First Call Resolution (FCR): The Formula, How to Measure It Honestly, and How to Fix a Bad Number
Updated for 2026.
First call resolution (FCR) is the percentage of customer issues fully resolved during the first contact, with no callback, transfer, or follow-up needed to close them out. The formula is: FCR = (issues resolved on first contact / total issues) x 100.
That definition is easy to state and surprisingly hard to measure. Most contact centers that report an FCR number are reporting something closer to “calls the agent believed they finished,” which is a different metric and a much friendlier one. This guide covers the formula, what actually counts as a resolution, the four ways teams measure FCR ranked by how much you can trust them, why FCR and average handle time pull against each other, and what to do when your number is bad.
The FCR Formula
The math is simple. The definitions inside it are where the work is.
FCR = (issues resolved on first contact / total issues in the period) x 100A worked example. Your support line took 2,400 calls last month. You count unique issues, not calls, so you first collapse repeat contacts about the same problem into one issue each. That leaves 2,050 distinct issues. Of those, 1,517 never generated a second contact within your measurement window.
FCR = (1,517 / 2,050) x 100 = 74%
Two things in that example do all the heavy lifting, and they are the two things most teams skip:
The denominator is issues, not calls. If you divide resolved calls by total calls, every repeat contact inflates the denominator and usually counts as its own “resolved” call, which quietly pushes your number up when service gets worse. Count the problem once, then ask whether it took one contact or four.
The measurement window has to be stated. “No second contact” means nothing without a time bound. Most teams use 7 days. Shorter windows (24 or 48 hours) flatter the number by missing slow-burn repeats like a billing correction the customer does not notice until the next statement. Longer windows (30 days) start catching genuinely unrelated new issues from the same customer. Seven days is a reasonable default; the important part is that you pick one, write it down, and never change it quietly, because changing the window changes the metric and makes your trend line a lie.
What Actually Counts as a “Resolution”
This is the argument every QA program eventually has. Write the ruling down before you start reporting, because your FCR number is only as meaningful as this definition.
A contact counts as resolved on first contact when the customer’s underlying need was fully met during that contact, and nothing further was required from the customer or from you.
Some specific cases and the honest ruling on each:
| Situation | Counts as FCR? | Why |
|---|---|---|
| Agent answered the question completely, call ended | Yes | The need was met |
| Agent transferred to a specialist who resolved it, one continuous call | Usually yes | The customer contacted you once; count it, but track transfers separately |
| Agent promised a callback and delivered it | No | Resolution required a second contact |
| Issue resolved, customer calls back about a different issue | Yes | Different issue, different denominator entry |
| Agent said “you should hear back in 3 to 5 days” | No | The issue is open, not resolved |
| Customer called, then opened a chat about the same thing an hour later | No | Cross-channel repeats are repeats |
| Correct answer given, customer did not like the answer, no repeat contact | Yes | FCR measures resolution, not agreement |
| Agent closed the ticket, customer calls back in 6 days | No | Inside the window, it is a repeat |
The cross-channel row is the one most teams get wrong. If your FCR is computed from phone data alone and your customers can also email, chat, or open a portal ticket, your number is structurally overstated. Every deflection to another channel looks like a resolved call. If you cannot unify channels yet, at least say so when you report the number, so nobody makes a staffing decision on a metric that is measuring the wrong thing.
Why FCR Matters More Than Most Metrics
FCR is worth the measurement trouble because it is one of the few metrics that moves cost, satisfaction, and agent retention in the same direction. Most operational metrics force a trade. This one mostly does not.
It compounds on cost. Every repeat contact is a call you pay for twice: the same queue time, the same agent minutes, the same wrap-up, plus the extra handling of a customer who is now annoyed. Repeat contacts are the cheapest volume to eliminate because that volume should never have existed. If 26% of your issues take more than one contact, a meaningful share of your total call volume is rework, and rework is the only volume you can cut without cutting service.
It predicts satisfaction better than politeness does. Customers rarely rate a contact well when they have to explain the same problem a second time, no matter how warm the agent was. Repeat contact resets the customer’s effort meter to zero and makes the second agent inherit the first agent’s frustration. A high FCR is largely why teams with strong scores on empathy criteria also tend to post strong CSAT: the empathy is real, but the resolution is what the customer is actually rating.
It reads as a systems diagnostic, not an agent grade. This is the part most teams miss. A low FCR is usually not a people problem. It is a knowledge base that is out of date, an authority limit that forces escalation for a $30 credit, a fulfillment process that generates predictable follow-ups, or a product defect quietly manufacturing calls. FCR is the metric that surfaces those, because they all show up as the same symptom: the customer had to call back.
