What Is Grade of Service (GoS) in Telecom?
Grade of service is the set of traffic-engineering measures, such as probability of loss and dial tone delay, used to judge if a group of lines is adequate.
Réponse rapide
Grade of service (GoS) is, in ITU-T E.600's words, a number of traffic engineering variables used to provide a measure of adequacy of a group of resources under specified conditions. Examples it gives are probability of loss and dial tone delay. In practice a GoS target says how often a call may be blocked or delayed before the line group is considered too small.
Grade of service is the telephone engineer's answer to "how many lines is enough?" It does not ask for zero blocked calls, which would need an unaffordable number of lines. It asks for a stated, measurable level of blocking or delay and then sizes the group of lines to meet it.
The Definition
ITU-T E.600 defines grade of service as "a number of traffic engineering variables used to provide a measure of adequacy of a group of resources under specified conditions", and says these variables may be probability of loss, dial tone delay and similar measures. Two notes in the recommendation matter in practice. The values set as objectives are called grade of service standards, and the values actually achieved under real conditions are called grade of service results. A target and a measurement are therefore two different things, and a report should show both.
E.600 separates this from a wider idea, the quality of service variable, which it defines as any performance variable, such as congestion or delay, that a user can perceive. Grade of service is the narrower, engineering-side measure of whether the resources are adequate.
Comment ça marche
The method has three steps. First, estimate the traffic the group must carry, in erlangs. Second, choose the GoS target, for example the maximum acceptable share of calls that find every line busy. Third, use a traffic table or formula to find the smallest number of lines that meets the target for that traffic. ITU-T E.520 has this problem as its subject: its title describes the number of circuits to be provided in automatic and/or semiautomatic operation, without overflow facilities. It recommends reading the number of circuits from tables or curves based on the classical Erlang B formula, and for automatic operation it bases the loss probability on 1 percent during the mean busy hour. That 1 percent is the parameter E.520 sets for its own circuit groups, not a target every business must adopt.
An Example
This is an illustration only. Suppose a team is told that its busy-hour traffic is a certain number of erlangs and that its GoS target is "no more than a stated small share of call attempts blocked". The engineer does not hire anyone. She looks up the line count that keeps blocking under that share for that traffic. If next quarter's traffic grows, the same target needs more lines. If the target is tightened, it also needs more lines. The target and the traffic together set the number.
Why It Matters for Outbound and Inbound Teams
When you bring your own carrier, the number of simultaneous channels on a SIP trunk is a purchase decision, and grade of service is the language for making it sensibly. Too few channels and attempts are blocked at peak, which an outbound campaign sees as dials that never go out and an inbound line sees as callers who get a busy signal. Too many and you pay for capacity that sits idle. BYOC lets you connect your own SIP trunks and shape routing by destination, cost or quality, so the sizing conversation is with your own carrier, using your own figures.
In a contact center the same thinking appears in a different form. A service level target, such as a share of calls answered within a set number of seconds, is the agent-side counterpart of a line-side GoS target: a stated promise about waiting that the staffing is sized to meet. The agent-side calculation is covered in the Erlang C calculator. Treat the comparison as an analogy for planning, not as a claim that the two are the same standard.
What to Put in a GoS Target
- The variable. Name what is measured, such as the probability of loss or a delay.
- The level. State a number, not "low".
- The conditions. E.600 says the measure applies under specified conditions, so specify them, for example the busiest hour of the week.
- The measurement. Decide in advance how the achieved result will be calculated, so the standard and the result can be compared.
Foire aux questions
Is grade of service the same as quality of service?
No. In ITU-T E.600, a quality of service variable is any performance variable a user can perceive, such as congestion or delay. Grade of service is a set of traffic engineering variables measuring whether a group of resources is adequate. Grade of service feeds quality of service; it is not the whole of it.
What does a grade of service number look like?
It is a stated limit on a variable such as the probability of loss, meaning the share of attempts that find all resources busy. The limit is chosen by the operator or buyer, and E.600 calls such objectives grade of service standards.
What is the difference between a GoS standard and a GoS result?
The standard is the objective you set. The result is what you actually achieved under real conditions. E.600 gives both names so that targets and measurements are not confused.
How is GoS different from service level?
GoS is usually applied to lines and switching resources and concerns blocking or delay on those resources. Service level is applied to contact center queues and is the share of contacts answered within a target time. Both are targets that capacity is sized to meet.
Does a lower blocking target cost more?
It needs more lines for the same traffic, so it generally costs more. That is the reason to set the target deliberately and to state it as a number, rather than asking for no blocked calls at all.
Related terms and guides: SIP trunk, service level, Erlang C calculator, BYOC, erlang.
To see what a service target means in agents for each interval of a day, try the call center staffing planner.
Sources
- ITU-T E.600, Terms and definitions of traffic engineering (03/93) (source, checked November 7, 2026)
- ITU-T E.520, Number of circuits to be provided in automatic and/or semiautomatic operation, without overflow facilities (11/88) (Erlang B basis and loss probability, section 1.1 of the recommendation text) (source, checked November 7, 2026)
À lire aussi : How many SIP channels do I need? Erlang B sizing
Termes associés
Qu'est-ce que l'IVR (Réponse Vocale Interactive) ?
L'IVR (serveur vocal interactif) est un système automatisé qui accueille les appelants avec des invites vocales et les route par saisie ou voix.
What Is Jitter in VoIP?
Jitter is the variation in the arrival timing of voice packets. Learn how RTP measures it, what the formula means and how it affects call quality.
What Is Latency in VoIP and Voice AI?
Latency is the delay between speaking and being heard. See what ITU-T G.114 says on one-way delay, how it differs from jitter, and why voice AI needs it low.
Qu'est-ce que le MOS (Mean Opinion Score) ?
Le MOS note la qualité audio des appels de 1 à 5. Comment il est estimé en trafic réel, ce qui fait baisser le score, pourquoi préférer le score par appel.
Qu'est-ce qu'une architecture multi-locataire ?
L'architecture multi-tenant sert plusieurs clients sur une infrastructure partagée avec isolation stricte. Fonctionnement et ce qui varie.
What Is NAT Traversal? STUN, TURN and ICE
NAT traversal lets voice and WebRTC media cross routers that hide private addresses. Learn how STUN, TURN and ICE fix one-way audio for browser-based agents.
Prêt à voir DialerBee en action ?
Demandez une démo en direct de 15 minutes, ou démarrez un essai gratuit de 14 jours — sans diaporama, sans engagement. Les appels démarrent dès que votre propre trunk SIP ou PBX est connecté et approuvé.
Essai gratuit de 14 jours · crédit IA gratuit · sans carte · 11 langues · BYOC · contrôles qui contribuent à la conformité