Grade of Service (GoS)
Definition
Grade of Service (GoS)
Grade of Service (GoS) is the probability that a call is blocked or delayed past an acceptable wait during the busiest hour of the day. Expressed as a decimal, it shows how often a caller hits a busy tone or an overflow queue.
A GoS of 0.02 means two of every 100 busy-hour calls fail to connect on the first attempt. Planners use the figure to size trunks, agents, and lines against forecast traffic, week by week.
Unlike service level, which times answered calls, GoS counts the attempts the network could not accept at all.
That distinction matters when you are sizing infrastructure rather than headcount — trunks and bandwidth get bought months ahead, while agents can be added next week.
Key takeaways
- GoS is the ratio of lost calls to total offered calls during the busy hour.
- The industry benchmark for voice contact centers sits between 0.01 and 0.05.
- Erlang B and Erlang C convert forecast traffic and a target GoS into the trunks or agents you need.
- A rising GoS usually signals under-provisioned lines, not slow agents.
- One Erlang equals one hour of continuous call activity, so 38 Erlangs of busy-hour traffic needs at least 38 channels before any headroom.
How it works
GoS equals the calls lost during the busy hour divided by the calls offered in that same hour. A result of 0.03 means three of every 100 attempts were blocked. Most centers target 0.02 or lower on customer-facing lines.
The formula pairs with Erlang B, a queuing model that estimates blocking probability from traffic volume in Erlangs and the number of available channels.
One Erlang equals one hour of continuous call activity under the ITU-T E.600 traffic engineering standard. So 30 Erlangs of busy-hour traffic through 35 trunks yields a calculable blocking rate.
Erlang C is the sibling model — it assumes blocked callers queue instead of hanging up, which makes it the right tool for agent staffing while Erlang B sizes the trunks.
| GoS value | Blocked calls per 100 | Typical use case |
|---|---|---|
| 0.01 | 1 | Emergency and premium support lines |
| 0.02 | 2 | Standard customer service queues |
| 0.03 | 3 | Peak-season general banking queues |
| 0.05 | 5 | Internal or low-priority queues |
| 0.10 | 10 | Overflow-tolerant outbound campaigns |
The busy hour is not a guess. Pick the 60-minute window with the highest offered traffic across a representative period, usually four to six weeks, and size against that window rather than a daily mean.
Every step down in GoS costs money. Going from 0.05 to 0.01 on the same traffic needs more trunks and more idle capacity, so planners reserve the tightest targets for lines where a blocked call has a real price.
Planners feed forecast Erlangs, a target GoS, and average handle time into their planning models.
Workforce management suites turn that output into a roster, telling you how many trunks and agents to schedule in each half-hour interval.
Watch the busy hour, not the daily average. A queue averaging 0.01 across the day can still sit at 0.06 through a 30-minute spike, and those are the callers who complain.
Examples
GoS shows up wherever call traffic meets finite capacity. Three cases across banking, offshore technical support, and insurance show how planners pick a target, then live with the trade-off between blocked calls and the cost of idle trunks.
A retail bank running a 200-seat call center targets 0.01 on its fraud hotline and 0.03 on general banking.
During the December 2024 holiday peak the fraud line held at 0.008, fewer than one blocked call in 100. The general queue drifted to 0.06, six in 100, and triggered an emergency trunk order.
A Manila-based Business Process Outsourcing (BPO) provider supports a United States software client.
It provisions 45 session initiation protocol trunks against a forecast of 38 busy-hour Erlangs, seven trunks of headroom. Erlang B math predicts a GoS of 0.019, just inside the 0.02 service level agreement (SLA).
Push that same 38 Erlangs through 40 trunks instead and the blocking rate climbs, which is why the provider pays for seven spare circuits it hopes never to use.
An insurance claims line runs inbound call center operations through storm season.
Planners size for 0.05 rather than 0.02 ahead of the 2025 storm season — accepting five blocked calls in 100 in exchange for lower fixed trunk cost in the quieter months.
When the bank’s general queue hit 0.06, expect abandonment rate to climb as well, because blocked callers redial into an already busy queue.
Related terms
GoS overlaps with several contact center metrics but measures a distinct failure mode — capacity loss, not agent speed. The cluster below covers the metrics that sit either side of it in a weekly operations review.
- Service Level: the share of answered calls picked up within a target time, usually 80 percent in 20 seconds.
- Average Handle Time: mean duration of a call including talk, hold, and after-call work.
- Abandonment Rate: the share of callers who hang up before an agent connects, downstream of a bad GoS.
- Call Center: the operational unit whose capacity GoS measures.
- Workforce Management: the planning discipline that turns GoS targets into schedules.
- Inbound Call Center: the received-call operation where blocking shows up first.
FAQ
Five questions come up whenever a planner sets a GoS target: what good looks like, how it differs from service level, what breaks it, how to calculate it, and whether it means anything on digital channels.
What is a good grade of service?
A GoS between 0.01 and 0.02 is the accepted benchmark for customer-facing voice lines, per CX Central. Internal and overflow queues tolerate 0.05 without hurting the customer experience.
How is grade of service different from service level?
GoS measures the share of calls the system blocks outright during the busy hour. Service level measures how quickly answered calls get picked up. A center can hit its service level target and still fail its GoS if trunks are undersized.
What causes a poor grade of service?
Under-provisioned trunks, unexpected traffic spikes, and misconfigured routing rules are the three usual culprits. Agent shortages show up in service level or abandonment first, not in GoS. Carrier faults are a fourth cause you cannot schedule around.
How do you calculate grade of service?
Divide the calls lost during the busy hour by the total calls offered in that same hour. For forward planning, feed forecast Erlangs and a target GoS into an Erlang B calculator to size trunks.
Does GoS apply to digital channels?
The classical formula is a voice-network metric, so chat, email, and social queues use analogues such as concurrency limits and offered-versus-served ratios, though the underlying idea of capacity-driven loss still applies.
Ready to size your staffing against a target GoS? Explore the OA outsourcing hubs to shortlist vetted contact center partners.
Independent
