Free tool Abandonment rate calculator and formula
Abandonment rate is the share of contacts that joined the queue and left before anyone answered. Calculate it from your report, see what excluding short abandons does, or estimate it ahead of time with Erlang A — the queueing model that, unlike Erlang C, lets people hang up.
- Report or model
- Abandonment by headcount
- Runs locally
Abandonment rate
All numbers are sample figures. Use one interval and one queue at a time; mixing a quiet morning with a busy evening hides where people actually give up.
Calls or chats that entered the queue in the period.
Left before an agent answered, including the short ones.
Hung up within the threshold below, often before hearing the greeting.
Only used in the summary text. Common choices are 5 or 10 seconds; use whatever your report uses.
Including the time abandoners waited before leaving. Lets the calculator estimate average caller patience.
Calls arriving in one interval.
Seconds, or m:ss. Talk + hold + after-call work.
Logged in and taking calls in that interval.
How long callers wait, on average, before giving up if nobody answers. See the FAQ for how to estimate it.
For the "answered within" figure.
Shows the fewest agents that keep abandonment at or below this.
All attempts, including the ones that got a busy signal.
How long a call occupies a line. Seconds, or m:ss.
e.g. 1 for 1%.
Abandonment rate
7.0%
Abandoned ÷ offered.
Excluding short abandons
5.1%
Short ones removed from both sides.
Answered
2,232
Offered − abandoned.
Answer rate
93.0%
Estimated average patience (m:ss)
8:20
Average queue time ÷ abandonment rate.
Abandonment (Erlang A)
—
Share of all arrivals.
Abandoned calls per interval
—
Answered within threshold
—
Of all arrivals; abandons count as missed.
ASA of answered calls (s)
—
Had to queue
—
Occupancy
—
Handled workload ÷ agents.
Agents for the target
—
Fewest agents at or below it.
Blocking probability
—
Erlang B.
Blocked calls per interval
—
Carried traffic (Erlang)
—
Lines busy on average.
Lines for the target
—
Fewest lines at or below it.
Agents vs. abandonment
Green meets the target. The last column is what Erlang C predicts for the same staffing if nobody hung up.
Lines vs. blocking
Green meets the target.
Share these inputs: the values go after the # in the link, so they never reach a server.
https://tapcub.com/tools/abandonment-rate-calculator
Version v1.0Updated Update history
What the numbers tell you
Abandonment is the most visible sign of understaffing, and the easiest to misread when definitions differ between reports.
Abandonment rate
Abandoned ÷ offered, where offered is everything that entered the queue. If your system calls it "abandoned ÷ (answered + abandoned)", it is the same thing as long as every queued contact ends one of those two ways.
Short abandons
Callers who leave within a few seconds often misdialled or got their answer from the greeting. Many centres remove them from both the numerator and the denominator. Always say which version you report.
Erlang A estimate
Given agents, handle time and average patience, the model predicts the share who give up and how many are answered within the threshold. Abandoners count as not answered.
Erlang C vs Erlang A
Erlang C assumes infinite patience, so with a tight team it predicts long queues and no abandons. Real callers leave, which shortens the wait for those who stay — Erlang A captures that.
There is no single "normal" abandonment rate. It depends on the channel, what callers want and what the queue tells them while they wait. Compare against your own history and against your service level target, not against a number from somewhere else.
From a queue report to a staffing decision
Start with what happened, then use the model to see what more or fewer agents would change.
-
1
Measure
Take offered, abandoned and short abandons for one queue and one interval from your phone or chat report. Check how the report defines each one.
-
2
Estimate patience
Add the average queue time of all contacts, abandoners included. The calculator divides it by the abandonment rate to estimate average patience.
-
3
Model
Switch to Erlang A. Enter calls per interval, handle time, agents and that patience, then set a target abandonment rate.
-
4
Decide
Read the agents table. Each extra agent usually cuts abandonment by less than the one before, so put the line where the cost of one more person matches what you lose to abandons.
Abandonment rate, Erlang A and blocking
The direct rate is a count. The model treats arrivals as Poisson, handle times and patience as exponential, and serves calls first come, first served (Garnett, Mandelbaum & Reiman, 2002).
