Call abandonment rate is the share of callers who hang up while waiting in the queue, before an agent answers. Divide the calls abandoned in queue by the calls that entered it, then multiply by 100. A queue that took 600 calls in a half hour and lost 60 of them to hang-ups has an abandonment rate of 10 percent.
Call abandonment rate = calls abandoned in queue / calls offered to the queue x 100
In a spreadsheet with calls offered in B2 and calls abandoned in C2, the rate is =C2/B2, formatted as a percentage. If you also exclude short abandons, the callers who hung up within a few seconds, put that count in D2 and use =(C2-D2)/(B2-D2).
Most customer service centers can live with 3 to 5 percent, according to COPC, and the best large centers in the research literature lose 1 to 2 percent. Before you compare your figure with either, check what your platform counts as an abandon and what it divides by. Those two choices move the same half hour of calls from 7.2 to 12.0 percent, as the worked example below shows.
| What you want to know | Formula | Example from one half hour |
|---|---|---|
| Abandonment rate | Calls abandoned in queue ÷ calls offered × 100 | 60 ÷ 600 = 10.0 percent |
| Rate without short abandons | (Abandoned − short abandons) ÷ (offered − short abandons) × 100 | 42 ÷ 582 = 7.2 percent |
| Average time to abandon | Total wait of abandoned calls ÷ calls abandoned | 3,960 seconds ÷ 60 = 66 seconds |
| Share of abandons that were short | Short abandons ÷ calls abandoned × 100 | 18 ÷ 60 = 30 percent |
What counts as an abandoned call?
An abandoned call is one where the caller hangs up while waiting in an ACD queue, before reaching an agent. On the two platforms below, callers who leave the queue another way, through a callback, a timeout to voicemail or a transfer out, are not counted as abandoned. Calls that end in the phone menu before they reach a queue are not counted either.
Amazon Connect divides contacts abandoned by contacts queued, and "contacts queued for callback are not counted as abandoned" (Amazon Connect metric definitions). Genesys Cloud divides "Abandoned in Assignment" by "Offered to Assignment" (abandoned rate). It treats callbacks, transfers to voicemail and callers who return to the menu as flow-outs rather than abandons, and its abandon rate does include short abandons (what abandons include).
| Platform | Abandonment rate in its documentation | Callbacks and voicemail | Short abandons |
|---|---|---|---|
| Amazon Connect | Contacts abandoned ÷ contacts queued × 100 | Queued callbacks are not abandons | Counted in the rate; a separate "contacts abandoned in X seconds" count has presets from 15 to 600 seconds |
| Genesys Cloud | Abandoned in assignment ÷ offered to assignment | Flow-outs, not abandons | Included in the rate; an organization sets a short-abandon threshold, which can be left out of service level |
Documentation read November 12, 2026.
How do you calculate call abandonment rate?
Divide the calls that hung up in queue by the calls offered to the queue for the same interval, usually a half hour, and multiply by 100. Then write down three choices next to the figure: the short-abandon cutoff, whether callbacks and voicemail count as lost, and whether you divided by calls offered or by calls answered.
Here is one half hour in a single queue:
| Outcome | Calls |
|---|---|
| Entered the queue (calls offered) | 600 |
| Answered by an agent | 528 |
| Hung up while waiting | 60, of which 18 within 10 seconds |
| Took a callback | 8 |
| Sent to voicemail after a timeout | 4 |
Here is the same half hour under four definitions:
| Definition | Calculation | Abandonment rate |
|---|---|---|
| Every hang-up, divided by calls offered | 60 ÷ 600 | 10.0 percent |
| Short abandons removed from both counts | 42 ÷ 582 | 7.2 percent |
| Callbacks and voicemail counted as lost | 72 ÷ 600 | 12.0 percent |
| Hang-ups divided by calls answered | 60 ÷ 528 | 11.4 percent |
A center reporting 7.2 percent and another reporting 12.0 percent could be running identical queues. Inside your own reports, the risk runs the other way: switch the short-abandon cutoff or start counting voicemail as handled and the trend line moves with no change in what callers went through. Fix the definition once, print it on the report, and keep a second count of the callers sent to voicemail or a callback, because the first definition hides them.
Abandonment also reads best beside service level and average speed of answer, which our call center metrics guide works through on one half hour of figures.
The cutoff for short abandons varies by platform and by setting, so pick one and keep it. Those calls are real, though. Some are wrong numbers, and some are callers who heard there was a wait and decided against it, so report them in their own column rather than deleting them.
When do callers hang up, and what does it tell you?
The count of abandoned calls tells you how many callers you lost, and the time they waited before hanging up usually tells you why. Pull each abandoned call's wait in seconds, group the waits into bands and read the shape. A queue whose hang-ups cluster in the first few seconds has a different problem from one whose callers give up after three minutes.
