Service Levels and KPIs

How we agree service levels

We do not commit to service levels before seeing your contact data. That is not caution, it is arithmetic: a speed-of-answer target is a function of arrival pattern, handling time and headcount, and committing to one without those numbers means either guessing high and failing, or padding the team and overcharging you.

The sequence is: you share arrival patterns and current performance, we model what is achievable at what team size, and we agree targets in writing with the assumptions attached. If the assumptions change materially - volume doubles, a new product launches, handling time rises - the targets are revisited rather than quietly missed.

The measures we report

Voice

• Average speed of answer - how long a caller waits before a person answers.
• Service level - the proportion of calls answered inside a target time. More useful than the average, because an average hides a long tail.
• Abandonment rate - callers who hang up while waiting. The clearest signal that coverage does not match arrival.
• Average handling time - reported for capacity planning, never used as a target on its own. Pushed as a goal it produces rushed calls and repeat contact.

Chat

• First response time - how long before a person replies to the opening message.
• Average response time within a conversation - the number that decides whether a chat feels attentive or abandoned mid-thread.
• Concurrency achieved - how many conversations agents actually held, against what was planned.

Email and tickets

• First response time - before a human replies, not before an automated acknowledgement.
• Resolution time - before the problem is fixed. Reported separately and deliberately.
• Backlog age profile - how old the oldest untouched tickets are, which an average response time conceals entirely.
• Reopen rate - tickets we believed resolved and had not. The most honest measure in written support.

Across every channel

• First contact resolution - solved in one interaction, without the customer coming back.
• Repeat contact rate - the inverse view, and harder to flatter.
• Customer satisfaction - the customer's assessment, with free-text themes read rather than only counted.
• Quality score - our assessment against the agreed scorecard.
• Contact rate per resolved issue - for omnichannel accounts, how many contacts one problem consumed.

Why we report response and resolution separately

Because a single blended number can hide a bad operation. A supplier showing excellent first response times and reluctance to discuss resolution times is sending fast acknowledgements and not solving much. Both numbers together are honest; either alone is not.

The same applies to handling time and resolution. Falling handling time alongside rising repeat contact is not efficiency, it is failure moved downstream where it costs more.

What happens when a target is missed

Missed targets get an explanation with a cause and an action, in the reporting cycle they occurred in - not summarised away at quarter end. Where the cause sits with us, we say so and state what changes. Where it sits with volume outside the agreed assumptions, an upstream process change, or a system problem on your side, we say that too, with the evidence.

Where a contract includes service credits, they are defined in the services agreement with the measurement method and exclusions written down. A service credit clause without an agreed measurement method is decoration.

Reporting cadence

Daily operational figures for the team running the queue. A weekly pack with trends and exceptions. A monthly review covering performance, quality findings, root causes and actions with owners. Quarterly, a wider look at whether the team shape and coverage still match how your customers behave.

You get the same numbers we manage the operation with. Where our reporting and your own perception of service diverge, that gap is the most useful item on the agenda, and we would rather surface it than manage it.

A note on the figures published on this site

Performance figures shown elsewhere on this site are historical or representative. They reflect particular engagements over stated periods and depend heavily on channel mix, volume, process complexity and campaign. They are not commitments about what your operation would achieve, and no supplier can honestly offer their past numbers as your future targets before seeing your queue.

Next: how quality is scored and calibrated, how a team is brought live, or share your current numbers and we will tell you what is realistic.