- Direct: abandonment rate = abandoned ÷ offered; excluding short abandons = (abandoned − short) ÷ (offered − short)
- Erlang A (M/M/N+M): N agents, arrival rate λ, service rate μ = 1 ÷ AHT, abandonment rate θ = 1 ÷ average patience
- P(abandon) = θ × average queue length ÷ λ, which also gives P(abandon) = θ × average wait — used here to estimate patience from data
- Erlang B: no queue; a call that finds every line busy is blocked. That is a different event from abandoning
Abandonment rate
abandoned ÷ offered
168 ÷ 2,400 = 7.0%
Excluding short abandons
(abandoned − short) ÷ (offered − short)
(168 − 48) ÷ (2,400 − 48) = 5.1%
Average patience (estimate)
average queue time ÷ abandonment rate
35 s ÷ 0.07 = 500 s
Workload
A = calls × AHT ÷ interval
100 × 180 s ÷ 1,800 s = 10 Erlang
Blocking (Erlang B)
B(N, A) = (A^N ÷ N!) ÷ Σₖ (A^k ÷ k!)
20 lines, 15 Erlang → 4.56%
Same queue, different headcount
100 calls per 30 minutes, 3-minute handle time (10 Erlang), callers wait 2 minutes on average before giving up, threshold 20 seconds. Sample figures.
| Agents | Abandon (Erlang A) | Answered in 20 s (Erlang A) | Answered in 20 s (Erlang C) | Occupancy (Erlang A) |
|---|---|---|---|---|
| 10 | 13.7% | 64.0% | queue never clears | 86.3% |
| 11 | 9.4% | 74.2% | 39.0% | 82.3% |
| 12 | 6.2% | 82.6% | 64.0% | 78.2% |
| 13 | 3.8% | 88.9% | 79.6% | 74.0% |
| 14 | 2.3% | 93.3% | 88.8% | 69.8% |
With 11 agents Erlang C says only 39% are answered within 20 seconds; Erlang A says 74%, because the 9% who leave make the queue shorter for everyone behind them. A low abandonment rate needs the same staff as a good service level — the two are measuring the same queue.
Where abandonment numbers go wrong
Most disputes about abandonment come from definitions, not from arithmetic.
Comparing rates with different definitions
One report removes short abandons, another counts calls abandoned in the IVR, a third excludes callbacks. Write the definition next to the number.
Averaging over a whole day
A 4% daily rate can hide 15% at the lunchtime peak. Look at it by interval, the same way you staff.
Using Erlang C to predict abandonment
Erlang C has no abandonment in it. It overstates waits when people do leave and cannot tell you how many will.
Calling blocked calls abandoned
A busy signal means the call never reached the queue. That is blocking (Erlang B); count it separately.
Fewer chats left waiting
TapCub live chat is built around not leaving a visitor in an empty queue. What it has today:
Alerts on new waiting chats
Desktop, email, WeCom, DingTalk, Feishu, Slack or webhook — the moment a new conversation is waiting.
Learn moreWorking hours and offline messages
Outside the hours you set, the widget collects a message and an email instead of letting a visitor wait for nobody.
Learn moreAI answers first
Answers from your knowledge base, labelled as AI, and hands over to the queue when it is unsure or the visitor asks for a person.
Learn moreAbandonment rate FAQ
Still have a question?
Chat with the team behind TapCub — we usually reply within a few hours on business days.
How do you calculate abandonment rate?
Divide the contacts that left the queue before being answered by all contacts that entered the queue in the same period: 168 abandoned out of 2,400 offered is 7.0%. Do it per queue and per interval.
Should short abandons be excluded?
Many centres exclude callers who hang up within the first few seconds, because they usually never meant to wait. If you do, remove them from both the top and the bottom of the fraction, and say what threshold you used.
What is a good abandonment rate?
There is no universal figure. A sales line, a bank fraud line and a chat widget on a store have very different patience. Set a target from your own history and from what a lost contact costs you, then staff to it with the Erlang A view.
What is Erlang A?
A queueing model (M/M/N+M) that extends Erlang C with customer patience: anyone still waiting gives up at a constant rate. It predicts abandonment as well as waiting, and is usually closer to reality than Erlang C for a tight team.
How do I estimate average patience?
If patience is exponential, abandonment rate = average wait ÷ average patience. So average patience ≈ average queue time of all contacts (abandoners included) ÷ abandonment rate. Enter both numbers in "From my report" and the calculator does it.
Is blocking the same as abandonment?
No. Blocking happens when every line is busy and the call gets a busy signal or a "please call later" message — it never queues. Erlang B gives that probability. Abandonment is a caller who queued and then gave up.
Does this work for live chat?
The direct rate does: chats that ended before an agent joined ÷ chats that entered the queue. The Erlang A model assumes one contact per agent; for agents handling several chats at once, use the chat staffing calculator.
Update history
New features, fixes and data refreshes for this tool, newest first.
-
· v1.0
LaunchedFirst release.
Missing something or found an error? Tell us — requests that come up often go into the next update. All tool updates
Turn the agent count into a schedule
Once you know how many agents keep abandonment down, add breaks, meetings and leave to get the number to roster.