Blog · Board and management reporting · Customer service
Contact center analytics compares demand, handling workload, repeat contacts and cost across channels. Work through exported contact and case summaries, with checks that keep service and resolution measures distinct.
Contact center analytics compares recorded demand, service, handling workload and cost across channels and teams. Use exported contact summaries to understand the work performed, then connect them to a separate case register to check resolution and backlog. A useful report explains a decision with traceable counts rather than presenting a long list of percentages.
Suppose a US business-services company wants to know why its phone and email teams have similar average handling time but different costs per contact. The operations manager owns the comparison. Team leads confirm contact classifications, and finance confirms which expenses belong to each channel.
Start with a weekly contact export containing contact ID, case ID, channel, team, issue category, handling duration and service timestamps. Add a case movement summary and a cost allocation sheet for the same week. Remove customer names and message contents when they are unnecessary for the calculation.
For the wider measurement catalog, use contact center KPIs. This analysis follows the narrower chain from recorded demand to workload, cost and an assigned investigation.
A contact is an interaction. A case is the underlying issue, which may require several contacts and may cross channels. Keep an organization-wide case ID when comparing repeat work so an email after a phone call can be recognized as part of the same issue.
For this report, a repeat is a handled contact after the first contact on the same case within an agreed observation window. It is an interaction measure, not a count of customers who returned. The repeat-contact definition explains why the underlying issue and window matter.
Handling workload includes recorded interaction, hold and after-contact work. Zendesk's discussion of average handle time describes those components for phone calls and emphasizes that channel formulas differ. Here, email time means recorded active labor and follow-up work; waiting for a customer reply is excluded.
The weekly source summary is:
| Recorded measure | Phone team | Email team | Total |
|---|---|---|---|
| Offered contacts | 500 | 300 | 800 |
| Handled contacts | 450 | 240 | 690 |
| Interaction labor minutes | 2,250 | 1,680 | 3,930 |
| Hold minutes included in workload | 450 | 0 | 450 |
| After-contact work minutes | 900 | 240 | 1,140 |
| Repeat handled contacts | 90 | 24 | 114 |
| Allocated operating cost (USD) | 2,250 | 1,440 | 3,690 |
Offered counts represent contacts entering the respective channel's defined intake population. They are not unique cases. Handled contacts can include work received earlier, so offered minus handled is not a general backlog formula.
Use totals from the same handled-contact population. All handling fields are complete in this example, and the cost allocation includes the same agreed operating expense categories for both teams.
Handling workload = Interaction minutes + Hold minutes + After-contact work minutes
Average handle time = Handling workload / Handled contacts
Repeat-contact share = Repeat handled contacts / Handled contacts
Cost per handled contact = Allocated operating cost / Handled contacts
Phone workload is 2,250 + 450 + 900 = 3,600 minutes, or 60 hours. Email workload is 1,680 + 0 + 240 = 1,920 minutes, or 32 hours. Divide minutes by 60 to convert them to hours.
| Computed measure | Phone team | Email team | Combined |
|---|---|---|---|
| Handling workload, minutes | 3,600 | 1,920 | 5,520 |
| Handling workload, hours | 60 | 32 | 92 |
| Average handle time, minutes | 8.0 | 8.0 | 8.0 |
| Repeat-contact share | 20.0% | 10.0% | 16.5% |
| Cost per handled contact (USD) | 5.00 | 6.00 | 5.35 |
For phone, 3,600 / 450 = 8.0 minutes and 90 / 450 = 20.0% repeats. Email also averages 1,920 / 240 = 8.0 minutes, but only 24 / 240 = 10.0% of handled contacts are repeats.
Combined cost is 3,690 / 690 = $5.35 per handled contact. Combined repeat share is 114 / 690 = 16.5%. Money is rounded to two decimals and percentages and average minutes to one; calculations use unrounded inputs.
Equal average handling time does not imply equal cost. Phone has $2,250 / 60 = $37.50 of allocated operating cost per recorded handling hour; email has $1,440 / 32 = $45.00. Those are expense-allocation ratios, not hourly pay rates. Finance should inspect the allocation and nonhandling paid time before attributing the difference to agent productivity.
Reconcile the channel costs to the $3,690 expense allocation, and the handled count to 450 + 240 = 690 source records after duplicate contact IDs are resolved. The 92 workload hours must equal 60 + 32. Do not interpret them as total paid hours: training, breaks and other time are outside this handling calculation.
Use the service rule the customer actually experiences. The example company defines phone service as an answer within 60 seconds and email service as a first response within one business day. Each numerator follows the intake cohort used in its denominator. Measure the result after that cohort's response window closes.
| Channel | Eligible offered contacts | Met the stated target | Service calculation |
|---|---|---|---|
| Phone | 500 | 360 | 360 / 500 = 72.0% |
| 300 | 210 | 210 / 300 = 70.0% |
These percentages describe different service promises. Do not average them into a claim that the channels deliver equivalent service. Check the export's actual definition: AWS's contact-metric reference distinguishes handled contacts, abandoned contacts and service-level calculations, illustrating why a label alone is insufficient.
Covirage is not affiliated with Zendesk or Amazon Web Services.
Use distinct cases for the backlog movement. Assign each case to a stable reporting team for this example; there are no transfers, reopenings or other adjustments during the week.
| Team | Opening cases | New cases | Resolved cases | Closing cases |
|---|---|---|---|---|
| Phone | 40 | 300 | 310 | 30 |
| 20 | 180 | 150 | 50 | |
| Total | 60 | 480 | 460 | 80 |
The control is opening + new − resolved = closing: 60 + 480 − 460 = 80. Phone backlog falls while email backlog rises. Resolved cases can include opening cases, so resolved divided by new cases is not a first-contact resolution rate.
The phone lead should investigate repeat issues. The email lead should examine the aging and complexity of unresolved cases, then compare active work with available capacity. Use support backlog recovery capacity for the separate question of clearing an accumulated queue.
Bring your contact summary, case movement register and channel cost sheet for the same period. Covirage maps the columns, and its tools compute and check workload, resolution and cost comparisons; the external AI model explains the results with every figure traced to its rows. Create a free Covirage account and bring a sample of your own exports.
See Covirage for customer service teams for the business context.
A KPI list defines measures. Analysis connects the measures to a decision, such as why one team has more repeat contacts or why handling cost rose. It needs consistent source records, comparisons by channel or issue, and checks that establish whether the differences are real.
Compare them only after documenting what counts as a contact and how handling time is recorded. Phone elapsed time and email labor time are different. Use channel-specific service targets, and compare costs with the same expense boundary rather than assuming every interaction requires equivalent effort.
Not automatically. Repeat-contact share often counts interactions, while first-contact resolution follows cases or customer issues over an observation window. Different denominators and incomplete follow-up can prevent the percentages from being complements. Define and measure each population separately.
Report the missing population separately and calculate averages only for records with sufficient time information. Do not convert an unknown duration to zero. Explain the coverage alongside the average so a channel with weak recording does not appear artificially efficient.