Average Handle Time and First Call Resolution
AHT and FCR formulas, 2026 benchmarks by industry, and a real BPO account showing why cutting handle time without tracking resolution rate raises total cost.
Is it right for you?
- Calculate AHT from all three inputs (talk time, hold time, after-call work), not talk time alone.
- Exclude wrong numbers, abandoned calls, and unresolvable contacts from the FCR denominator before calculating the rate.
- Measure FCR through a post-call survey or an independent QA evaluation, never through agent self-report.
- Compare AHT against your own industry benchmark, not a blended cross-industry average.
- Set an FCR floor before setting an AHT target, and track both metrics on the same dashboard.
What AHT and FCR actually measure
Average handle time (AHT) measures how long an agent spends on one contact, from pickup to the end of after-call wrapup. First call resolution (FCR) measures whether that contact actually solved the customer's problem, or whether they had to reach out again. Teams that track AHT without watching FCR next to it tend to draw the wrong conclusion: a shorter individual contact is not efficient if a third of those contacts turn into a second, third, or fourth interaction about the same issue. This guide covers both formulas, the eligibility rules that keep FCR honest, 2026 benchmarks by industry, and a real BPO account of what happens when handle time becomes the only number anyone is measured against.
AHT is calculated as: AHT = (total talk time + total hold time + total after-call work) ÷ total number of interactions. All three inputs matter, and a dashboard that only shows talk time understates the real number. Talk time is the minutes an agent is actively on the line or in the chat window. Hold time is any stretch where the customer is parked, waiting on a lookup or a transfer. After-call work (ACW), sometimes called wrap-up, is everything the agent does once the contact ends and before the next one can start: tagging the ticket, logging notes, updating the CRM record.
A worked example: a 12-agent support team logs 600 eligible contacts in a week. Talk time across those contacts totals 1,800 minutes, hold time adds 300 minutes, and after-call work adds another 300 minutes. That is 2,400 minutes of total handle time across 600 contacts, an AHT of exactly 4 minutes per contact. Move any one of those three inputs and AHT moves with it, which is why a team that looks fast on talk time alone can still post a mediocre AHT once after-call paperwork is added back in.
FCR is calculated as: FCR = (eligible contacts resolved on first contact ÷ total eligible contacts) × 100. The word doing the real work is eligible. SQM Group, the research firm that benchmarks roughly 500 call centers a year, defines an eligible contact as one where the agent had a genuine chance to resolve the issue: a billing question, a service problem, an account change. Wrong numbers, abandoned calls, test calls, and contacts that disconnected before the agent understood the issue get excluded from both sides of the fraction [SQM Group, checked 2026-09-01]. Skip that filtering step and FCR gets diluted by contacts that were never resolvable in the first place, which flatters a team that is actually struggling.
Continuing the same team: of the 600 contacts that week, 40 were wrong numbers or disconnected before the issue was understood, leaving 560 eligible. A post-call survey, where the customer confirms resolution rather than the agent, found 431 of those 560 were resolved without a repeat contact. FCR = 431 ÷ 560 × 100, about 77%, which lands inside the healthy range covered in the next section.
One more distinction worth knowing before comparing a number against anyone else's: First Call Resolution measures a single touchpoint, typically phone. First Contact Resolution applies the same idea to any channel, chat, email, a web form. One Contact Resolution (OCR) is stricter still: it counts an issue as resolved only if the customer never had to use a second touchpoint at all, not even a follow-up email, to close it out. SQM Group's benchmarking puts OCR at roughly 10% lower than FCR for the same population of contacts, because some issues closed on the phone quietly reopen by email a day later [SQM Group, checked 2026-09-01]. If a help desk tool reports one blended resolution number across channels, check which of the three it is actually computing before trusting it.
