Resolution Rate

Resolution rate is the percentage of customer issues or inquiries that are successfully resolved, out of the total received, over a period. It is a core measure of how effectively a support operation, human or automated, actually solves the problems customers bring to it.

Resolution is the point of support, so resolution rate gets close to measuring what matters most: did the customer's problem actually get solved? A high resolution rate means customers leave with their issue handled; a low one means problems linger, escalate, or recur.

What resolution rate measures

Resolution rate is resolved issues divided by total issues, over a period. The crucial part is defining "resolved": genuinely solving the customer's problem, not just closing a ticket. A ticket marked resolved that the customer reopens, or that leads to a repeat contact, was not truly resolved, which is why resolution rate is often paired with repeat-contact and reopen rates to keep it honest.

Related resolution metrics

First-contact resolution is an especially valued variant, solving the issue in one interaction, because it correlates strongly with both efficiency and customer satisfaction.

How resolution rate works

An issue comes in, is worked (by a human, self-service, or AI), and is either resolved or escalated/unresolved; the resolution rate is the share that reach genuine resolution.

Resolution rate sits alongside deflection rate in customer operations: deflection measures how many issues are handled without a human, resolution measures how many are actually solved (by any channel). A healthy operation wants both high, issues resolved, and increasingly resolved through efficient self-service and AI.

Why resolution rate matters

• It measures the point of support. Solving problems is the job; resolution rate measures it most directly.
• Customer satisfaction. Resolved issues, especially on first contact, drive satisfaction and loyalty.
• Efficiency. Unresolved issues create repeat contacts and escalations that cost far more.
• Retention. Customers whose problems go unsolved are at real risk of churning.

Resolution rate and AI

AI has raised the stakes for resolution rate. A customer agent that can take actions, not just answer, can genuinely resolve issues rather than deflect them, lifting resolution rate while reducing human load. But the same honesty principle applies: an AI that closes interactions without truly solving the problem inflates the metric while harming customers. Measuring resolution alongside reopen and satisfaction rates keeps automated resolution genuine.

How to calculate resolution rate

The formula is resolved issues divided by total issues received in the same period, multiplied by 100. If a support team received 1,200 issues in a month and genuinely resolved 1,020 of them, the resolution rate is 85%. The arithmetic is easy; the definitions decide whether the number means anything.

• Define resolved from the customer's side. An issue is resolved when the customer's problem is gone, not when an agent changes a status. A common rule is to count an issue as resolved only if it has not been reopened, and the customer has not contacted you again about the same problem, within a fixed window such as seven days.
• Decide what counts in the denominator. Spam, duplicates and messages that need no answer should be excluded consistently, or the rate moves for reasons unrelated to performance.
• Separate the channels. Resolution by email, chat, phone, self-service and AI agents behaves differently. A blended number hides which channel is doing the work.
• Track first-contact resolution alongside it. First-contact resolution, sometimes called first call resolution, shows how many issues were solved without the customer having to come back.

A worked example

Consider a team that reports a 92% resolution rate and still sees satisfaction falling. The illustrative breakdown tells the story:

Once reopens are removed, the real rate is closer to 79%, and fewer than six in ten customers get an answer on the first try. The team was optimizing for closing tickets quickly. The fix is not to work harder but to change the definition, then look at which issue types generate the most repeat contacts. Usually a small number of categories, such as billing questions or a confusing setup step, create most of the reopens, and fixing the underlying cause lifts resolution more than any staffing change.

How to improve resolution rate

• Give the first responder the tools to finish the job. Access to account data, permission to issue small refunds and clear escalation paths all turn handoffs into resolutions.
• Keep the knowledge base current. Outdated articles produce confident wrong answers, which show up later as reopens.
• Route by issue type. Sending technical problems straight to someone who can fix them avoids the transfer loop.
• Review unresolved and reopened cases weekly. Each one is a small diagnosis of a gap in process, product or documentation.
• Let AI act, not only answer. An agent that can check an order, reset access or update a record resolves; one that only links to an article often deflects.

Resolution rate also connects to revenue. Unsolved issues are an early churn signal, which is why teams that track customer satisfaction and churn rate by account often find the same names on both lists. Our customer retention statistics gather the external research on that link.

Common resolution rate mistakes

• Closing tickets, not solving problems. Marking issues resolved to hit the number while customers come back.
• Ignoring reopens. A resolution rate that does not account for repeat contacts overstates success.
• Speed over resolution. Optimizing for fast closes can lower genuine resolution.
• One number, no context. Resolution rate needs first-contact resolution and satisfaction alongside it to mean anything.

Resolution rate measures the heart of support, whether customer problems actually get solved. Defined as genuine resolution (not just closed tickets) and read alongside reopen and satisfaction rates, it is one of the truest gauges of whether a support operation, human or AI, is doing its job.