Contact centre service level targets
80% in 20 seconds is the most common contact centre SL target — but it is a convention, not an optimum. Understanding what different SL targets actually cost in headcount, and what they deliver in customer experience, is the foundation of a rational SLA setting process.
The 80/20 benchmark: where it came from
The 80/20 benchmark (80% of calls answered within 20 seconds) originates from a 1970s AT&T study of telephone customer satisfaction. Researchers found that customer dissatisfaction increased significantly after waits longer than 20 seconds, and that answering 80% of contacts within this threshold produced acceptable satisfaction levels across a broad range of customer types.
The benchmark was adopted widely in the 1980s and 1990s as the telephony industry standardised around AT&T's research. It has no specific regulatory basis in most sectors: it is industry convention.
Important caveat: The 80/20 benchmark was developed for voice calls in an era before digital channels. Customers today have different expectations depending on channel. 80/20 may be appropriate for voice but is clearly wrong for email (should be hours, not seconds) and may be too slow or too fast for chat depending on the customer expectation set. Use 80/20 as a starting point for voice, not as a cross-channel rule.
Industry service level benchmarks
| Sector / channel | Common SL target | Rationale |
|---|---|---|
| Emergency services (999/112) | 90% in 5 seconds | Life-safety service; regulatory standard. No SL miss is acceptable. |
| NHS 111 (non-emergency health) | 90% in 60 seconds | Regulated by NHSE; clinical safety obligation. Long calls create natural queue. |
| Financial services (FCA-regulated) | 80–90% in 20s (voice); 100% in 5 days (email) | FCA Consumer Duty; high customer harm from delayed resolution. |
| Insurance claims (first notice of loss) | 90% in 20s | Customer distress context; claim delay costs money. |
| Retail banking (standard service) | 80% in 20s | Industry convention; competitive expectation. |
| Utility companies | 80% in 30s | Ofgem performance reporting; slightly relaxed vs. banking. |
| E-commerce / retail | 80% in 20s (peak); 70% in 30s (standard) | SL target typically relaxed outside peak season. |
| Telecommunications / ISP | 80% in 20s | Ofcom guidance (not mandatory); industry norm. |
| Live chat | 90% in 30–60s first response | Customer expectation of near-instant response in chat. But queue management differs from voice. |
| 80–95% within 24h or 5 business days | Async channel; SL measured in hours/days not seconds. Highly sector-dependent. | |
| Social media (DM) | 80% within 1–2 hours | Reputational urgency drives faster response than email but slower than chat. |
The Erlang C headcount cost of SL targets
Worked example: 200 calls/hr, AHT 7 min, target 80% in 20s
| SL target | Agents needed | vs. 70% SL | Occupancy at this headcount |
|---|---|---|---|
| 70% in 20s | 26 | — | ~85% |
| 80% in 20s | 28 | +2 (8% more) | ~79% |
| 85% in 20s | 29 | +3 (12% more) | ~76% |
| 90% in 20s | 31 | +5 (19% more) | ~71% |
| 95% in 20s | 35 | +9 (35% more) | ~63% |
| 80% in 10s | 32 | +6 (23% more vs. 70/20) | ~69% |
Erlang C estimates. Actual values depend on specific interval volume patterns and AHT distribution. Use the Erlang C calculator for your specific inputs.
The last 10pp of SL are the most expensive
Moving from 70% to 80% SL requires 2 additional agents in this example, an 8% headcount increase. Moving from 85% to 95% SL requires an additional 6 agents, a 21% headcount increase for the same 10pp SL gain. The Erlang C non-linearity means each percentage point of SL improvement costs progressively more as you approach 100%.
Tighter SL targets reduce occupancy — which is good for agents
A 95% SL target requires more agents, which means each agent handles fewer contacts per hour and therefore lower occupancy. This improves agent wellbeing, reduces burnout, and typically improves adherence. The headcount cost of a higher SL target is partially offset by lower attrition and higher productivity per agent.
SL and ASA give different pictures of the same queue
A contact centre hitting 80% in 20s might still have an ASA of 8 seconds, because 80% of contacts are answered immediately and the 20% waiting drag the average up. Reporting ASA makes the queue look better than SL reporting. If your stakeholders are asking for SL improvement, measure SL, not ASA.
Service level vs. average speed of answer
Service level (SL) — recommended
Percentage of contacts answered within a threshold (e.g., 80% in 20s). SL is sensitive to queue spikes: a period where 20% of contacts wait 5+ minutes is visible in the SL metric.
Use for: operational management, performance reporting, customer commitment
Average speed of answer (ASA) — misleading
Mean waiting time across all contacts. Long waits are diluted by the large number of fast answers. Can look good even when 15% of customers are waiting 5+ minutes.
Avoid for: primary performance metric. Use only for trend analysis alongside SL.
Service level target questions
What does 80/20 mean in a contact centre?
80% of contacts answered within 20 seconds. Industry convention originating from a 1970s AT&T study, not a universal optimum. Sectors with higher urgency or regulatory obligation use stricter targets (90% in 5s for emergency services; 90% in 20s for FCA-regulated financial services). Use 80/20 as a voice SLA starting point, but choose your target based on your sector, customer expectations, and headcount budget.
How does service level target affect contact centre headcount?
Erlang C produces a non-linear relationship. In a worked example (200 calls/hr, 7 min AHT): 70% SL needs 26 agents; 80% SL needs 28 (+8%); 90% SL needs 31 (+19%); 95% SL needs 35 (+35%). The last 10pp of SL are the most expensive. Moving from 70% to 95% SL doubles the headcount premium compared to moving from 70% to 80%.
What is the difference between service level and average speed of answer?
SL: % of contacts answered within a threshold; sensitive to queue spikes; best for operational management. ASA: mean waiting time across all contacts; long waits diluted by fast answers; can look good even when 15% of customers wait 5+ minutes. Use SL for performance reporting; avoid ASA as the primary metric.