that hides both states inside one average will look healthier than it is. Keep the two lists separate in the weekly review.

Limitations, conclusion, and sources

The result can support a change to the callback script, the coverage plan, or the owner assignment. It cannot prove that the customer was satisfied or that the issue was resolved. Pair the share with a sample of call outcomes and complaints before drawing a conclusion about service quality. Preserve the original record when a definition changes and label the transition.

Run a baseline review, then repeat after any change to the cadence, the dialer, or the shift pattern. A weekly cadence is practical for a busy calling floor, while a monthly review may suit a stable campaign. Archive the export, the coding sheet, and the notes so another reviewer can reproduce the measurement. Use the same unit and window next time so the comparison remains fair.

Sources: FCC consumer guidance on unwanted calls; FTC guidance on the Telemarketing Sales Rule; NIST Privacy Framework; U.S. Small Business Administration guidance on managing business finances. Each source was checked on November 6, 2026 and is used for general operating context, not as a local performance benchmark. Record the definition and the sample size beside the result so the review can be repeated.