It shows up in attrition. Agents who cannot resolve issues absorb the frustration of customers on their second and third attempt while lacking the tools to fix anything. That is one of the most reliable burnout patterns in a contact center, and it is fixed by changing what agents are permitted and equipped to do, not by coaching them to sound calmer.
FCR belongs alongside the other core call center metrics rather than replacing them, but if you can only trend one number weekly, this is a strong candidate.
The Four Ways to Measure FCR, Ranked by Honesty
Not all FCR numbers are the same number. Here is what each method actually measures.
| Method | What it measures | Trustworthiness |
|---|---|---|
| Agent self-report (disposition code) | Whether the agent thought they were done | Low |
| Post-call survey question | Whether the customer thought they were done, at that moment | Medium |
| Repeat-contact analysis in phone/CRM data | Whether the customer came back within the window | High for phone-only, blind to cross-channel |
| Transcript analysis of every call | What was actually said, resolved, promised, or deferred | Highest, requires scoring every call |
Agent self-report is the most common method and the least reliable. The agent codes the call “resolved” at the moment they hang up, which is precisely the moment they have the least information. They do not know whether the fix held, whether the credit posted, or whether the customer understood. Add even mild pressure on an FCR target and the disposition code stops describing reality at all.
Post-call surveys (“Was your issue fully resolved today?”) are better because the customer answers, but they carry heavy response bias. The customers most likely to respond are the delighted and the furious, and the survey still fires before the follow-up problem appears.
Repeat-contact analysis is the strongest structural method: pull contacts by customer identifier, group by issue, and check for a second contact within the window. It is objective and it is retrospective, which means it catches the repeats the agent could not have predicted. Its weakness is matching. Contacts have to be tied to the same customer and the same underlying issue, which requires either decent CRM discipline or intent detection on the contacts themselves.
Transcript analysis closes the last gap by looking inside the call. A repeat-contact model tells you an issue took three calls. Transcript analysis tells you why: the first agent gave an answer that was wrong, the second promised a callback that never happened, the third finally had the authority to issue the credit. That is the difference between a metric and a fix.
Why Self-Reported FCR Is Almost Always Inflated
If your dashboard says 89% and your customers seem to disagree, this section is the reason. Four mechanisms push self-reported FCR up, all of them without anyone lying:
- The agent grades their own work at the least informed moment. Resolution is confirmed later, by the customer, or not at all.
- The target changes the behavior it measures. Put FCR on a scorecard as a self-coded field and you have asked agents to score themselves on a number that affects them. Disposition codes drift toward “resolved” within weeks.
- Ambiguity resolves in the optimistic direction. “I told them to try it and call back if it does not work” gets coded as resolved. So does “I sent it to billing.” Neither is.
- Cross-channel and repeat-caller blindness. The follow-up arrives by email, or from the customer’s spouse, or from a different phone number, and the join never happens.
The result is a number that looks stable in the high 80s while actual first contact resolution sits well below it. The tell is a persistent gap between FCR and CSAT, or a repeat-caller rate in your phone data that does not square with the reported FCR. When those two disagree, believe the phone data.
FCR and AHT Pull Against Each Other
This is the trade-off nobody puts on the dashboard, and it explains most self-inflicted FCR problems.
Resolving an issue on the first contact usually takes longer on that contact. The agent asks the extra clarifying question, checks whether the credit actually posted, walks the customer through the setting instead of emailing a link, confirms the fix worked before closing. All of that raises average handle time.
Push AHT down hard and agents optimize the only way they can: wrap up faster, promise a callback, transfer instead of learning, get off the call. FCR falls, repeat volume rises, and total handle time across the issue goes up even though the per-call number improved. You made the reported metric better and the operation worse.
A worked version of that trade:
- Scenario A: AHT 6:30, FCR 78%. For 1,000 issues, that is roughly 1,220 contacts and about 7,930 minutes.
- Scenario B: AHT cut to 5:15, FCR falls to 62%. Now 1,000 issues generate roughly 1,610 contacts and about 8,450 minutes.
The faster team spent more total minutes and delivered a worse experience. This is why AHT should never be a standalone target. Pair it with FCR and watch the combination, or watch total handle time per resolved issue, which is the number that actually reflects cost:
Minutes per resolved issue = total handle minutes / issues resolvedThe one legitimate way to improve both at once is to remove work rather than rush it: better knowledge at the agent’s fingertips, wider resolution authority, and fixing the upstream defects generating the calls. Those shorten calls and prevent repeats.
