Contact centre metric

Call abandonment rate

Q: What is a good call abandonment rate for a contact centre?

Industry benchmarks for call abandonment rate: under 3% is excellent; 3–5% is acceptable for most operations; 5–8% indicates a staffing or routing problem; above 8% is poor and indicates persistent understaffing or badly designed IVR routing. Regulated sectors (financial services, NHS) have tighter expectations — the FCA has cited abandonment rates above 5% as indicators of poor consumer support outcomes.

Q: What is the difference between call abandonment rate and service level?

Service level measures the percentage of calls answered within a target time (e.g. 80% within 20 seconds). Abandonment rate measures the percentage of callers who give up waiting before being answered. They are related but different: you can have a good service level (80%) and still have high abandonment if a minority of callers wait very long. You can also have low abandonment and poor service level if callers are patient. Both metrics are needed for a complete picture of queue performance.

Q: Why does Erlang C not model call abandonment?

Erlang C assumes infinite patience — every caller waits until an agent is available. In reality, callers have a patience distribution: most will wait for the first 20–30 seconds, then the probability of abandonment increases steadily with wait time. The Erlang A model (also called the M/M/N+M queue) adds a patience parameter to Erlang C, producing more realistic estimates for abandonment and ASA under different staffing levels. In practice, most contact centres use Erlang C for staffing and track abandonment separately as a KPI.

Q: Does a callback / virtual queue eliminate abandonment?

A callback offer transforms abandonment from a lost contact into a deferred contact — the caller requests a callback instead of waiting. This effectively eliminates queue-driven abandonment for the callers who accept the offer (typically 30–60% of callers when wait time exceeds 2 minutes). However, it does not reduce the staffing requirement — the contact still needs to be handled. Callbacks also shift contacts from real-time to deferred, which affects intraday planning: callbacks accumulate and need to be worked in addition to new inbound volume.

Call abandonment rate is the percentage of inbound contacts that disconnect before being answered by an agent. It is the most direct measure of queue performance and the first indicator that staffing levels are failing — callers vote with their feet.

Formula and definition

Abandonment rate formula

A = (Calls abandoned ÷ Calls offered) × 100

Calls abandoned — contacts that disconnected while waiting in queue (not during IVR routing)

Calls offered — total contacts that entered the queue (after IVR routing, before agent answer)

Worked example

1,000 calls offered · 55 abandoned = 5.5% abandonment

IVR drop-off vs. queue abandonment:Many ACD systems report "abandonment" including contacts that dropped out during IVR routing — before reaching a human queue. These are fundamentally different problems (navigation failure vs. staffing failure) and should be reported separately. Use your ACD reporting to distinguish queue-phase abandonment from IVR-phase drop-off.

Industry benchmarks

< 3%ExcellentWell-staffed operation with short queue times

3–5%AcceptableTypical for good contact centres; minor peaks cause occasional abandonment

5–8%PoorStructural staffing gap or IVR routing problem

> 8%Very poorPersistent understaffing; regulatory risk in some sectors

Benchmarks vary by sector. Financial services and healthcare operations are typically held to the lower end of acceptable (<3–5%) due to regulatory and patient safety expectations. Outbound blended operations may report higher apparent abandonment because outbound contacts are sometimes included in queue metrics.

Abandonment, service level, and ASA

Abandonment, service level, and ASA are all different views of the same underlying queue. They are correlated but not identical — you need all three to get a complete picture.

Service level

% of calls answered within Ts

As SL improves (more agents, shorter wait), abandonment falls. Inversely correlated. The most useful metric for staffing decisions.

ASA

Mean wait across all answered calls

ASA is a lagging indicator — it can look acceptable even if a tail of callers waited very long. Abandonment captures those callers; ASA misses them (they never answered).

Abandonment

% of queued contacts that hung up

A leading indicator of staffing failure. Rises steeply when SL drops below roughly 70–75%. At 80/20 SL, abandonment is typically 2–5%; at 60% SL, abandonment is often 10–15%+.

The abandoned caller problem: When a caller abandons, most ACD systems do not count that call in SL (since it was never answered). This means your SL% can look acceptable while abandonment is rising — the abandoned calls exit the denominator. Always report SL and abandonment together to prevent gaming.

