Workforce Research
Estimating the cost of repeat calls with public wage data
A bounded method for estimating repeat-call cost with public wage data and local call records.

Table of contents
A repeat call is often treated as a customer annoyance, not a cost. The second call takes time, and that time has a wage behind it. This review shows a bounded way to estimate the cost with public wage data and a local call sample. The result is a planning figure, not an audited cost.
Key takeaways
- Count a repeat call only when it repeats the same issue.
- Use a loaded hourly cost, not only the base wage.
- Separate the agent time from the customer time.
- Report the estimate as a range and state the assumptions.
Question and scope
This review asks how a small team can estimate the labor cost of repeat calls using public wage data and its own records. The unit is one repeat call linked to an earlier call about the same issue. The estimate covers the agent's time, not the customer's time or the business impact. The goal is a planning figure, not an accounting entry.
The scope covers inbound service calls handled by a human agent. It excludes automated calls and calls that were transferred without a new conversation. A separate method would be needed for those cases. The scope keeps the estimate comparable across teams.
The review also separates a repeat call that was caused by an incomplete first answer from one caused by a new development. Only the first kind is avoidable through better intake. A broader count may overstate the cost that the team can control. The reason codes are what keep the estimate honest.
Define the repeat window before collecting the sample. A call about the same issue within three days may be a repeat, while a call after two weeks may be a new request. State the window and apply it consistently. A clear window keeps the estimate comparable.
What public wage data can provide
The U.S. Bureau of Labor Statistics publishes occupational employment and wage data at bls.gov. The data can give a base hourly wage for customer service representatives and related roles. It is a national or regional figure, not a specific business's payroll. The estimate should say which figure was used.
The base wage is not the full cost of an hour of work. Payroll taxes, benefits, and overhead add to the employer's cost. A business can use a simple multiplier from its own payroll or a clearly stated assumption. The multiplier should be visible in the report. A low base wage can understate the real cost.
The BLS data is updated on a schedule and may not match the current market in a specific city. Use the most recent figure available and note the year. If the business has its own payroll data, use that instead. Public data is a fallback, not a substitute for the business's own numbers.
Use a loaded cost that includes the employer's payroll taxes and benefits. A base wage alone understates the cost of an hour. If the business does not have a loaded rate, use a stated multiplier and show it. The reader should be able to reproduce the calculation.
Note the source and the year of the wage figure in the report. A figure from an older year may not reflect the current market. If the business has a current payroll rate, use it and say so. The reader should know how old the data is.
Collecting the local repeat-call sample
Match calls by the caller number, the issue, and a short time window. A call about the same issue within a few days is a likely repeat. A call about a different issue is not. Record the reason for the repeat and the agent who handled the first call. The reason is what makes the estimate useful.
Measure the duration of the repeat call from the connection to the end. Do not include the wait time before the call was answered unless the analysis is about the customer's total time. State which clock was used. A clear definition prevents the estimate from drifting.
Include the time spent on any internal follow-up that the repeat call triggered, such as a note, a transfer, or a callback. The agent's time is not only the talk time. A short sample with a stopwatch or a system report is enough to start. The estimate can be refined later.
Include the time spent on the first call as well as the repeat. A repeat is often the visible part of a longer problem that started with an incomplete answer. Report the two together where the data allows. A single repeat cost can understate the total effort.
Check whether the repeat call was handled by the same team or a different one. A transfer can add time and confusion. Record the team and the handoff. A repeat that crosses teams may cost more than a simple second call.
Include the time spent waiting for a system or a supervisor during the repeat call, because that time is part of the agent's cost even when it is not talk time. A short observation sample can capture it. If the system records only talk time, state that the estimate is a lower bound. A clear limitation is better than an invisible gap.
Calculating a range
Multiply the repeat-call minutes by the loaded hourly cost and divide by sixty. Do the calculation for a low and a high wage assumption instead of a single number. Show the range and the assumptions in a small table. A range is more honest than a false precision.
Report the number of repeat calls in the sample and the share of all calls they represent. A cost estimate without the volume can look larger or smaller than it is. The reader should see both the unit cost and the total. A small unit cost across a large volume can still matter.
Separate the avoidable share from the total. If half of the repeat calls were caused by a new development, the avoidable cost is smaller. Apply the same reason codes to every call and state the share. The owner can then decide whether a process change is worth the effort.
Present the estimate as a range with the assumptions beside it. A range signals that the number is a planning tool, not an audited figure. Show the low and high wage assumptions and the repeat volume. A clear table is easier to discuss than a single number.
Show the calculation in a small table with the minutes, the loaded rate, and the result. A table is easier to check than a paragraph. Include the low and high assumptions. A transparent calculation is more useful than a polished headline.
Limitations and next action
The estimate cannot capture the customer's time, the lost goodwill, or the revenue that never arrived. It also cannot prove that a process change would remove the repeat calls. Treat the result as a starting point for a test. A conservative estimate is more credible than an inflated one.
Choose one change to test, such as a clearer first-call script or a callback that closes the loop. Record the date and measure the same repeat-call reasons after the change. Do not change the script and the staffing at the same time. A controlled test gives the team a better answer.
Share the estimate with the team that answers the phones and ask whether the reasons match their experience. A number that the team does not recognize will not lead to action. A short conversation can reveal a cause that the data missed. The estimate is a tool for the conversation, not a verdict.
Ask the team whether the repeat reasons match their experience. A reason code that the team does not recognize may be too broad. A short conversation can improve the categories before the next sample. The estimate should support a better conversation, not replace it.
Separate the repeat calls that happened on the same day from those that happened later in the week. A same-day repeat may reflect a misunderstanding that could be fixed with a clearer close. A later repeat may reflect a problem that was not resolved at all. Report the timing. The two patterns suggest different changes and different costs. A simple split makes the estimate more useful to the team that will act on it.