KPIs and metrics

Service Level Explained: The 80/20 Target and Its Limits

What the 80/20 service level means, how to calculate it, why it is a convention and not a law, where it misleads and how to pick a target that fits.

Call editorial team · Published · 5 min read

"We run an 80/20." Many contact centre managers say it as if it were a physical constant. It is not. It is a convention that took hold because it is easy to state and to measure: 80 of every 100 calls answered within 20 seconds. Understanding what it does and does not tell you is the difference between using it and being used by it.

A contact centre team at their desks with a large screen showing live queue figures
Illustrative image

This guide explains how service level is calculated, where 80/20 comes from, its blind spots, and how to choose and tune a target. It connects to Average Speed of Answer, Answer Rate and Abandon Rate and the broader designing service levels.

What service level means

Service level has two numbers: a share of calls and a time. The share is the proportion of calls answered within the time. So an 80/20 target reads "answer 80 of every 100 calls within 20 seconds". A 90/30 target would be "90 of every 100 within 30 seconds".

Calculating service level for an interval

  1. Count the calls offeredAll calls that arrived for the queue in the period.
  2. Count answered within the thresholdCalls picked up by an agent within the target time, such as 20 seconds.
  3. Decide how to treat abandoned callsSome teams include them as failures; others exclude very short abandons. Pick a rule and document it.
  4. DivideAnswered within threshold divided by calls offered (or by your chosen base).
  5. Compare with the targetLook at it by half hour, not just for the day.

Because definitions differ, always check how a report calculates service level before you compare it with a number from another team or a vendor.

Where 80/20 comes from

The 80/20 figure grew out of early call-centre practice and became a default in planning tools and contracts. It stuck because it sits at a sensible balance: high enough that most callers wait only briefly, low enough that you do not need a large idle team to achieve it. It is a rule of thumb, not a standard set by any regulator, and many organisations use something different.

Choosing a different target is not failing. A busy emergency line may aim for something much tighter. A back-office queue where callers expect a callback may aim looser. What matters is that the target reflects the reason people call and what a wait costs them.

The limits of service level

Service level is useful, but it tells only part of the story.

What service level can hide

  • The unlucky tailThe other calls that waited much longer are invisible. A queue can meet 80/20 while a few callers wait many minutes.
  • Abandoned callsIf abandons are excluded or counted loosely, a good-looking score can hide callers who gave up.
  • Averaging across the dayA quiet morning can offset a disastrous lunch hour in a daily figure.
  • Quality of the callA fast answer to a call that then goes badly still fails the caller.
  • Small numbersWith only a few calls in an interval, one slow answer can swing the figure wildly.

Choosing a target that fits

Start from callers, not from a number you saw in a book. Ask what a wait means for each type of call, what callers do when they wait (hold, abandon, call back) and what your team can realistically staff.

Different calls, different targets

Urgent or time-critical lineGeneral enquiry line
What a wait costsA caller may be in difficulty; every second mattersA caller can usually wait or try again
Typical choiceTighter answer time and generous staffingLooser answer time, with callback offered
Alternatives to waitingClear route to another urgent numberCallback or voicemail
Review rhythmCheck often, adjust fastReview monthly with volumes

Service level and staffing

Service level depends mostly on having enough agents available when calls arrive. The relationship is not linear: improving from a moderate level to a very high one often needs disproportionately more staff, because the extra agents spend more time idle. That is why chasing a very tight target can be expensive in effort even when money is not discussed. Forecasting methods such as Erlang C help estimate the staff needed; see Erlang C Explained and Forecasting Call Volume.

Real life is bumpier than models. Calls arrive in bursts, agents take breaks, handle times vary and some calls need a specialist. Use the model to get close, then use real-time monitoring to adjust.

Improving service level when it falls short

A practical order of attack

  1. Find the bad intervalsLook at the half hours where you miss, not just the daily total.
  2. Check the scheduleAre people on shift when calls arrive? Compare scheduled with required staffing.
  3. Reduce avoidable callsFix the causes of repeat calls and confusing menus.
  4. Shorten handling carefullyRemove wasted steps without rushing callers; see Reducing Average Handle Time.
  5. Offer alternativesCallbacks and a clear online route reduce peak pressure.

A worked example

Imagine a small support line that receives 200 calls in a morning. It answers 150 within 20 seconds, answers 30 more after longer waits and loses 20 callers who hang up. Depending on the rule, the service level is 150 out of 200 (counting abandons as misses) or 150 out of 180 (leaving them out). The first is lower and tells a more honest story about callers. Neither shows that three of the 30 late answers took several minutes.

The invented numbers are there to make the point: the same morning produces different scores under different rules, so the rule matters as much as the result.

Reporting it honestly

Present service level with its definition, the interval it covers and the volume behind it. If you exclude short abandons, say so. If you changed the target, show the old and the new. Honest reporting builds trust with your team and with whoever reads the numbers. For the wider set of measures, see Call Center KPIs.

Service level review

  • Write down the definition your reports use
  • Agree separate targets for each queue or call type
  • Check service level by half hour, not just by day
  • Look at longest wait and abandon rate alongside it
  • Compare scheduled staffing with forecast need
  • Offer callbacks for long-wait peaks
  • Revisit the target at least twice a year

Doing it in Call

In Call, you set a queue's answer target on its Thresholds tab under Queues and SLA: you enter the share of calls to answer and the number of seconds to answer them within. You can also set alert limits for the longest wait, the number of waiting calls and the abandon rate. These limits change what is raised as an alert, not how calls are routed. The Intervals tab in Reports shows, half hour by half hour, calls offered, answered and abandoned, with average speed of answer and handle time, and a service level projection for each queue with its target. The wallboard shows service level, calls waiting and longest wait as live tiles. Check the screens for exactly how each figure is defined before comparing with other systems.

Frequently asked questions

What does 80/20 mean?

It means 80 of every 100 calls are answered within 20 seconds. It is a common service level target, though not a rule, and many teams choose different numbers.

Is 80/20 a legal or regulatory requirement?

No. It is an industry convention. Some contracts or sector rules set their own answer-time requirements, so check the ones that apply to you.

Should abandoned calls count against service level?

Practice varies. Some teams count them as not answered in time, others exclude very short abandons. Choose a rule, apply it consistently and state it when reporting.

Why can service level look fine while callers complain?

It measures the share answered in time, not how long the others waited, how many hung up or how the call went. Look at longest wait, abandon rate and satisfaction alongside it.

How do I choose a target?

Start from why people call and what waiting costs them, then check what you can staff. Set separate targets for different queues and revisit them after changes in volume or opening hours.

Is a higher target always better?

No. Very tight targets need disproportionately more staff and can leave agents idle. Pick the level that serves your callers at a workload your team can sustain.

Set an answer-time target for a queue in Call

In the demo console you can try incoming calls in a simulation.

Related guides