2026 benchmarks by industry, and why the blended number is nearly useless
Vendors and analyst reports love to publish a single AHT number and a single FCR number as though every support queue looks the same. They do not. Sprinklr's 2026 contact center data puts the blended average handle time across all sectors at 6 minutes 10 seconds, and SQM Group's benchmarking puts the average FCR rate at 71% [Sprinklr, cited via Kayako's 2026 benchmark compilation, checked 2026-09-01; SQM Group, checked 2026-09-01]. Neither number says much on its own. A retail team fielding order-status calls and a SaaS support desk walking a customer through an API integration bug are not playing the same game, and benchmarking one against the other's average is how a team ends up chasing the wrong target.
| Sector | Typical AHT (2026) | FCR target range |
|---|---|---|
| Retail & eCommerce | 3-5 min (voice) | 75-85% |
| Financial services | ~4 min 45 sec | 70-78% |
| Healthcare | ~6.6 min | 75-82% |
| Telecommunications | 2-4 min hold, 8-12 min technical | 65-75% |
| SaaS / tech support | 7-10 min | 72-80% |
| Travel & hospitality | 5-7 min | 70-78% |
| Utilities & energy | 6-8 min | 68-76% |
Those sector ranges are drawn from Sprinklr's and Sobot's 2026 contact center benchmarking as compiled by Kayako, not from a primary audit of each industry. Treat them as directional, useful for spotting whether a team is wildly out of range, not as a precise target to hit [Sprinklr and Sobot data, cited via Kayako, checked 2026-09-01]. NICE's own glossary entry on the metric states plainly that no universal AHT standard exists, because different industries and call types carry fundamentally different benchmarks, and that AHT should be pushed down only without compromising service quality [NICE, checked 2026-09-01]. That caveat is the entire subject of the next section.
The trade-off nobody puts on the dashboard
A mid-sized BPO account, about 800 seats, handling tech support for a major electronics brand, ran an 8-minute AHT target and hit it consistently. Every individual call cleared 8 minutes, agents earned their bonuses, and the leadership dashboard looked clean. Then the operations team pulled a second number: first call resolution sat at 61%. Nearly four in ten customers were calling back about the same issue, and once those repeat contacts were tracked and added up, the real time to actually close a problem averaged 23 minutes spread across multiple calls, almost three times the 8-minute figure anyone was being scored against [Etech Global Services, checked 2026-09-01].
That same team ran a 90-day pilot afterward: 50 agents, no AHT target, permission to stay on a contact until the issue was actually closed. Average handle time went up 2.3 minutes per contact. FCR climbed from 61% to 84%. Repeat contacts fell 35%, and customer satisfaction scores rose 18 points. Once total cost, including every repeat contact, transfer, and escalation, was modeled against the pilot group, the slower-looking operation came out 14% more efficient than the fast one [Etech Global Services, checked 2026-09-01]. A longer call that actually closes the issue can cost less than three short ones that do not.
That is one operator's account of one queue, not a controlled study, and the exact multiplier will not carry over to every team. What does carry over is the mechanism: push agents to close contacts fast without tracking whether the issue got solved, and agents rationally optimize for the number they are measured on, wrapping up quickly and letting the next contact become someone else's problem later. The fix is not abandoning AHT. It is refusing to move the number without an FCR floor sitting on the same dashboard, tracked continuously instead of reconstructed from memory after volume spikes, see our contact center analytics software guide for tools built to hold both metrics side by side.
Measuring FCR without letting agents grade their own work
The fastest way to get a meaningless FCR number is to let agents self-report it. An agent asked whether their own call resolved the issue has every incentive, conscious or not, to say yes, and self-reported FCR routinely runs well above what a customer would confirm if asked directly. SQM Group treats this as settled in its own methodology: FCR has to be measured externally, through the customer's account of what happened, or through a systematic evaluation of the interaction itself, never through an agent's own say-so.
