First Call Resolution Benchmarks: What Is a Good FCR Rate?

First call resolution (FCR) is the percentage of customer issues resolved on the first contact, with no callback, second channel or follow-up needed. In SQM Group's 2024 benchmark of more than 500 North American call centers, the average FCR across all industries was 69%, and individual centers ranged from 43% to 88%.

Those numbers are only useful if you measure FCR the same way. This page covers the formula, the published benchmarks by industry and call type, and how FCR changes once an AI voice agent answers part of your calls.

How do you calculate first call resolution?

Divide the number of issues resolved with no repeat contact by the number of unique issues customers raised, then multiply by 100. The hard part is defining "no repeat contact": you have to pick a callback window and decide what counts as the same issue.

First call resolution formula: FCR = (Issues resolved with no repeat contact in the window ÷ Unique issues raised) × 100

10,000 first calls in a month; 3,000 of those customers call back about the same issue within 7 days. 7,000 resolved ÷ 10,000 unique issues × 100 = 70% FCR.

Pick the callback window per call reason before you calculate, and keep it fixed when you compare periods.

There are two ways to collect the inputs, and SQM's guide to calculating FCR describes both:

• Internal FCR uses your own systems (CRM, ACD, IVR, call records) to check whether the same customer called back about the same issue within a window, typically somewhere between 1 and 30 days. SQM suggests different windows by call reason: a claim that needs investigation may need 30 days, a status inquiry perhaps 5.
• External FCR asks the customer in a post-call survey. SQM counts a call as resolved on first contact only if the customer says it was resolved, took one call, and did not need another channel.

What is a good first call resolution rate?

By SQM's standard, a good FCR rate is 70% to 79%, and world-class is 80% or higher. In SQM's breakdown of the centers it benchmarks, only 5% reach world-class, 46% sit in the good range and 49% fall below 70%.

FCR is worth chasing because it moves the other numbers. SQM reports that every 1% improvement in FCR raises the interactional Net Promoter Score by 1.4 points for the average call center. Every call that is not resolved the first time also comes back as a second call, which adds to volume and handle time. That is why FCR belongs next to average handle time on any operations report.

How does first call resolution vary by industry and call type?

Part of the industry gap is the mix of calls each industry gets. In SQM's 2024 results by industry, retail, not-for-profit and insurance call centers led with an average FCR of 73% to 75%. Energy, health insurance and financial services sat at a moderate level, while technical support, government and telecommunications remained in the "needs improvement" range. SQM notes that no telecommunications company has reached world-class FCR in its 25-plus years of benchmarking.

The bigger spread is by call type, which is the comparison to use when you set targets:

• General inquiries: 2024 average FCR: 73%. What drives it: Answer usually known on the call.
• Account maintenance: 2024 average FCR: 72%. What drives it: Agent can make the change directly.
• Orders: 2024 average FCR: 71%. What drives it: Depends on stock and order system access.
• Billing: 2024 average FCR: 69%. What drives it: Disputes often need a second team.
• Claims: 2024 average FCR: 61%. What drives it: Investigation happens after the call.
• Technical support: 2024 average FCR: 60%. What drives it: Fixes may need testing or a visit.
• Complaints: 2024 average FCR: 48%. What drives it: Often need authority the agent lacks.

If 40% of your calls are complaints and technical support, your blended FCR will sit below a center whose calls are mostly general inquiries, even if your agents are better. Benchmark each call reason against its own row, not your total against the industry average.

Why does your FCR rate not match the benchmark?

Usually because the measurement differs, not the performance. SQM finds that internal FCR measurement tends to run 10% to 20% higher than external, survey-based measurement, mostly because of how organizations define a repeat call. A customer who calls back from a different number, or about a related problem, can slip past an internal check.

Compare your FCR only with benchmarks measured the same way. SQM finds internal, system-based FCR tends to run 10% to 20% higher than survey-based FCR. Match the method and the callback window before you compare.

Two other choices shift the number. A shorter callback window makes FCR look better, and excluding hard call types (transfers, escalations, complaints) from the denominator makes it look much better. Write your definition down before you compare it with anyone.

How does an AI voice agent change first call resolution?

From the customer's side, nothing about the definition changes: one contact, problem solved. What changes is who resolves it and how you track it.

• Transfers inside the same call. If the AI agent hands the caller to a person who resolves the issue on that call, the customer made one contact. Decide up front whether your FCR counts that as resolved, and report AI-only resolution separately.
• Callbacks across channels and agents. A caller who talks to the AI agent on Monday and a human on Wednesday is a repeat contact. Your FCR calculation has to join AI and human call records by customer, or it will overstate both.
• After-hours calls. Calls that used to end in a voicemail and a next-day callback can now be resolved on the first call, which lifts FCR for that segment.

A related metric is how often the AI agent finishes calls without a person at all, covered in the guide to measuring call containment. Containment without resolution is not a win. For a published example of resolution on real calls, Anker reported an 80.4% case resolution rate for Tier-1 support calls handled by a Retell AI agent, against a pilot target of 80%. That is one company's resolution measure, not an industry FCR benchmark.

How do you measure FCR for voice agent calls in Retell?

Retell does not ship a single FCR number, because your callback window and issue definition are yours to set. It gives you the parts to build one:

1. Define resolution per call. Under Post call extraction in the agent editor, add a boolean field such as issue_resolved and a selector such as call_reason. You can also customize the prompt behind the built-in call_successful field. Retell fills these from the transcript after every call, as described in the post-call extraction documentation.
2. Find repeat callers. The Contacts page under Data keeps one record per phone number with a count of related conversations and the latest conversation time. For an exact 7-day or 30-day window, pull call records through the list calls API and match on phone number and timestamp.
3. Chart it. In Analytics, chart Call Counts broken down by call_reason and issue_resolved to see resolution by call type.
4. Find the gaps. AI quality assurance reports a call resolution rate and a table of the questions callers ask most, each with its own resolution rate.

How do you improve first call resolution?

Start with the causes. SQM's 2024 analysis of non-FCR calls attributed 49% of errors to the organization (policies, processes, systems), 38% to the agent and 13% to the customer. The organizational fixes take longer but pay back across every call. For both human and AI agents, the most common levers are:

• Fill knowledge gaps in the order of the lowest-resolving questions.
• Give the agent the ability to act (update the booking, issue the refund) instead of only taking a message.
• Carry context forward. Retell's contact memory lets an agent pick up where the last call ended, so a caller who rings back does not start over.
• Transfer with context when a person is needed, so the human resolves it on the same call. Retell's warm transfer can brief the receiving agent before the caller joins.

The surest way to know what FCR an AI agent will reach on your calls is to measure it on your calls. Retell runs pilots on real call traffic, so you can compare resolution by call reason against your current baseline before you commit. Talk to the Retell team about a pilot.