Customer Support Resolution Cost per Case

A support team spends money on people, software, outsourcing and management. Dividing total spend by every incoming message may make handling look cheap even when customers must contact the team repeatedly.

Customer support resolution cost per case focuses on issues actually resolved within a defined period, while keeping unfinished work visible. Define a case, since one customer issue can produce several emails, chats and calls, and link those contacts under one case identifier when they concern the same underlying need.

Define resolution too, because a closed ticket is not necessarily a solved problem, so establish whether the customer outcome is verified, accepted or left unreopened for a suitable period. Set the period so that the same dates apply to costs and resolved cases, describing how work begun earlier and completed later is treated.

Choose the cost scope, which may include staff compensation, outsourced fees, software licences, facilities and management overhead, and list the components for fair comparisons. Distinguish variable cost, since some costs change with each case while rent and platform subscriptions may not, so the average is not the incremental cost of one extra case.

Allocate shared resources with an allocation method rather than charging every salary to support, and include escalation effort by deciding whether engineering or finance time spent resolving a case enters the numerator. Handle work in progress, because a month with many newly opened cases but few completions can make cost per resolved case look high, so report backlog alongside the ratio.

Avoid artificial closures, since agents who close tickets merely to improve the denominator can make repeat contacts and customer satisfaction worse, and define whether a reopened case remains one case or is a new one after a stated interval, applying the same rule over time. Reconcile volume by counting unique resolved cases from the case system, removing duplicate records and test tickets consistently.

Track channel mix, because a simple self-service question and a complex phone escalation use different resources, and check case severity, since a safety or compliance issue can legitimately cost more to solve and a lower cost is not automatically a better outcome. Time-tracking can estimate case-level cost, but missing or inconsistent logs can bias comparisons between teams, and an outsourced provider may charge per contact, seat or outcome, so translate charges under the contract rather than assuming each ticket incurs a fixed fee.

Exclude unrelated refunds, because a goodwill refund may be a recovery cost, not an operating cost of handling the case, and present both if useful while avoiding inconsistent inclusion. APQC measures average cost per inbound contact, which is a useful but different denominator and is not interchangeable with cost per resolved customer issue, while Zendesk describes cost per resolution as spend tied to a fully solved issue rather than merely a ticket closure.

Pair the number with repeat contact and satisfaction, and watch automation effects, since a chatbot that diverts easy questions may leave human agents with harder cases, so a rising human-case average could reflect the new mix, not worse efficiency. Compare like cohorts, because B2B technical cases, routine consumer billing questions and regulated complaints have different complexity, and use trends to find repetitive issues that a clearer product or process could prevent, then check whether both cost and customer outcome improve.