- Manish Jain
View
Share
Most inbound contracts promise a call center service level, such as 80% of calls answered within 20 seconds. Yet few buyers see what that number does to staffing, or what a tighter target would cost. So the choice often comes down to habit. This guide models the trade-off with standard queueing math. It shows how many agents each target needs and what callers gain. It also explains why forecast accuracy and contract wording matter more than the headline number.
We wrote it for operations, finance, and customer experience leaders who set or buy inbound service levels.
What a Service Level Actually Promises
A service level has two parts: a share of calls and a time limit. For example, 80/20 means 80% of calls answered within 20 seconds. The remaining 20% can wait much longer, so the target says nothing about the worst experiences. That is why it helps to track average speed of answer and abandonment alongside it.
The standard way to size a queue for a service level is the Erlang C model. A widely cited review of call center research by Gans, Koole, and Mandelbaum explains the model. It links call volume, handle time, and staffing. Standards bodies also treat service levels as something to define and agree with the client. ISO 18295-1, the international standard for customer contact centers, sets requirements for doing exactly that.
The Staffing Cost of Each Target
The table models agents needed on the phones for five common targets. It assumes a six-minute average handle time and steady call arrivals. The model also ignores abandonment, so treat the results as illustrative, not as a quote.
| Service level target | Agents at 100 calls/hour | Agents at 200 calls/hour | Agents at 400 calls/hour | Average wait at 200 calls/hour | Agent occupancy at 200 calls/hour |
|---|---|---|---|---|---|
| 70% in 60 seconds | 13 | 23 | 44 | 50 sec | 87% |
| 80% in 20 seconds | 14 | 25 | 46 | 15 sec | 80% |
| 90% in 30 seconds | 15 | 26 | 48 | 9 sec | 77% |
| 90% in 10 seconds | 15 | 27 | 49 | 5 sec | 74% |
| 95% in 20 seconds | 16 | 28 | 50 | 3 sec | 71% |
Model limitations: These estimates use Erlang C, assuming random call arrivals, a constant average handle time, and no abandonment. They show the agents needed to meet each target under those assumptions. Real staffing plans also need to account for interval-level demand, handle-time variation, shrinkage, skills, breaks, and scheduling constraints.
However, the results surprise many buyers. Moving from 80/20 to 90/30 adds only one agent at 200 calls an hour, about 4% more staff. At 400 calls an hour, it adds two agents, also about 4%. Meanwhile, the average wait at 200 calls an hour falls from 15 seconds to about 9. In other words, a tighter target often costs less than expected.
Remember that the table counts agents on the phones at the same moment. Scheduled headcount will be higher after you account for breaks, training, absence, and other time off the phones. The final staffing difference between targets will depend on each program’s shrinkage, scheduling constraints, and minimum coverage requirements.
Turning Staffing Into Cost
Agents on the phones are not yet a budget. To compare targets in dollars, use one simple formula:
Incremental monthly cost = additional scheduled FTEs x fully loaded monthly cost per FTE
Getting to additional scheduled FTEs takes three steps. First, model the extra agents each interval needs at the tighter target, because the gap changes through the day. Second, convert those agent-hours into scheduled FTEs by allowing for shrinkage, meaning breaks, training, absence, and meetings. Third, apply your scheduling rules and minimum coverage, which can add or remove whole positions.
For example, take a line open 60 hours a week, with 40-hour FTEs and 30% shrinkage. Keeping one extra agent on the phones for every open hour then needs about 2.1 scheduled FTEs. Multiply that by the fully loaded monthly cost for your delivery model. The result is the price of the tighter target. These assumptions are illustrative, so replace them with your own.
Why Forecast Accuracy Matters More Than the Target
The bigger risk is not the target you pick. It is staffing to a forecast that turns out wrong. The table below keeps staffing fixed at the 80/20 level and lets volume run higher than planned.
