Quality of experience vs quality of service: what each measures, and why the gap matters

Quality of service (QoS) measures the network itself, such as availability, latency and packet loss. Quality of experience (QoE) measures what the user actually receives from an application. A network can meet its QoS targets while QoE is poor, which is why measuring only QoS leaves the real question, whether the work got done, unanswered.

The two definitions

Quality of service describes the behaviour of the network. It is the domain of availability, latency, jitter and packet loss, and the TM Forum defines it as the collective effect of service performance that determines the degree of user satisfaction. Quality of experience describes the other end of the same story: what the person or application using the network actually gets. The ITU defines quality of experience as the degree of delight or annoyance of the user of an application or service. QoS is measured on the wire. QoE is measured where the work happens.

Why the gap matters, and why it widens across an estate

The two are related but not the same, and the space between them is where most assurance problems live. A circuit can be available and lightly loaded, every QoS metric green, while a video call stutters or a point of sale hangs, because QoE depends on the whole path and the application, not just the link. This is the watermelon SLA: green on the outside, red where the work is done.

Across a distributed estate the gap widens, for a simple reason. When performance is rolled up across many sites and many carriers into a single monthly figure, the average hides the worst site and the worst hour. A network that is 99.9 per cent available on paper can still contain a branch that is unusable every afternoon, and a QoS dashboard reported as an estate wide average will never show it. The more sites there are, the more the average conceals.

Moving from QoS to QoE

The industry has been shifting toward experience for several years. Gartner defines digital experience monitoring as the measurement of the availability, performance and quality of the user experience of applications, and a growing number of organisations now write experience level agreements, commitments framed around what the user receives rather than what the interface reports. The common thread is that the network being up is treated as necessary but not sufficient. The test is whether the applications that matter actually performed.

How to measure QoE across a distributed estate

Three properties make QoE measurable rather than anecdotal:

• Per site, not an estate average, so the bad branch is visible instead of averaged away.
• At the level of the applications the site depends on, not a generic list, so it reflects the real work.
• Continuous, so the bad afternoon is on the record.

Measured that way, quality of experience stops being a complaint and becomes a number you can act on, and reconcile against the provider’s QoS report.

How XVOLV does it

XVOLV measures experience on both sides of the network. Beacon measures the real experience of the applications you nominate across cellular and wireless sites, and forecasts where performance is heading before a user reports it. Link does the same for wireline circuits. Both test the applications you choose, per site and continuously, so the estate wide average never hides the branch that is failing. The argument is set out in the white paper Up Is Not Working.

Frequently asked questions

What is the difference between quality of service and quality of experience?

Quality of service (QoS) measures the network itself, such as availability, latency and packet loss. Quality of experience (QoE) measures what the user actually receives from an application. QoS is measured on the network; QoE is measured where the work happens.

What is quality of experience?

Quality of experience is the degree of delight or annoyance of the user of an application or service, as the ITU defines it. It reflects the whole path and the application, not just whether the network link is up.

Why does QoS look fine when users are unhappy?

Because QoS measures the link, not the application experience. A circuit can be available and lightly loaded while an application still performs poorly. Reported as a monthly or estate wide average, QoS can read green while specific sites or hours are red. This is the watermelon SLA.

How do you measure QoE across many sites?

Measure per site rather than as an estate average, at the level of the applications each site actually uses, and continuously rather than as a snapshot. That exposes the worst site and the worst hour instead of averaging them away.

What is an experience level agreement (XLA)?

An experience level agreement is a commitment framed around what the user actually experiences, rather than only the network metrics an SLA traditionally covers. It uses experience measurement to hold a service to the outcome, not just the connection.

Sources

• ITU — definition of quality of experience (Recommendation P.10/G.100): the degree of delight or annoyance of the user of an application or service. itu.int
• Gartner — definition of digital experience monitoring (DEM). gartner.com
• TM Forum, SLA Management Handbook (GB917) — definition of quality of service. tmforum.org