Skip to content

Bananaflow MCP is live.

Read the docs

Call Center Metrics: AHT, FCR, CSAT Explained

Jaseem Abbas, founder of Bananaflow, in glasses and a dark top against a pale background.Jaseem Abbas

Published
Updated
Cover for "Call Center Metrics: AHT, FCR, CSAT Explained": a white isometric line diagram of six gauges on one board: six low tiles in two rows of three, marked with a clock, a check, a smile, a cross, a battery and an arrow, labelled AHT, FCR, CSAT, Abandoned, Occupancy and Connect, with FCR lit, drawn over a soft, out-of-focus cyclamen in muted rose, its slender swept-back petals streaming sideways, with a cool sky blue haze at the left end and a warm peach glow at the right.

At 9:05 every morning, the supervisor of a 20-agent customer care floor in Chennai opens one sheet. It shows yesterday's calls, and six numbers decide how the day starts: how long the average call ran, how many callers got their answer the first time, how many rated the call well, how many hung up in the queue, how busy the agents were and, on the outbound side, how many dials were answered. Everything the floor does that day is a reaction to those six numbers.

This guide defines each one, gives the formula, works a sample day through them and shows how an Indian call floor uses them. It also warns about the traps: every one of these numbers can be pushed in the wrong direction by someone who is trying to look good.

Six metrics with their formulas and the sample day's values. Each arrow shows the healthy direction; occupancy is best in a middle band, and AHT should fall only while FCR holds.

What are call center metrics?

Call center metrics are the counts and rates a team uses to see how well it is answering or making calls. They cover speed, quality, cost and effort. Each one is a number worked out from call records, so a good one is clearly defined, easy to compute and hard to fake.

A call center, as Wikipedia's entry on the call centre describes it, is an office that deals with a large volume of enquiries by telephone, so the volume alone makes measurement necessary. No manager can listen to every call. The metrics are the summary that tells them where to listen.

Most of these numbers have a common problem: they have more than one definition. "Handling time" may or may not include the time after the call ends. "Resolution" may be counted over a day or a week. Write your definition down before you start, and apply it the same way every time.

What is average handling time (AHT)?

Average handling time is the average time an agent spends on one call from start to finish, including any hold time and the wrap-up work after the caller hangs up. The formula is total talk time plus hold time plus after-call work, divided by the number of calls handled.

On the sample day below, the floor handled 1,080 calls. Agents spent 4,320 minutes talking, 540 on hold and 540 on after-call work, which is 5,400 minutes. AHT is 5,400 ÷ 1,080, which is 5.0 minutes.

Use AHT to staff and to cost. A longer AHT means more agents per hundred calls and a higher cost per call, which our guide to cost per call in India turns into rupees. Do not use it to rank agents on its own. An agent pushed to cut AHT will rush callers, and the repeat calls that follow cost more than the minutes saved. Our AHT guide covers the full form, ACW and benchmarks.

What is first call resolution (FCR)?

First call resolution is the share of issues that are solved on the first contact, so the customer does not need to call again. The formula is issues resolved at first contact divided by total issues, as a percentage. The hard part is the window: choose how long you wait before deciding that a repeat call is the same issue.

In the sample day, 756 of the 1,080 calls were resolved at the first contact and not repeated within 7 days. FCR is 756 ÷ 1,080, which is 70%. The 7 days is a choice made for the example. Pick your own window and keep it.

FCR is the best check on AHT. If AHT falls and FCR falls with it, you have made calls shorter and problems longer. Customers ring back, your volume rises and your cost climbs. Match FCR to the reason for the call, since a billing question and a technical fault resolve differently.

What is customer satisfaction (CSAT)?

CSAT is the share of surveyed customers who rate the interaction as good. The formula is the number of satisfied responses divided by the number of responses, as a percentage. "Satisfied" usually means the top two points on a five-point scale. Wikipedia's entry on customer satisfaction describes how the measure is used.

