Call Center KPIs That Actually Matter: FCR, AHT, CSAT

The call center KPIs that are most critical are first call resolution (FCR), average handle time (AHT), customer satisfaction (CSAT), service level, and occupancy — five numbers that together show whether an operation is effective, efficient, and sustainable. Everything else is derivative.

Most metric guides list twenty or thirty. That is not rigour; it is noise, and it is why so many operations track everything and manage nothing. Five numbers carry the signal. The rest are diagnostics you reach for when one of the five moves in a direction you don’t like.

This guide gives each of the five a plain definition, its formula, an honest word about benchmarks, and — the part vendor listicles leave out — the trap hiding inside it. Then it covers the trade-offs nobody puts on a dashboard: what breaks when you optimize one of these numbers in isolation.

The Five Core KPIs

Three questions decide whether a contact operation is healthy. Is it effective — does it actually solve things (FCR, CSAT)? Is it efficient — does it absorb demand without waste (AHT, service level)? Is it sustainable — can it keep doing this next quarter with the same people (occupancy)? Five numbers, three questions. 

While there may be other operational, speed or agent related metrics – like Net Promotor Score (NPS) – a measure of overall customer loyalty, Abandonment Rate – how many callers hang up before an agent answers, Quality Assurance (QA) Score –  the audit score evaluation of service quality standards – these are secondary KPIs often linked to primary ones. 

First Call Resolution (FCR) — did we actually solve it?

FCR is the share of contacts resolved on the first attempt, with no follow-up needed.

Formula: (Contacts resolved on first contact ÷ total contacts) × 100

Now the part that decides whether your FCR number means anything: who gets to say it was resolved. There are three common methods, and they do not produce the same figure. You can ask the agent to flag resolution — the cheapest method, and the one most vulnerable to optimism and to targets. You can ask the customer at the end of the interaction. Or you can define it structurally: no repeat contact from the same customer, on the same issue, within a fixed window — seven days is a common window, checked through the CRM.

Those three methods can be measured on the same week of the same operation and produce meaningfully different results — the agent-flagged number is almost always the flattering one. This is why an FCR benchmark quoted without its measurement method is close to meaningless, and why the sector’s published “industry standard” figures scatter so widely. Before you compare your FCR to anyone’s — a benchmark, a competitor, a provider’s proposal — establish how each side counts a resolution. If a provider quotes you an FCR figure and cannot immediately tell you which method produced it, you have learned something more useful than the number.

Get the measurement honest and FCR becomes the most valuable number on the floor. Every point of genuine improvement removes repeat contacts, which cuts volume, which relieves the queue, which lifts satisfaction — one metric quietly improving three others.

And when FCR is genuinely low, the cause is rarely the agents. It is almost always one of three things: they lack the information (the knowledge base is thin or out of date), they lack the authority (every exception needs a supervisor, so nothing is resolved on the first call by design), or the contact reached the wrong person in the first place. Fix those and FCR moves. Exhort agents to “resolve on first contact” without fixing them, and all you have done is teach the floor to flag optimistically.

Average Handle Time (AHT) — a capacity number, not a whip

AHT is the average total time an agent spends on one interaction: talk time, hold time, and the after-call work required to finish it.

Formula: (Total talk time + total hold time + total after-call work) ÷ total contacts handled

Its legitimate use is arithmetic: forecast your volume, multiply by AHT, and you know how many agents you need on the floor at 6pm on a Thursday. AHT is how a schedule gets built.

Its illegitimate use is as an agent speed target — and it is, in our experience, the most abused number in this industry. Pressure agents to bring AHT down and they will comply, but not by becoming better. They will rush explanations, skip the confirmation question that would have caught the real problem, and transfer calls that they could have solved. The minutes come off AHT and reappear as tomorrow’s repeat contacts. You did not save time; you moved it, and paid a customer’s patience for the privilege.

The honest answer to “how do we reduce AHT” is: never by pressure. Reduce it through better knowledge bases so agents stop searching, cleaner routing so calls arrive at someone qualified, and automated after-call work so wrap-up doesn’t devour the minutes. Those cut handle time without cutting quality — because they remove work rather than rushing it.

Customer Satisfaction (CSAT) — the number your customers give you

CSAT is the percentage of customers who report being satisfied after an interaction, gathered by a short post-contact survey.

