Busy hour

Busy hour is the 60-minute window in a contact centre’s day when inbound volume hits its highest sustained level. Planners use it to size agents, phone trunks, and chat seats, so service-level targets hold even when every queue peaks at once.

The term comes from traffic engineering. Danish mathematician Agner Krarup Erlang first modelled telephone exchange load in 1909, and his formulas still take busy-hour volume as their core input today.

Skip the measurement and the staffing model collapses. Every downstream number — agent headcount, shift patterns, seat licences, even the floor space you lease — traces back to that single 60-minute window.

Modern omnichannel floors complicate the picture. Voice, chat, and email rarely peak together, so most planners now track a separate busy hour per channel and per market rather than one number for the whole site.

Key takeaways

• Busy hour drives staffing maths everywhere, from a 20-seat help desk to a 5,000-seat BPO floor.
• Forecasters average the highest-volume hour across the busiest 10 days of a month, per ITU method.
• Omnichannel operations track separate busy hours per channel, because voice, chat, and email peak at different times.
• Missing the busy hour by 10% of agents pushes abandonment past most BPO service-level agreements.
• Erlang C converts busy-hour volume into agent count, and shrinkage of 25–35% sits on top.

How it works

A workforce planner logs contact volume in 15- or 30-minute increments for at least 30 days, finds the consecutive 60 minutes carrying the highest average load, then feeds that figure into an Erlang C calculator to size agents.

The International Telecommunication Union defines two variants: time-consistent busy hour fixes the same clock window daily, while average daily peak hour lets the peak float. Most BPOs use the fixed version.

Channel patterns diverge sharply. Voice traffic in a US retail account usually peaks late morning Eastern, then again after work, while live chat runs flatter and surges late evening, per Zendesk’s 2024 CX Trends Report.

Email behaves differently again. Queues spike on Monday mornings and after holiday weekends, a backlog pattern the US Bureau of Labor Statistics tracks in its 2024 customer-service occupation outlook.

Forecasters then layer shrinkage on top of the raw Erlang requirement: paid breaks, training, attrition, and unplanned absence. A 25–35% allowance is standard for outsourced call centres running tier-one support.

Two numbers make or break the model: forecast accuracy and adherence. A forecast within 5% of actual volume still fails if only 80% of scheduled agents are on the queue when the peak lands.

Interval-level detail matters more than the hour itself. Planners schedule in 30-minute slots, so a peak that starts at 10:15 needs coverage from 10:00, not from the top of the busy hour.

One more wrinkle: busy hour moves. Daylight saving shifts, product launches, and billing-cycle dates all drag the peak by an hour or more, so most BPO planners re-baseline the window every quarter.

Get the window wrong and the cost shows up twice. Overstaff and you pay idle agents through the whole shift; understaff and you pay service-level penalties plus the churn that follows a bad queue day.

Examples

Large BPO providers publish busy-hour staffing models because the whole commercial case rests on them. Retail, telco, and banking accounts each show a different peak shape, and the site footprint gets built around it.

Concentrix, the world’s largest pure-play customer experience provider, flexed its Manila and Mexico City floors for Black Friday 2023. Its playbook absorbed a 3.4x spike between noon and 1pm US Pacific time, the documented busy hour for e-commerce voice.

Teleperformance, the French-headquartered CX group, described a similar pattern for its UK telco accounts in its 2024 annual report: a 9am to 10am peak from commuters reporting overnight outages.

That single hour justified real estate. The firm added 1,200 seats across South African and Egyptian sites purely to cover it, because the contract’s penalty clause bit on busy-hour service level, not the daily average.

Inside the Philippines, Sutherland and TaskUs run split-shift rosters tied to US East Coast busy hours. Peak demand at 11am New York time lands on a fully staffed Manila floor at midnight local.

Banking shows a third shape. Offshore teams supporting UK current accounts see the peak clustered around pay dates and the first working day of the month, so the busy hour is calendar-driven rather than clock-driven.

Smaller operations feel it just as hard. Take a 40-seat Cebu help desk supporting an Australian SaaS client: its peak lands at 9am Sydney time — 6am in Cebu — so the busy hour sets the entire shift premium.

Related terms

Busy hour sits inside a cluster of contact-centre planning terms. Each one either feeds the calculation, constrains it, or measures what happens when the forecast misses, so it helps to keep the distinctions straight.

• Erlang C: the staffing formula that converts busy-hour volume into an agent count at a chosen service level.
• Average Handle Time: the per-contact duration that, multiplied by busy-hour volume, sets the total workload.
• Service Level Agreement: the contract clause defining how busy hour gets measured and penalised.
• Workforce Management: the wider discipline that turns busy-hour forecasts into published shift schedules.
• Shrinkage: the share of paid time agents aren’t handling contacts, added on top of busy-hour staffing.
• Abandonment Rate: the proportion of callers who hang up during busy hour, the headline failure metric.

FAQ

How do you calculate busy hour?

Pull at least 30 days of contact volume in 15- or 30-minute increments, then find the consecutive 60 minutes with the highest average load. Use that figure as your planning input. Most workforce-management platforms automate this and refresh weekly.

Is busy hour the same as peak hour?

In day-to-day BPO conversation, yes, and the two get used interchangeably. In formal traffic engineering, busy hour follows ITU definitions, while peak hour can mean any local maximum, including a one-off Black Friday surge.

How many busy hours should you plan for?

One per channel, per market. A US-facing voice queue might have one busy hour, a UK email queue another, and a global chat queue a third. Forecasting them separately stops both over- and under-staffing on the same floor.

What service level should you hit during busy hour?

The widely cited default is 80/20: 80% of contacts answered within 20 seconds. The COPC CX Standard treats anything under 70/30 as a structural problem rather than a bad day, which is why BPO contracts price it separately.

Does AI change busy-hour planning?

It softens the curve without removing it, because the residual human queue still peaks and still needs the same Erlang maths.

If your busy-hour maths says you need more seats than you can hire locally, browse vetted providers on the Outsource Accelerator directory.