On the sample day, 300 callers answered the survey, and 240 rated the call 4 or 5. CSAT is 240 ÷ 300, which is 80%. Note the size of the sample: 300 responses out of 1,080 calls is 28%, and the people who answer surveys are not a random slice of callers. Very happy and very angry callers answer more often. Our guide to customer feedback calls shows how to sample callers so the score is fair.

A related measure, the net promoter score, asks how likely the customer is to recommend you and is a better view of the relationship than of one call. Use CSAT for the call, and NPS for the brand.

What is abandonment rate?

Abandonment rate is the share of inbound callers who hang up before they reach an agent. The formula is abandoned calls divided by calls offered, as a percentage. Calls offered means every call that came in, answered or not.

In the sample day, 1,200 calls came in and 120 callers hung up first. Abandonment is 120 ÷ 1,200, which is 10%. Some of those callers hung up within a few seconds, perhaps because they misdialed, and many floors ignore abandons under a short threshold. Decide your threshold and write it down.

Abandonment shows how the queue feels to the caller. It rises when volume spikes, when agents are on break together and when callers are put through a long menu first. A related measure is service level: the share of calls answered within a set time, such as a given number of seconds. The time and the target are your own choice. To find how many agents a service level needs, see Erlang C.

Do not confuse this with the dialer idea of an abandoned call, where a person answers an outbound call and no agent is there. Our guide to auto dialers covers that case.

What is occupancy?

Occupancy is the share of an agent's available time spent handling calls. The formula is handling time divided by handling time plus time spent waiting for the next call, as a percentage. It measures how busy people are while logged in and ready, and not how many hours they were paid for.

In the sample day, agents handled calls for 5,400 minutes and waited ready for 1,800 minutes. Occupancy is 5,400 ÷ (5,400 + 1,800), which is 75%.

Both extremes hurt. Low occupancy means paid time with nothing to do. Very high occupancy means no breathing space, and tired people give worse answers and leave sooner. Read it next to CSAT and attrition: if occupancy climbs and CSAT drops, the floor is overloaded.

What is connect rate?

Connect rate is the share of outbound dial attempts that a person answers. The formula is answered calls divided by calls dialed, as a percentage. It tells you whether your list, your timing and your numbers are working before anyone speaks.

In the outbound sample, an agent dialed 3,600 numbers and 1,080 were answered. Connect rate is 1,080 ÷ 3,600, which is 30%. Pair it with what happens next. A high connect rate with few outcomes points to the script or the offer. A low connect rate points to old numbers, the wrong hours or a poor retry plan.

Be clear about what counts as answered. Voicemail, an answering machine or a phone menu is not a person, so a good report separates them.

How does an Indian call floor use these numbers?

A call floor uses these numbers on three clocks: a daily huddle, a weekly review and a monthly plan. The daily one is for fixes, the weekly for coaching and the monthly for staffing and cost. The numbers are the same, but the question changes. For what to put on the screen, see what to put on a call center dashboard.

  • Metric: AHT

    Formula
    (Talk + hold + after-call work) ÷ calls handled
    What it tells you
    How long a call takes, and so staffing and cost
    Read it with
    FCR
  • Metric: FCR

    Formula
    Issues resolved at first contact ÷ total issues
    What it tells you
    Whether callers get what they called for
    Read it with
    AHT
  • Metric: CSAT

    Formula
    Satisfied responses ÷ total responses
    What it tells you
    How callers rate the call
    Read it with
    Response count
  • Metric: Abandonment rate

    Formula
    Abandoned calls ÷ calls offered
    What it tells you
    How many callers give up before an answer
    Read it with
    Service level, volume by hour
  • Metric: Occupancy

    Formula
    Handling time ÷ (handling time + ready time)
    What it tells you
    How busy agents are while available
    Read it with
    CSAT, attrition
  • Metric: Connect rate

    Formula
    Answered calls ÷ calls dialed
    What it tells you
    Whether outbound numbers and hours work
    Read it with
    Outcomes per connected call

In a Chennai-style customer care floor, the daily huddle covers abandonment by hour and AHT by queue: did the 10 am spike overload the Tamil queue? The weekly review covers FCR and CSAT by agent and by call reason, where team leads pick calls to listen to together. The monthly plan uses AHT, occupancy and volume forecasts to decide how many seats to staff for the festival season.

