First contact resolution formula, with examples

The basic formula divides the problems solved in the first contact by all the problems customers raised, then multiplies by 100. If 340 of the 400 customers who wrote to you last month never had to write again about the same thing, your first contact resolution was 85%.

Call centres call it first call resolution, and SQM Group, which benchmarks call centres, says the two names are “commonly used interchangeably”. Zendesk counts one-touch resolutions instead, and Intercom counts conversations closed on first contact.

For a founder answering support alone, every problem that needs a second message costs you twice. Deacon is an AI customer support agent that answers your customers from your own docs in seconds, and asks one question back when a question is too vague to answer. It has a free plan.

The first contact resolution formula, four ways

The versions differ in what they divide by. MetricNet, which benchmarks IT service desks, sets out two. Gross FCR divides by every contact. Net FCR leaves out contacts that could never be solved in one go, such as a hardware repair, and those left out are called carve-outs. MetricNet calls net FCR “by far the more relevant”, and says most organisations don’t track gross FCR at all.

ContactBabel’s guide to first-contact resolution, from an analyst firm that studies contact centres, adds two methods that count only new problems. Here are all four on a month at a small software company. It got 500 support emails, 450 about a new problem and 50 following up on one not yet solved. Of the new problems, 20 couldn’t have been solved in one reply, such as a bug that needed a fix. The reply they got settled 360 of the emails.

One month reads anywhere from 69% to 80%. ContactBabel says there’s no single correct method, and that what matters is picking one that fits your data and sticking to it. The last two give you one chance per problem, which is how a customer sees it.

How to tell whether a problem was solved first time

The formula is the easy part. Deciding what counts as solved first time takes one of four approaches, and they don’t agree.

Ask the customer. SQM’s survey counts a call as resolved first time only when the customer answers yes to “Was your call resolved?”, one call to “How many calls did you make to resolve your call?” and no to “Did you use another contact channel after you used the call center?” Measures that don’t ask the customer come out 10 to 20% higher, partly because some customers who weren’t helped never call back, or try another channel instead. CSAT covers when a survey like this goes out and who answers it.

Watch for a repeat contact. If the same customer gets in touch about the same problem within a set window, the first contact failed. Genesys’s report uses 7 days. SQM puts the window anywhere from 1 to 30 days and suggests setting it by the reason for the call, such as 30 days for a claim and 5 for a status question. Every repeat contact is work for the customer, which is what a customer effort score asks them about.

Count replies in your helpdesk. Zendesk’s one-touch resolution is a ticket solved with a single reply from your team, or none, and a ticket the customer reopens stops counting once your team replies again. Freshdesk counts a ticket only if the customer sent nothing after their first message, and divides by resolved and closed tickets, leaving reopened ones out. Intercom counts a conversation closed once with one teammate replying, and a close by a bot doesn’t count. For email, MetricNet’s guide says a resolution within one business hour of the email arriving counts as FCR.

Let the person who took it decide. This is the weakest. NiCE’s glossary calls it “the simplest but least accurate” method, because “agents are motivated to declare resolution to hit performance metrics even when the underlying issue may persist”.

What is a good first contact resolution rate?

On the phone, 70 to 79% is what SQM calls good. Its current call centre average is 71%, and it calls 80% or more world class, a level 5% of call centres reach. Tickets and chats look very different. At the median company on Freshworks’ products, 18% of tickets were resolved in the first interaction, against 83% of chat conversations. These are the published figures we found with a named sample, read on 23 September 2026.

By industry, SQM’s 2026 chart puts retail on top at 77% and tech support at 64%. Every centre counts FCR its own way, so ContactBabel calls it “far more suitable for use internally, rather than comparing it against external benchmarks”. Compare this month with your last few.

When a high first contact resolution rate is a bad sign

Customers who keep asking something simple, such as when their subscription renews, push FCR up, because those questions are easy to solve first time. Zendesk’s guide to FCR says a high rate “could indicate that your self-service resources are lacking”.

The reverse happens when self-service works. ContactBabel found the UK trend “declining very slightly” as “the easier interactions go to self-service”, which left contact centres the harder work. If a chatbot starts answering your simple questions, the ones that still reach you are harder to settle in one reply. Your rate can fall while customers get more help, so read it beside how many questions reach you at all.

First contact resolution vs resolution rate and first response time

The gap between the first two can be wide. In Freshworks’ 2024 data, the median company resolved almost all of its tickets in the end, and 18% of them in the first interaction. NiCE’s glossary adds that a high FCR can sit beside a low resolution rate, when first fixes don’t hold and customers call back about the same issue.

Solving more questions first time with Deacon

Deacon answers in the conversation where the question is asked, in seconds, from your own docs and in the customer’s language. When a question is too vague to answer well, it asks one question back instead of guessing at what they meant. If three clarifying questions haven’t settled it, Deacon stops asking, sums up what the visitor said in one sentence and asks for their email, so you can reply without asking them to start over.

Your site can also tell Deacon who’s asking, such as the customer’s plan, and Deacon takes that into account when it answers. An answer that fits their plan is less likely to bring them back.

When your docs don’t cover a question, Deacon says it doesn’t know instead of guessing, and asks for the visitor’s email. That matters here, because a confident wrong answer looks like a first contact resolution until the customer comes back. Every question your docs didn’t cover has an Answer this button. Write the answer once and it becomes part of what Deacon knows, so the next person who asks can get it the first time, and the more you answer, the less Deacon needs you.

It also shows you where repeat questions come from. Topics group similar questions, count how many people asked and label most groups as a how-to, a bug report, a feature ask and so on. A bug report that keeps coming back is the fix that will stop the repeat contacts.

Deacon doesn’t know whether a problem stayed solved, so it has no first contact resolution rate to give you. A conversation closes after twenty quiet minutes, so a customer who comes back later about the same thing starts a new one, and each conversation shows whether the visitor has chatted before. Filter by Unanswered and Wants a person to find the conversations that didn’t go well, and try it all on the free plan, which needs no card.