Service desk KPIs for owners: cost per call and wait time

e-automate provides service statistics, but cost per call and queue wait need separate reading. This guide defines cost per call and wait time using the cited industry sources.

MetricNet’s 2012 paper asks whether a lower cost per contact is good when abandonment rises at the same time. In our view, the answer depends on reading the two numbers together, so we start with a pair.

Why these two

MetricNet’s 2012 paper says cost per contact is the most effective cost metric for a service desk. Cost per contact measures what the desk spends for each contact. Wait time measures one part of the service experience. We pair them because cost alone does not describe queue performance.

Cost per call: count the whole desk

ICMI’s 2003 call center metrics tutorial defines cost per call as total costs divided by total calls for a given period. MetricNet’s average cost per contact uses a different denominator: all inbound contacts, IVR-contained ones too. Its operating expense counts salaries, overtime, benefits and incentive pay, plus contractors, facilities, telecom, software, training, travel, office supplies and desktop computing.

In a 2012 MetricNet analysis, labor-related costs (salaries, benefits, incentive pay and contractors) made up 67% of all costs for the average service desk. That’s an old average, so use it to decide where to look first, not as a target.

Two habits make the figure useful. Work it out the same way monthly, using the same list of costs, so a change means something. ICMI also recommends that cost per call be differentiated by each channel of contact. Phone, email and chat shouldn’t share one average without anyone noticing.

ICMI notes that a climbing cost per call can be a good sign, depending on the variables driving it up. Fewer calls can spread fixed costs over a smaller volume and push cost per call up. So investigate the reason for an increase before deciding whether it reflects poorer performance.

e-automate’s Service Statistics report lists average labor cost per call, average parts cost per call and average response time, broken out by rep, branch or company. Do not assume those two fields equal total desk operating cost, so verify which costs the report includes before using it. The report only covers calls with at least one labor record, so before you divide by a call count, verify that the call population in the report matches the base in your own calculation.

Wait time: track average speed of answer and service level

HDI’s 2023 article on IT service center metrics defines average speed of answer as the average time a customer waits in a queue, by voice or chat, before being connected to a Service Center analyst. ICMI gives the formula: total delay divided by total number of calls, drawn from the same data as service level.

In our view, interval-level reporting makes changes easier to see. ICMI says service level should be calculated for each reporting interval, not averaged across days, weeks or months. ICMI defines service level as the percentage of contacts answered within a set number of seconds.

Abandonment counts calls dropped before anyone answers. HDI defines the call abandonment rate that way, and notes that a high rate could indicate the team is understaffed or unable to handle the call volume, which ultimately reduces customer satisfaction.

Because people directly experience queue delays, include wait time on the owner’s dashboard.

Read the two numbers together

Review any cost change alongside average speed of answer, service level, abandonment and channel mix before drawing a conclusion.

HDI says the abandonment standard depends on the service being provided. We therefore do not prescribe a universal target here. We recommend using your own recent trend as a baseline when no comparable external target is available.

Set up the dashboard

1. Total the desk’s annual operating costs, including the labor, telecom, software and facility costs on MetricNet’s list, and divide by annual inbound contacts, including IVR-contained contacts, following MetricNet’s definition.
2. Split that cost per contact by channel, such as phone, email, chat or another separately measured contact channel.
3. Record average speed of answer: total delay divided by the total number of calls.
4. Record service level as the share of contacts answered within a set number of seconds, for each reporting interval you choose.
5. Record abandonment: the share of calls disconnected before anyone answers.
6. Put cost per contact, average speed of answer, service level and abandonment side by side for the same period, then compare them with the previous period.

Our guide, Build vs buy: how to add AI agents to your team, is one place to start.

Sources

• ICMI, Call Center Metrics: Key Performance Indicators (KPIs), tutorial, copyright 2003.
• MetricNet, MetricNet’s Seven Most Important KPI’s for the Service Desk, 2012 (file dated January 2012).
• MetricNet, Introduction to IT Service Desk Metrics: Definitions and Key Correlations, page updated October 19, 2026.
• HDI, Michael Hanson, The Metrics That are Valuable to IT Service Centers, March 15, 2023.
• ECI, e-automate online help (version 250): Service Statistics.
• Auralis customer results (internal, anonymized).