Formula: (Satisfied responses ÷ total responses) × 100

Two traps live in the collection, not the calculation. The first is timing: ask immediately after resolution and you are measuring the interaction; ask three days later and you are measuring the customer’s mood, their memory, and whatever else happened to them since. The second is response bias: the delighted and the furious answer surveys; the mildly-content majority ignore them. A CSAT score built on a thin response rate is a poll of your extremes.

CSAT, NPS, or CES — which? They answer different questions and are not interchangeable. CSAT asks was this interaction good? — the right question for judging a contact operation. NPS (Net Promoter Score) asks would you recommend this company? — a relationship and brand question, which a single support call rarely moves. CES (Customer Effort Score) asks how hard did we make this for you? — often the sharpest predictor of whether a customer quietly leaves. Use CSAT to run the floor; use NPS and CES to understand the relationship. Swapping one for another because it produces a nicer number is self-deception with a methodology.

One more caution: CSAT is a lagging number and a blunt one. It tells you something went wrong; it rarely tells you what. A score that drops four points this month could be a queue problem, a policy the agents cannot defend, a product fault they are absorbing the anger for, or a survey that started reaching a different set of customers. CSAT is where you start the investigation, not where you finish it.

Improving CSAT is a discipline of its own — survey design, driver analysis, and service recovery — and we cover it properly in how to improve CSAT.

Service Level — were calls answered fast enough?

Service level is the percentage of contacts answered within a target time. It is usually written as a pair: 80/20 means 80% of calls answered within 20 seconds.

Formula: (Calls answered within the threshold ÷ total calls answered) × 100

Note what that pair really is: a choice, not a law. There is nothing sacred about 80/20 — it is a convention, and a bank’s fraud line and a furniture retailer’s order line have no business promising the same one. Set the threshold your customers’ patience actually justifies, then staff to it — and be honest that a tighter promise is not free. Every second you shave off the threshold costs agents on the floor, because you are buying the ability to absorb a surge that may not come. A demanding service level is a real commercial decision, not a statement of ambition.

The trap is averaging. The Average Speed of Answer (ASA) is the mean wait time before a call is picked up. A day, a week, or a month can post a comfortable service level while containing a stretch of misery: the 6-to-8pm block where the queue collapsed, every caller waited four minutes, and a third of them hung up. The daily average absorbed it. The customers did not. Service level must be read in intervals — typically half-hours — or it will hide precisely the failures you are paying it to expose. In Iraq this is not an academic point. Contact patterns here have their own shape — Ramadan evenings, in particular, concentrate demand into hours that a schedule built on a monthly average will systematically under-staff. Interval-level planning is the difference between a service level you report and a service level your customers experience.

Occupancy — the sustainability number

Occupancy is the share of an agent’s logged-in time spent actually handling contacts, including after-call work.

Formula: (Handling time ÷ logged-in time) × 100

Distinguish it from utilisation, which measures productive time against paid time — a broader number that includes training, meetings, and breaks. Occupancy asks: when an agent is on the floor and available, how much of that time are they working? Confusing the two is a common way to accidentally overstate how hard a team is being pushed.

The trap here is the most expensive in this article, because it disguises itself as good news. High occupancy looks like efficiency. Sustained high occupancy is burnout with a spreadsheet. An agent at 95% occupancy has almost no gap between conversations — no recovery, no time to think, hours of continuous emotional labour. What follows is predictable and always in the same order: quality erodes first, then errors and absence rise, then people leave. Then you are recruiting and training replacements, at which point your “efficient” operation is paying for the same work twice.

You cannot run a contact operation at full occupancy for the same reason you cannot run a motorway at 100% capacity: occupancy in the mid-80s with strong resolution is a better operation than the mid-90s with customers calling back. Headroom is not waste. Headroom is what makes the other four numbers achievable.

The Supporting Cast — Metrics That Explain the Five

These are diagnostics. They do not belong on the executive dashboard; they belong in the conversation about why one of the five moved.

Abandonment rate — the share of callers who gave up while waiting. It is service level’s shadow: when service level slips, abandonment is where the damage shows.

Adherence — whether agents are where the schedule expects them, when it expects them. A forecast can be perfect and the floor still empty at the moment of peak demand; adherence is how you find out.

Shrinkage — the paid hours that are not available for handling contacts: training, meetings, breaks, leave, absence. Forecast without it and you will be short-staffed every single day while your maths insists you are fine.

QA score — what your monitoring actually found in sampled conversations, scored against an agreed scorecard.

