The Metric Has a Mandate: Auditing the Behaviour a Service KPI Commissions

Service KPIs are usually presented as instruments of oversight. In practice, they are instruments of instruction. A scorecard tells managers what to notice, teams what to protect and customers what the organisation is prepared to trade away. When one measure dominates — average handling time in a contact centre, utilisation in a shared service, SLA compliance in an outsourced arrangement — it stops describing the operating model and starts becoming it. The behavioural audit set out here is a way to test that influence before the scorecard hardens into habit.

Begin with behaviour, not the dashboard

A conventional metric review asks whether performance is improving. A behavioural audit asks a different question: what does this measure make people do? That shift matters because the operating model is not only the process map, the governance forum or the service catalogue. It is also the set of daily choices that the scorecard makes rational. If a team learns that closing a contact quickly is rewarded more reliably than resolving the underlying issue, the model has changed — even if no one has redrawn the diagram.

The audit should therefore start with observed behaviour. Review decision logs, sample contacts, escalation routes, handover notes and the language used in team meetings. Ask frontline staff what they would do differently if a given metric disappeared. Ask team leaders which conversations they avoid because the number is under pressure. The answers usually reveal more about the operating model than the formal design documentation.

The four common distortions

Most service scorecards contain a small number of dominant measures. Each has a legitimate purpose, and each can distort when treated as a standalone target.

• Average handling time (AHT) rewards ending contact quickly. It can displace thorough diagnosis, accurate noting and ownership of complex cases. It also hides an assumption that contacts are broadly interchangeable and that repeat demand is outside the service's control.
• First-contact resolution (FCR) rewards closure at the first interaction. It can displace careful triage when closure is recorded prematurely, and it hides an assumption that first contact is always possible or desirable. Some issues are better resolved through a planned callback or a specialist handover.
• Utilisation rewards visible busyness. It can displace the slack needed for surge absorption, coaching, improvement work and informal problem-solving. It hides an assumption that demand is predictable and that all useful work is captured in the system.
• SLA compliance rewards clock management. It can displace fair prioritisation and end-to-end customer time, particularly when the SLA starts at the wrong point or measures only one segment of the journey. It hides an assumption that queue time is the customer's main cost.

The audit test is simple: for each metric, record what work it rewards, what work it displaces and which operating-model assumption it leaves unexamined. Then look for evidence of the displaced work in repeat contacts, reopens, escalations, backlog ageing, customer effort and colleague workload.

Case study: the number that became the model

A UK shared-service centre handling billing and collections had made AHT the primary measure reviewed each week at director level. The official operating model described specialist triage, end-to-end ownership and a single queue for complex accounts. In practice, the scorecard commissioned something else. Agents learned to avoid complex accounts, transfer them to the back office and keep notes brief. Informal queue splits appeared. Team leaders coached to the handle-time target because that was the number that shaped their own reviews.

Over 18 months, AHT fell by 9 per cent. The headline looked strong. But repeat contacts within seven days rose by 14 per cent, reopened cases increased and the back office absorbed work that the front line no longer wanted to own. The operating model had not been redesigned on paper. It had been redesigned in practice: triage became informal, escalation routes multiplied and ownership thinned at the point of contact.

The lesson is not that AHT is a bad metric. It is that a dominant metric acquires decision rights. It determines what gets investigated, what gets celebrated and what gets tolerated. The behavioural audit makes that acquisition visible.

Pair distorted measures with counter-metrics

Counter-metrics are not a licence to add targets until the scorecard becomes unreadable. They are a check on displacement. A practical pairing might link AHT with a seven-day repeat-contact rate; FCR with quality audit outcomes and reopen rates; utilisation with service-level performance during surge periods; and SLA compliance with aged work and customer-effort indicators.

The design principle is to measure the work the primary metric might push elsewhere. If AHT improves while repeat contacts rise, the service may be buying speed with rework. If FCR improves while reopens increase, closure may be outrunning resolution. If utilisation improves while backlog ageing worsens, the service may be optimising visible activity rather than flow.

Keep the counter-metric set small. Two or three paired measures are usually enough to expose the most important distortions. More than that and the scorecard starts to lose the very clarity that made it useful in the first place.

Run a short shadow-scorecard trial

A shadow scorecard is a low-risk way to test whether the current measures are commissioning the behaviour the operating model needs. Run it for six to eight weeks alongside the official scorecard, without changing formal targets or publishing league tables. Select one or two teams, define the behaviour you expect to see, and collect evidence through contact sampling, decision logs, short frontline interviews and weekly review notes.

The trial should ask three questions. First, do the paired measures reveal displacement that the primary metric conceals? Second, do team leaders and frontline staff change their decisions when the paired measures are discussed? Third, is the evidence strong enough to retain, amend, pair or retire the original measure?

A shadow trial is not a performance-management exercise. It is an operating-model diagnostic. Its output should be a short decision paper for the service governance forum: what the metric rewards, what it displaces, which assumptions it hides and what the organisation intends to do about it.

Audit questions for the next scorecard review

Before the next quarterly review, put six questions to each dominant KPI. What does this measure make easy? What does it make expensive? Which customer or colleague group carries the displaced work? What assumption about demand, work or handover does it leave unexamined? Which counter-metric would reveal displacement within one reporting cycle? And who has the authority to change the measure when the evidence warrants it?

The scoreboard is never a neutral window on the operating model. It is part of the model. Auditing the behaviour it commissions is therefore not an exercise in metric diplomacy. It is a practical test of whether the service is being managed towards the outcomes it claims to value.