The Failure Modes That Fake a Good FCR
Once FCR becomes a target, these appear. Watch for all five:
- Premature closure. “That should take care of it, call back if not.” Resolution declared, nothing verified.
- Callback laundering. The agent schedules the follow-up themselves, so the second contact is outbound and never counts as a repeat.
- Channel deflection. “Just email support and they will sort it out.” The repeat leaves the phone system, and so does your visibility into it. This shades into call avoidance when it becomes habitual.
- Denominator gardening. Short measurement windows, dropping calls under 60 seconds, excluding transfers. Each exclusion is defensible on its own and the stack of them is not.
- Disposition drift. The “resolved” code becomes the default because it is first in the dropdown and nobody audits it.
The common thread: every one of these is invisible in aggregate reporting and obvious in a transcript. That is the argument for measuring FCR from what was said, not from what was coded.
How to Fix a Bad FCR Number
A low FCR is a symptom. Diagnose before you coach, because coaching an agent about a policy limit they cannot change is the fastest way to lose their trust.
Start by pulling your repeat contacts, not your resolved ones. Take the last 100 issues that required two or more contacts, read what happened on the first one, and sort each into a bucket:
| Root cause | What it looks like | The actual fix |
|---|---|---|
| Knowledge gap | Agent gave an incomplete or wrong answer | Fix the knowledge base article, not the agent. Then coach. |
| Authority gap | Agent had to escalate a small credit, waiver, or exception | Raise the resolution authority limit and define its bounds |
| Process gap | Resolution genuinely requires another team or a wait | Redesign the handoff, or set expectations so precisely no callback is needed |
| Systems gap | Agent could not see order, billing, or account history | Fix the data access; no amount of coaching substitutes for visibility |
| Expectation gap | It was resolved, the customer did not believe it | Change the closing: confirm, restate, tell them what happens next and when |
| Upstream defect | Same product or billing issue generating repeat volume | Route it to the team that owns the defect with the call evidence attached |
Most teams find that the first three buckets hold the majority of repeats, and that only one of those three is a coaching problem. That is the whole reason root cause analysis belongs in an FCR review rather than a straight leaderboard.
Then make FCR coachable at the call level by putting observable criteria on your scorecard:
- Did the agent confirm the issue is fully resolved before closing?
- Did they verify the fix rather than assume it, where verification was possible?
- If follow-up was genuinely required, did they state the specific action, owner, and timeframe?
- Did they proactively address the obvious next question, the one that generates the callback?
- Did they resolve within their authority instead of escalating by reflex?
Those are behaviors a reviewer can point at in a recording, which is what makes them coachable. If you need a starting structure, the call center scorecard templates already weight first contact resolution as the heaviest category in the inbound support template. And before you compare agents on any of it, run a calibration session, because two reviewers who disagree about what “resolved” means will produce an FCR number that measures the reviewer.
The Measurement Problem, and What Changed
Everything above assumes you can see what happened on the first contact. In a manual QA program you cannot. A thorough review takes 15 to 20 minutes per call, so most teams review 1 to 5% of calls, and the repeat-contact pairs you actually need to study are rarely in that sample. You end up with an accurate count of repeats and no idea what caused them.
This is the specific gap call center quality assurance software closes. Voxjar does not record your calls; it analyzes the recordings your phone system already produces, transcribes them, and scores 100% of them against your own scorecard with the supporting quote pulled from the transcript. That makes FCR measurable from what was actually said rather than from a disposition code: you can flag every call containing an unfulfilled callback promise, every escalation that was inside the agent’s authority, every close that skipped confirmation, and see which of the root cause buckets above is producing your repeat volume.
It also makes the number honest in the other direction. When the metric comes from the transcript instead of the agent, there is nothing to game, and FCR can go back to being a diagnostic instead of a target people manage around. For how this fits into a broader quality program, see the guide to call center QA and customer experience.
Start With Your Own Calls
FCR is worth the trouble because it is the rare metric where the customer’s interest, the agent’s interest, and the cost line all point the same way. The failure is almost never that teams do not care about it. It is that they measure it from the one source with a reason to be optimistic, then coach agents about a number that was describing a broken process all along.
The fastest way to find out what your real FCR looks like: take a handful of calls where the customer had to contact you twice and read what happened on the first one.
Score a real call with Voxjar’s free AI evaluation. Upload a recording, apply a scorecard with resolution criteria on it, and see the score, the reasoning, and the exact transcript moment behind it. Plans start at $99 a month with unlimited users, so the whole team can work from the same evidence.