Calibration — the practice of having evaluators score the same recorded call and comparing results until they converge. Skip it and your QA score is not a measurement; it is a collection of individual opinions wearing a percentage sign.

The Trade-Offs Nobody Puts on the Dashboard

Here is what the metric listicles will not tell you, because it complicates the dashboards they are selling. These five numbers pull against each other. Optimise one alone and you will pay for it somewhere else.

AHT versus FCR. Squeeze handle time and resolution quality falls — the rushed call becomes tomorrow’s second call. The minutes you saved come back as volume, and you have traded a slower resolution for two faster failures. If AHT falls while FCR falls with it, you have not improved efficiency; you have degraded the service and hidden the evidence in a different column.

Occupancy versus sustainability. Running hot flatters this quarter’s numbers and mortgages the next one. Attrition, retraining, and quality decay arrive with a lag of a few months — which is exactly long enough for the manager who pushed occupancy to have been congratulated before the bill lands.

Targets versus gaming. Any metric that becomes a target will be gamed — not by bad people, but by ordinary people responding rationally to how they are judged. AHT targets produce transfers. FCR targets produce optimistic resolution flags. Survey-score targets produce agents coaching customers on how to answer the survey. The defence is not tighter targets or closer surveillance; it is measurement design: customer-confirmed resolution instead of agent-flagged, calibrated QA instead of a single evaluator, and metrics reviewed in pairs so that no number can be improved silently at another’s expense.

The practical defence is simpler than it sounds: never read these numbers alone. AHT is meaningless without FCR beside it. Service level is meaningless without abandonment. Occupancy is meaningless without attrition. Any metric that can only be seen in isolation can be improved in isolation — and something else is quietly paying for it.

Manage the five as a system. Every serious operation has watched someone win on one metric and lose the operation.

Tailoring Metrics to Your Operation

These metrics can also vary based on the type of operations a company is dealing with.

If setting up a new call center, a company’s main focus should be speed and foundational quality. Setting baselines rather than high-performance targets immediately is recommended. Prioritize tracking Service Level to ensure callers aren’t waiting, and Quality Assurance (QA) Scores to ensure agents are handling calls correctly.

In case of improving an existing one, focus is redirected to efficiency and customer experience optimization. Catering more to reducing friction from First Call Resolution (FCR) to see if agents are solving problems on the first try, and Average Handle Time (AHT) to identify bottlenecks in the process. 

When targeting Inbound Support, your main focus is Customer satisfaction, accuracy, and speed. Customer Satisfaction (CSAT) and First Call Resolution (FCR) are your top priorities. You want to make sure the customer is happy and doesn’t have to call back. 

For Outbound Sales, track Conversion rates and activity volume. Conversion Rate and Call-to-Close Ratio are crucial to track how many calls are made before a sale is finalized, and which agents are performing best. 

Building a Scorecard That Works

Keep the five core numbers permanently. Add two or three contextual metrics that match what your operation actually is — abandonment for a high-volume helpline, QA score for a regulated or high-risk queue, adherence for an operation with sharp intraday peaks.

Review weekly, not daily. Daily review invites reaction to noise; monthly review is too slow to act. Calibrate QA monthly. Give every metric a named owner, because a number that belongs to everyone is watched by nobody.

And accept the last, slightly uncomfortable truth: a scorecard that never causes an argument is a scorecard nobody is using. The disagreements — about whether that call was really resolved, about why the evening service level slipped — are where the operation actually improves.

What This Page Doesn’t Cover (and Where It Lives)

Improving CSAT specifically — survey design, driver analysis, and how to recover a failed interaction — has its own guide, linked in the CSAT section above.

Outbound campaigns run on a different set of primaries: contact rate, conversion, attempts per record. Judging an outbound campaign by inbound numbers, or the reverse, is a classic management error and a subject for another day.

And if you are earlier in the journey — asking what a partner actually takes over and how the model works before you worry about how to measure it — start with what call center outsourcing is.

Final Guidance

Five numbers. Three questions. Is the operation effective, is it efficient, and can it survive its own efficiency?

Every one of those numbers can be made to lie — by measuring resolution generously, by averaging away a terrible evening, by calling burnout productivity. Which is why the measurement method matters more than the target, and why an operation that argues about how it counts is healthier than one that quietly agrees to be flattered.

These are the numbers we report to clients every week, with the method written down beside them, across our call center services in Iraq.