Why Erlang C doesn't model abandonment

Erlang C assumes every caller waits indefinitely — infinite patience. In the real world, callers have a patience distribution and abandon at increasing rates as wait time grows.

The patience distribution

Studies consistently show that abandonment probability is low in the first 20–40 seconds (callers accept a short wait), then rises steeply between 40–120 seconds, then plateaus as only the most patient callers remain. The exponential decay model is: P(abandonment by time t) = 1 − e^(−t/θ), where θ is the mean patience (typically 60–180 seconds for inbound voice queues).

Erlang A: the model that includes abandonment

The Erlang A model (M/M/N+M queue) adds a patience parameter to Erlang C. It produces more realistic estimates for abandonment and ASA under different staffing levels. In practice, most contact centres use Erlang C for staffing requirements (because it is conservative — it slightly overstates agent needs vs. Erlang A) and track abandonment as a separate operational KPI from ACD reports.

Practical implication

If your Erlang C model says you need 15 agents to hit 80/20 SL, and you staff 15 agents, abandonment should be approximately 3–5% under typical patience assumptions. If your actual abandonment is higher at the same staffing level, your callers are less patient than average — or your ACD is mis-classifying IVR drop-off as queue abandonment.

Five levers to reduce call abandonment

Improve staffing to SL target

Impact: HighEffort: Medium

The primary driver of abandonment is wait time. Erlang C gives the minimum seated agents to meet your SL target — hitting that target reduces abandonment directly. For every percentage point improvement in SL (e.g. 75% → 80% in 20s), abandonment typically falls 0.5–1.5 percentage points.

Offer a callback / virtual queue

Impact: HighEffort: Low

Give callers the option to receive a callback rather than waiting. 30–60% of callers who would otherwise abandon accept a callback when wait times exceed 90–120 seconds. Abandonment falls immediately but the underlying contact volume is unchanged — callbacks must be resourced.

Publish expected wait time on IVR

Impact: MediumEffort: Low

Callers with no wait time information abandon faster than callers who know they will wait 3 minutes. Publishing a realistic estimated wait time reduces premature abandonment by setting expectations. Unrealistic (optimistic) wait time estimates backfire — they cause anger-driven abandonment late in the queue.

Improve digital self-service deflection

Impact: MediumEffort: High

Contacts that are deflected to web, app, or IVR self-service never enter the voice queue. Every 1% volume reduction from deflection reduces abandonment because the queue shortens. Focus deflection on high-volume, low-complexity contact reasons (account balance, status update, booking confirmation).

Improve IVR routing efficiency

Impact: MediumEffort: Medium

Abandoned callers who drop out in the IVR routing tree (before reaching a queue) inflate raw abandonment figures but represent a different problem from queue-driven abandonment. Audit your IVR flow for dead-ends, excessive menu depth, and options that do not match caller intent. An IVR audit typically reduces IVR-phase drop-off by 20–40%.

Call abandonment questions

What is a good call abandonment rate for a contact centre?

Under 3% is excellent; 3–5% is acceptable for most operations; 5–8% indicates a staffing or routing problem; above 8% is poor and indicates persistent understaffing. Regulated sectors (financial services under FCA Consumer Duty, NHS) typically hold to tighter expectations of under 3–5%.

What is the difference between call abandonment rate and service level?

Service level measures the percentage of calls answered within a target time. Abandonment measures the percentage of callers who gave up before being answered. They are correlated but different: you can have a good service level and still have moderate abandonment if a long tail of callers waits past the target. Report both together — SL alone can miss a rising abandonment problem.

Why does Erlang C not model call abandonment?

Erlang C assumes infinite patience — every caller waits until answered. Real callers have a patience distribution and abandon at increasing rates as wait time grows. The Erlang A model adds a patience parameter and produces more accurate abandonment estimates. In practice, most operations use Erlang C for staffing (it is slightly conservative) and track abandonment from ACD reports separately.

Does a callback / virtual queue eliminate abandonment?

A callback offer transforms abandonment into a deferred contact — the caller requests a callback instead of waiting. 30–60% of callers accept when wait time exceeds 90–120 seconds. This eliminates queue-driven abandonment for those callers but does not reduce the total contact volume. Callbacks also shift contacts from real-time to deferred, which must be planned in intraday scheduling.