SQM lists two credible measurement methods. The first is the post-call survey: ask the customer directly, right after the contact, whether the issue was resolved without needing to reach out again. The second is an AI-powered QA evaluation, where a rules-based model reviews the call recording or transcript and determines resolution independently of what either the agent or the customer reports. Run across SQM's client base, the two methods land within a point of each other, 71% for the survey approach against 72% for the AI-evaluated approach, agreeing on individual-contact outcomes 93% of the time [SQM Group, checked 2026-09-01]. That alignment matters for a smaller team without survey budget: a competent QA process reviewing recordings can substitute for a formal survey program and land close to the same number. See our contact center QA software guide for tools built for that review.
Keep the eligibility filter from the first section in mind when setting either method up. A survey or QA review that scores every single contact, including the wrong numbers and the two-minute disconnects, will produce a number nobody can act on. Score only the contacts where an agent genuinely had a chance to resolve something.
Bringing AHT down without breaking FCR
Most AHT bloat is not agents typing slowly, it is agents searching. A shared inbox or help desk tool that forces an agent to leave the ticket and hunt through a separate wiki or chat thread for an answer adds minutes that never show up as talk time, they show up in hold time or after-call work instead. Putting knowledge base search inside the same window as the ticket, not adjacent to it, is usually the single biggest lever available before touching anything about how agents talk.
Scripting and decision-tree tools can cut AHT, but only when they are built to get an agent to the correct answer faster, not simply to any answer faster. A script that shortcuts to a canned close before the actual issue is confirmed resolved will lower AHT and quietly wreck FCR in the same move. See our call center scripting software guide for the difference between tools built for accuracy and tools built purely for speed.
Understaffing is a common, underappreciated driver of rushed calls. An agent who can see the queue backing up behind them will unconsciously speed through the current contact, whether or not anyone told them to. If AHT climbs specifically during peak intervals and FCR dips at the same time, the root cause is often a staffing gap, not an agent or script problem. See our call center workforce management software guide for the forecasting side of that fix.
The order of operations matters most. Establish a real FCR baseline, survey or QA evaluation, not agent self-report, before touching AHT targets at all. Pick a floor the number cannot drop below, then work on the actual time-wasters (bad routing, missing knowledge access, tool-switching) instead of pressuring agents to talk faster. Track both numbers on one view. A dashboard that only shows AHT will eventually reward exactly the wrong behavior.
FAQ
What counts as a good AHT figure?
There is no single correct number. NICE's glossary entry for the metric warns that benchmarks shift by industry and call type, so a figure that looks slow for one queue is normal for another. The 2026 cross-sector average works out to a bit past six minutes, yet a fast retail order-status line clears 3-5 minutes while a SaaS technical queue often runs 7-10 minutes. Weigh a team's result against peers doing the same kind of work, not against that catch-all average.
What is a good FCR rate?
SQM Group, which tracks roughly 500 centers annually, reports an industry FCR average of 71%. A rate between 70% and 79% counts as healthy, and only about one center in twenty clears 80% or higher, the threshold SQM labels world-class.
Does lowering AHT actually save money?
Not by itself. The same BPO operations write-up cited above described an electronics support queue running 800 seats where agents hit a fixed eight-minute call ceiling every time, yet only 61 out of every 100 callers had their problem solved before hanging up. Add the callbacks back in and the true time needed to close a single issue stretched to 23 minutes across separate calls. A later trial that dropped the ceiling entirely raised individual call length but cut overall cost by 14% once those callbacks disappeared.
FCR versus the stricter OCR metric, what actually changes?
FCR credits a resolution the moment one touchpoint, a phone call included, closes the loop. OCR sets a higher bar: nothing counts as solved if the customer had to circle back on any channel afterward, a follow-up note included. Pulling from the same pool of contacts, SQM Group finds OCR scores land about a tenth lower than the matching FCR figure.
Can agents accurately self-report their own FCR?
Not reliably. Letting the person who handled the call decide whether it counts as resolved invites optimistic scoring. SQM Group's approach instead leans on confirmation from the customer, gathered right after the interaction, or a separate rules-based review of the recording, precisely to keep that incentive out of the number.