A study of caller patience at an Israeli bank's call center, using a year of 1999 call records, found two peaks in the rate at which waiting callers hung up (Brown et al., Journal of the American Statistical Association, 2005). The first came a few seconds in, when a "please wait" message played for the first time and callers who did not want to wait probably realized they were in a queue. The second came at about 60 seconds, when the system played the message again. The authors concluded that the message "increases customers' likelihood of hanging up for a brief time thereafter," which may work against its purpose.
Use this as a starting point for your own data:
| When the hang-ups cluster | Likely cause to check | Look at first |
|---|---|---|
| Within a few seconds of entering the queue | Callers learning there is a wait and leaving, plus wrong numbers | What the greeting says, and a fixed short-abandon cutoff for reporting |
| Just after a recorded message plays | The message itself | What it says and how often it repeats |
| Spread through long waits in a few half hours | Too few people on the phones for those intervals | Staffing against the forecast for those half hours, and shrinkage |
| Mostly in one call reason or menu option | Callers with less patience for that request, or another way to get it done | Abandonment split by call reason |
| Rising week after week at the same staffing | More calls, or longer ones | The trend in calls offered and in average handle time |
A time-to-abandon worksheet
Export one row per queued call for the period, with the wait in seconds in column B and the outcome in column C (answered, abandoned, callback or voicemail). Put the start of each band in column E and its end in column F, then fill these formulas down:
| Measure | Formula, for calls in rows 2 to 601 |
|---|---|
| Abandonment rate | =COUNTIF(C2:C601,"abandoned")/COUNTA(C2:C601) |
| Rate without abandons under 10 seconds | =COUNTIFS(C2:C601,"abandoned",B2:B601,">=10")/(COUNTA(C2:C601)-COUNTIFS(C2:C601,"abandoned",B2:B601,"<10")) |
| Average time to abandon, in seconds | =AVERAGEIFS(B2:B601,C2:C601,"abandoned") |
| Abandons in one band | =COUNTIFS(C$2:C$601,"abandoned",B$2:B$601,">="&E2,B$2:B$601,"<"&F2) |
Bands of 0 to 10, 10 to 30, 30 to 60, 60 to 120, 120 to 300 and 300 seconds or more work for most queues. Add a band edge at each point where a recorded message plays, so you can see whether hang-ups jump right after it. Run the worksheet for your worst half hours separately from the whole day, because a daily average blends the intervals you need to fix with the ones that are fine.
What is a good call abandonment rate?
COPC, which publishes performance standards for customer experience operations, puts the abandonment rate most customer service environments can live with at three to five percent, and suggests choosing that target first, then reading the service level that delivers it from your own queue history. The best large centers in the research literature lose 1 to 2 percent.
| Source | Figure | What it covers |
|---|---|---|
| COPC (2024) | 3 to 5 percent | The rate it says most customer service environments can live with |
| Gans, Koole and Mandelbaum (2003) | 1 to 2 percent at most | Large best-practice centers, which the authors call "the exception, rather than the rule" |
| Same paper | 2.8 percent | Daily average across 12 call centers of a large US health insurer, with a 31 second average speed of answer |
| Brown et al. (2005) | About 20 percent | Callers who asked for an agent at the Israeli bank's call center in 1999 |
Sources read November 12, 2026.
COPC adds that the relationship between service level and abandonment differs from one organization to the next, because it depends on how urgent the call is and what else the caller can do. In its airline example, a caller rebooking a flight canceled by weather will wait longer than one asking about seats or meals, who may call another time. So set the target for each queue or call reason rather than for the center as a whole. The service level guide covers how to set the threshold that goes with it.
Why does abandonment jump when a few people are missing?
Abandonment rises with waiting time, and waiting time rises steeply once a queue runs near capacity. Koole and Mandelbaum describe a "remarkably linear relation between the fraction of abandoning customers and average waiting time" (Queueing Models of Call Centers: An Introduction), so a longer average wait shows up almost directly as more hang-ups.
Gans, Koole and Mandelbaum show how little it takes. Fitting a queueing model that allows for abandonment to one half hour at a large center, they estimated that "the absence of only 5 agents (out of the 223 working) would likely result in almost a doubling of both the ASA and the fraction abandoning." Five agents is about 2 percent of that floor.
Plot each half hour's abandonment rate against its average wait from your own data, and the slope tells you roughly what a change in staffing or volume will cost in lost callers. Staff the peaks with the queueing math rather than a ratio: our Erlang C calculator returns the people you need for a service level target. Erlang C assumes every caller waits until answered, so when waits run long it can show more agents than you need, which is the safe side when you are trying to bring abandonment down.
How do you reduce call abandonment rate?
Find the half hours and call reasons where callers give up, then change what happens to them during the wait. The usual levers are staffing those intervals, rewriting announcements that prompt hang-ups, offering a callback, and answering some call types at once with self-service. Count the calls that never reach your queue too, because none of them shows up in the rate.