Indian floors add three twists. First, language: the same number often needs separate queues for Hindi, English and a regional language, and the numbers should be read per language, since a good average can hide a failing queue. Second, peaks: festival sales, exam results and fuel or gas booking days push inbound volume far above normal, so hour-by-hour views matter more than daily averages. Third, mixed work: many floors do outbound follow-ups in the quiet hours between inbound spikes, and the numbers for each need to be kept apart.

For how inbound and outbound jobs differ in what they measure, read our guide to inbound and outbound calls.

How do you stop the numbers from being gamed?

Measure pairs, sample real calls and change one target at a time. Any metric that carries a bonus will be pushed, and pushed in the cheapest way. AHT drops because agents hang up early. FCR rises because agents mark cases closed. CSAT rises because agents ask only happy callers to rate. Outcome codes are gamed too, as our list of disposition codes explains.

Four habits prevent most of this:

  1. Pair measures that pull against each other. AHT with FCR, occupancy with CSAT.
  2. Listen to calls. Numbers say where to listen. Only listening says what happened.
  3. Control the survey. Send it to every caller, or a random sample, and not to the agent's choice.
  4. Keep definitions fixed. If you change a definition, mark the date on the chart.

How do AI voice agents change the numbers?

An AI voice agent produces a record for every call, so the counts are there without manual tagging. You can see how long calls ran, how they ended and what the agent collected. The same measures apply, and a few change meaning.

  • AHT becomes the minutes per call, and also the cost per call, since the price is per minute.
  • Occupancy matters less, since an agent is not tired and does not wait idle at a desk. The limit is the number of lines in use at once. Our Erlang C guide shows how to size those lines.
  • FCR still matters most. A call that ends quickly without solving the problem is worse than a long one that does.
  • CSAT and abandonment work as before, and the handoff to a person becomes a number to watch.

Bananaflow keeps a transcript, a recording, the minutes used and the data collected for each call. A QA step can score a call and return tags, sentiment and a summary, and call outcomes are recorded as dispositions, which are outcome codes. Reports show KPIs and trends and export to CSV. See analytics for what the reports show and call QA for how scoring works.

How Bananaflow fits

Bananaflow builds AI voice agents for inbound and outbound calls in English and 10 Indian languages, billed per minute. Because each call leaves a record, you can start with the same six numbers you use today and compare an agent's calls with your team's on equal terms.

See the pricing page for rates.

FAQ

Frequently asked questions

  • What is average handling time (AHT) and how is it calculated?

    AHT is the average time an agent spends on one call, including hold and after-call work. The formula is total talk time plus hold time plus after-call work, divided by calls handled.

  • What is first call resolution (FCR)?

    FCR is the share of issues solved on the first contact. The formula is issues resolved at first contact divided by total issues. Choose how long you wait before a repeat call counts as the same issue.

  • What is occupancy in a call center?

    Occupancy is handling time divided by handling time plus the time agents wait ready for the next call. Very low means idle paid time. Very high leaves no breathing space and can hurt quality.

  • What is connect rate?

    Connect rate is the share of outbound dial attempts that a person answers: answered calls divided by calls dialed. Voicemail and phone menus should not count as answered.

  • How do you stop call center metrics from being gamed?

    Measure pairs such as AHT with FCR, listen to real calls, send surveys to every caller and keep each definition fixed. Any metric tied to a bonus gets pushed.

Written by

  • Jaseem Abbas, founder of Bananaflow, in glasses and a dark top against a pale background.

    Jaseem Abbas

    Founder, Bananaflow

    Jaseem founded Bananaflow to build voice AI that understands how people in India really speak, including the switch between languages in the middle of a call.

Similar articles

Bring us one process. We'll pilot it on your real calls, with your team as the fallback.

Plan a pilot