Expected wait time
Definition
Expected wait time
Expected wait time (EWT) is the forecast a contact centre gives a caller about how long they’ll wait in queue before a live agent answers the line. Centres use it to set caller expectations, route traffic, and staff the floor against real demand.
The metric matters because callers tolerate a known wait far better than an unknown one. A 2024 Zendesk Customer Experience Trends report found 73% of consumers switch to a rival after several bad service experiences.
EWT is one of three closely watched queue metrics, alongside average speed of answer and average handle time. Together they form the operating dashboard for any voice or chat queue.
Wait time is also a published number in some sectors. Telecoms regulators report carrier call-answering performance, so those figures reach the public, while most retail and banking queues stay behind the vendor’s own dashboard.
Key takeaways
- EWT forecasts a caller’s queue time from queue position, average handle time, and live agent availability.
- Under 30 seconds is excellent for voice; under two minutes is still tolerable.
- Three levers cut EWT: sharper routing, better forecasting, and callback or virtual queueing.
- Callers abandon roughly 5% to 10% of calls once the wait crosses two minutes.
- Most centres announce EWT through an interactive voice response (IVR) prompt or a chat-window timer.
How it works
Expected wait time is a forecast, not a stopwatch reading. The queueing system divides the callers ahead of you by the rate agents are clearing calls, then adjusts for staffing and recent handle time. It refreshes every few seconds.
Most platforms run a variation of the Erlang C formula, the same maths telecoms have used since the 1910s to size trunk lines, refined with live agent data.
The International Telecommunication Union, the United Nations agency that sets telecoms standards, still publishes Erlang reference tables for capacity planning.
| Input | What it means | Typical source |
|---|---|---|
| Callers in queue | Position ahead of the current caller | Automatic call distributor (ACD) live counter |
| Average handle time | Mean talk plus after-call work, last 30 days | Workforce management software |
| Active agents | Logged-in agents in an “available” state | ACD live counter |
| Shrinkage allowance | Breaks, training, and absence removed from headcount | Workforce management forecast |
| Callback take-up | Share of queued callers who opt for a return call | ACD queue report |
| Output: EWT | (Callers in queue × average handle time) ÷ active agents | Quoted to the caller |
So a queue with eight callers, a three-minute handle time, and four active agents produces an EWT of six minutes. The IVR announces a friendly rounded version, usually padded by 10–15% so the system under-promises.
Forecast accuracy decays fast in volatile queues. A sudden outage or a marketing send can double arrivals in minutes, and any EWT built on last month’s handle time will understate the wait until the model catches up.
EWT differs from average speed of answer (ASA) in one key way: ASA is measured after the fact, while EWT is predicted live. A centre quoting a four-minute EWT against a one-minute ASA has a broken forecast.
Every extra minute in queue costs money. Abandonment climbs steeply once the wait passes two minutes, and abandoned sales calls rarely come back, which is why finance teams treat EWT as a revenue metric rather than an operations one.
Examples
EWT shows up in every queue you’ve ever joined, but the published benchmarks swing hard by sector and channel. Voice queues run longest, chat sits in the middle, and self-service deflection quietly shrinks both.
Banking — JPMorgan Chase, 2024. The bank’s mobile-app chat quotes an EWT under 45 seconds after a 2023 push moved password resets and card replacements to self-service bots. Voice EWT still runs three to five minutes at weekday peaks.
Telecoms — BT Group, 2023. UK regulator Ofcom published complaint data showing BT’s average voice EWT climbed to nine minutes during a Q1 outage spike. Adding callback as the default queue option pulled it under three minutes by Q4.
Healthcare — Kaiser Permanente, 2024. Kaiser publishes EWT bands on its appointment line: under 90 seconds for routine bookings, under 30 seconds for the advice nurse. Triaging callers by intent inside the IVR pulls urgent EWT down without extra capacity.
Offshore delivery — Manila and Cebu, 2024. Philippine providers tender service-level agreements of 80% of calls answered inside 20 seconds, the legacy 80/20 target. Blended shifts that follow the buyer’s working day make that arbitrage hard to copy onshore.
Channel matters as much as sector. The same team usually posts a shorter EWT on chat, because agents run two or three conversations at once, while voice stays strictly one-to-one and every minute of handle time lands on the queue.
Related terms
EWT sits inside a tight cluster of contact-centre metrics, and reading one without the others is how dashboards get gamed. These six terms define the queue you’re actually managing.
- Average speed of answer: the actual wait callers experienced, measured after the call ends.
- Average handle time: talk time plus after-call work, the denominator behind every EWT forecast.
- Service level: the share of calls answered inside a target threshold, classically 80% in 20 seconds.
- Abandonment rate: the proportion of callers who hang up before an agent answers.
- Interactive voice response: the menu system that quotes EWT and routes traffic to the right queue.
- First call resolution: the share of issues solved on the first contact, which trims repeat traffic.
FAQ
What is a good expected wait time?
For voice support, under 30 seconds is excellent and under two minutes is acceptable. For live chat, under 45 seconds is the working benchmark, per the 2024 Gartner Customer Service Survey. Anything past five minutes triggers measurable abandonment.
How is expected wait time calculated?
Multiply the callers ahead of you by the rolling average handle time, then divide by the count of active agents. The ACD refreshes that number every few seconds from live queue data.
What’s the difference between EWT and ASA?
EWT is the predicted wait quoted to the caller in real time. Average speed of answer is the actual wait measured after the call ends. A wide gap between the two points to a forecasting or staffing problem.
Can EWT be reduced without hiring more agents?
Yes. Tighten IVR routing so callers land in the right queue first time, lift first call resolution to cut repeat traffic, and offer callback so queued callers stop holding a line. Many centres shave 30% to 40% off EWT before adding headcount.
Why do call centres add padding to the announced EWT?
A 10–15% buffer absorbs forecasting error and creates a small positive surprise when the caller connects sooner than promised. Under-promising on wait time consistently beats exact quoting on satisfaction scores, especially in high-volume voice queues.
Does EWT apply to chat and email too?
Live chat quotes EWT directly, while email queues use a slower version called expected response time, measured in hours.
To benchmark or trim your own EWT, compare vetted contact-centre providers in the Outsource Accelerator directory and read their service-level offers side by side.
Independent