- Staff the half hours where callers give up. The worksheet shows which ones they are. Move breaks, meetings and training out of those intervals before adding headcount, and read the average speed of answer beside the abandonment rate, since ASA excludes everyone who hung up.
- Rewrite or space out the announcements. If the worksheet shows hang-ups right after a recorded message, change what the message says or how often it plays, then run the worksheet again for the same intervals.
- Offer a callback, and count it honestly. A caller who takes a callback leaves the queue without being counted as abandoned on either platform above. Report callbacks requested and callbacks completed beside the abandonment rate, so a callback nobody returns doesn't vanish from the numbers.
- Answer the call types that don't need a person right away. An option in the phone menu that resolves a call type immediately takes those callers out of the line before they have a chance to give up. Make sure callers who would rather wait for a person keep their place when they decline it.
- Count the calls that never queue. Calls that arrive when the center is closed usually hear a message and end, and callers who hang up in the menu never enter a queue. Neither appears in the abandonment rate, so track them as separate counts.
That gap can be large. In our own data, 37.6 percent of calls made to business lines arrived outside Monday to Friday, 8am to 6pm, in the caller's local time (call center statistics from 181,555 AI agent calls). A center that closes at 6pm and on weekends would never see those callers in its abandonment rate, because they never entered a queue to abandon.
Is the 3 percent abandonment rule the same thing?
No. The three percent figure is a federal rule for outbound telemarketing, not a benchmark for inbound queues. Under the Telemarketing Sales Rule, an outbound call is abandoned if a person answers and is not connected to a sales representative "within two (2) seconds of the person's completed greeting."
The safe harbor requires technology that keeps abandonment to "no more than three (3) percent of all calls answered by a person, measured over the duration of a single calling campaign, if less than 30 days, or separately over each successive 30-day period" (16 CFR 310.4, read November 12, 2026). The denominator is calls answered by a person, and the window is the campaign. How that ceiling limits dialer pacing is covered in predictive dialer vs AI agents.
Taking calls out of the queue with an AI agent
A call an AI agent answers never waits in your queue, so it can't be abandoned there. Telvana's agents answer immediately, at any hour, and most centers start with one new line in the phone menu, offered when the wait climbs: "Press 1 to talk to our AI assistant now." The option transfers callers to the agent's number, and callers who would rather wait for a person keep their place in line (hold-queue relief).
Nights, weekends and spikes can be forwarded to the agent as overflow, which covers the calls that never reached your queue (after hours and overflow). When a caller needs a person, the agent transfers the call to the number you choose, under rules you set. Every call is recorded, transcribed, summarized and tagged by outcome, so you can set the calls it finished against the queue's abandonment rate before and after. Plans start from $399 a month, based on call volume, plus a one-time setup fee.
Frequently asked questions
How do you calculate call abandonment rate?
Divide the calls abandoned in queue by the calls offered to the queue in the same interval and multiply by 100. If 600 calls entered the queue in a half hour and 60 hung up before an agent answered, the rate is 10 percent. Note whether you excluded short abandons and whether callbacks and voicemail count as abandoned.
What is a good call abandonment rate?
COPC puts the rate most customer service environments can live with at 3 to 5 percent, and large best-practice centers in the research literature lose 1 to 2 percent. The right target depends on the call: callers with an urgent problem wait longer than callers with a casual question, so set it per queue or call reason.
What is a short abandoned call?
A short abandon is a caller who hangs up within a cutoff of a few seconds after entering the queue. Platforms let you set the cutoff, and Genesys Cloud includes short abandons in its abandon rate. Report them in their own column rather than deleting them, since some are wrong numbers and some are callers who decided not to wait.
Do callbacks count as abandoned calls?
Usually not. Amazon Connect does not count contacts queued for callback as abandoned, and Genesys Cloud treats callbacks and transfers to voicemail as flow-outs rather than abandons. Track callbacks requested and callbacks completed beside the abandonment rate, so callers who were never called back still show up somewhere.
What is the maximum permissible call abandonment rate?
For outbound telemarketing in the US, the Telemarketing Sales Rule's safe harbor allows abandonment of no more than three percent of calls answered by a person, measured per calling campaign or per 30-day period. An outbound call counts as abandoned if the person isn't connected to a sales representative within two seconds of their completed greeting.
How do you calculate call abandonment rate in Excel?
With calls offered in B2 and calls abandoned in C2, use =C2/B2 and format it as a percentage. To exclude short abandons, put their count in D2 and use =(C2-D2)/(B2-D2). With one row per call, =COUNTIF(C2:C601,"abandoned")/COUNTA(C2:C601) returns the rate from the outcome column.
See which of your calls could skip the queue
Book a demo and bring your interval report and your top call types. You'll hear a Telvana agent handle one of those calls, and we can look at which half hours lose the most callers.