| Line size | Actual volume | Calls answered in 20 seconds | Average wait |
|---|---|---|---|
| 100 calls/hour (14 agents) | On forecast | 86% | 16 sec |
| 100 calls/hour (14 agents) | 10% above forecast | 74% | 36 sec |
| 100 calls/hour (14 agents) | 20% above forecast | 57% | 87 sec |
| 200 calls/hour (25 agents) | On forecast | 84% | 15 sec |
| 200 calls/hour (25 agents) | 10% above forecast | 63% | 52 sec |
| 200 calls/hour (25 agents) | 20% above forecast | 26% | 4.7 min |
| 400 calls/hour (46 agents) | On forecast | 81% | 16 sec |
| 400 calls/hour (46 agents) | 10% above forecast | 39% | 2.0 min |
| 400 calls/hour (46 agents) | 20% above forecast | Queue keeps growing | Not stable |
A 10% forecast miss at 200 calls an hour drops the service level from 84% to about 63%. The average wait rises from 15 seconds to about 52. At 20% above forecast, waits stretch to several minutes. And at 400 calls an hour, the queue stops being stable at all. Larger lines are the most exposed, because they run at higher occupancy with less slack. So a precise target staffed against a weak forecast delivers far less than a modest target staffed accurately.
Why Larger Queues Hit Targets More Cheaply
Also notice that occupancy rises with volume at every target. Larger queues absorb random bursts of calls more easily, so they need less spare capacity per call. The same research review describes this as an economy of scale. It is also why splitting one line into several small queues raises staffing. Our guide to bilingual call routing shows the effect for a separate Spanish line.
How to Choose the Right Service Level
So match the target to what a delay costs on each line, not to habit. The table offers starting points to test against your own data.
| Line type | What a delay costs | Starting point to test |
|---|---|---|
| Sales and reservations | Lost revenue when callers hang up | A tighter target, such as 90/30 or better |
| Retention and cancellations | Lost customers | A tighter target, plus priority routing |
| Urgent service (outages, fraud, clinical) | Risk and complaints | A tighter target, with tracked abandonment |
| General service and account questions | Some frustration; callers often retry | 80/20 or similar |
| Low-urgency or back-office queues | Little, if a callback is offered | A looser target with callbacks |
Then check the result against abandonment and repeat calls. If abandonment stays high at your target, the target is too loose for that line. On the other hand, idle agents and near-zero waits mean you are paying for capacity you do not need.
Service Level Terms to Get Right in Your Contract
Still, two contracts with the same 80/20 target can deliver very different experiences. The wording below decides which one you get.
- Measurement interval: A daily average can hide bad half-hours. Measure by interval, and set a floor for the worst intervals.
- Abandoned calls: State whether abandoned calls count against the target, and define any short-abandon threshold.
- Forecast responsibility: Agree who forecasts, how often, and what happens when volume runs above the forecast range.
- Remedies: Tie credits or penalties to sustained misses, not single intervals.
- Reporting: Require interval-level reports with service level, average wait, and abandonment together.
Our guide to writing a call center RFP shows how to build these terms into a vendor evaluation. For written channels, see our guide to email response time standards.
How an Outsourcing Partner Helps
Hitting a service level every interval therefore takes accurate forecasting, flexible scheduling, and the ability to add agents quickly. That is hard for a small in-house team covering a single line. A partner running inbound call center services can pool capacity and staff to interval-level forecasts. It can also report against the contract terms above. For quality, the same reports should sit alongside your call center quality assurance results.
Planning an Inbound Support Program?
Compare service level targets, staffing requirements, and reporting expectations before finalizing your outsourcing plan. Discuss your expected call volumes, average handle time, and operating hours with SkyCom.
Conclusion
A call center service level is a business decision, not a default. In the model, tighter targets often add only a few agents. Yet a 10% forecast miss can undo the target entirely. So choose targets by line, staff them to accurate forecasts, and write contracts that measure every interval. That combination protects callers and keeps staffing costs predictable.
Frequently Asked Questions
What is a good call center service level?
It depends on what a delay costs on each line. 80/20 is a common starting point for general service. However, sales, retention, and urgent lines often justify a tighter target. Test any target against abandonment and repeat calls.
What does 80/20 mean in a call center?
80/20 means agents answer 80% of calls within 20 seconds. It says nothing about how long the other 20% wait, so track average wait and abandonment alongside it.
How much more does a 90/30 service level cost than 80/20?
In our illustrative model, 90/30 needs about 4% to 7% more agents on the phones than 80/20. Your figure depends on handle time, volume, and scheduling.
Why do call centers miss service level targets?
Most misses come from volume running above forecast, short-term absence, or longer handle times. Even a 10% forecast miss can drop service level sharply, so forecast accuracy matters as much as staffing.
How should service level be measured in a contract?
Measure by interval, such as every 30 minutes, rather than as a daily average. Also define how abandoned calls count, and report service level with average wait and abandonment.
