Service Operating Model All articles
Governance and Risk

The Last Dropdown: How Closure Codes Decide What Your Service Model Learns

Service Operating Model
The Last Dropdown: How Closure Codes Decide What Your Service Model Learns

Photo: Heba Essa, CC BY-SA 4.0, via Wikimedia Commons

Service operating models are usually articulate about intake. Channels, eligibility rules, triage matrices, priority logic and response clocks are documented, argued over and tested against real demand. The moment a case leaves the system receives a fraction of that attention — typically a dropdown near the bottom of a form, selected by an agent under time pressure, often after the substantive work is already done.

That asymmetry deserves scrutiny. Closure is the point at which the operating model converts activity into record. It is where the organisation decides what happened, what caused it and whether anything should change. A closure code is not a label applied to finished work; it is an instruction about what the business will believe six months later, when the people who did the work have moved on.

Why the close is the least governed moment

Three structural reasons explain the neglect.

First, closure is largely invisible to the customer. Whether a case was closed as resolved, as no further action or as customer did not respond rarely changes what the customer experiences next. Because nobody complains about the code, nobody reviews it.

Second, closure tends to lack a natural owner. Case management teams own throughput, quality teams own sampling, finance owns cost per contact. The code set itself — its definitions, its boundaries, its maintenance — frequently belongs to no one in particular, inherited from a system implementation that finished years ago.

Third, codes are cheap to add and expensive to retire. A reporting requirement arrives, a new code is created, and the old one remains because a historic dashboard still references it. The set grows; the definitions blur.

What a closure code actually commissions

Closure codes are not passive. They feed, and therefore shape, a set of measures that many service leaders treat as objective: first contact resolution, often derived from a closure code rather than from observed customer behaviour; repeat contact rates, which depend on whether the next contact is attributed to the previous case; defect and fault logs, which capture only what someone chose to record at closure; and complaint root-cause reporting, which usually rests on the agent's selected cause rather than on investigated fact.

They also commission behaviour. If a code removes a case from a denominator — advice given closing a query without a follow-up task, or no fault found removing an engineering visit from a defect count — agents and team leaders will learn that lesson quickly and apply it consistently. That is not cynicism; it is a rational response to the incentives the model has built.

A useful analytical split is to sort the code set into three families and check that each is doing its own job:

Family Question it answers Typical failure
Outcome What state is the customer in now? Overlaps with effort codes, so agents pick the easier one
Cause What produced this contact? Selected by guesswork, then reported as fact
Effort What did we do about it? Used as a substitute for outcome

Where these three are collapsed into a single list, reporting degrades. Cause analysis becomes contaminated by effort, and outcome measures drift towards what the team did rather than what changed for the customer.

The discretion problem

Most closure code sets contain at least two codes an agent could reasonably choose for the same case. Some ambiguity is unavoidable, but the scale of the overlap should be known and actively managed. Two diagnostics are worth running.

The first is concentration. If the two most-used codes account for a very large share of all closures, the set is functioning less as a taxonomy and more as a binary choice between fine and not fine. Its learning value is close to zero.

The second is drift. Compare the distribution of codes for a comparable case type across teams, sites or periods. Wide variation usually indicates differing local definitions rather than differing reality, which means a national report built on those codes is comparing vocabularies rather than outcomes.

A practical closure audit

The following sequence can be run in a fortnight by a small team with access to case records and a sample of agents.

  1. Recover the definitions. Collect the code set, its documented definitions and dates of introduction. Where no written definition exists, ask three team leaders to define the same code independently and compare their answers.
  2. Sample twenty closures per major code. Read the case notes, not just the fields. Ask of each: does the code accurately describe what happened, and does the note contain the evidence for it?
  3. Test reversibility. Establish whether a closed case can be reopened, by whom, and whether the record shows both the original closure and the reopening. A model that cannot represent reopening will systematically under-report failure demand.
  4. Trace the downstream use. Identify every report, dashboard and target that consumes the code. For each, ask what decision it informs and what would change if the code were wrong.
  5. Run the concentration and drift diagnostics set out above, and agree in advance the threshold at which the code set is revised rather than reinterpreted.

What good looks like

A defensible closure regime has a small number of mutually exclusive outcome codes, a separate and clearly evidenced cause set, and an explicit route for reopening. It has a named owner. It distinguishes between what the customer reported, what the agent observed and what was independently verified — three different levels of evidence that should never be averaged together.

Above all, it treats resolved as a claim requiring support. If the operating model cannot say, from its own records, why a case was closed and who decided that, the closure code is not a control. It is a habit with a field name.

All Articles

Related Articles

Commissioned by Configuration: Five CRM Settings That Quietly Write Your Service Model

Commissioned by Configuration: Five CRM Settings That Quietly Write Your Service Model

When a Gesture Becomes Policy: Governing Discretionary Remedies in Service Operations

When a Gesture Becomes Policy: Governing Discretionary Remedies in Service Operations

Ranked by Rumour: Auditing the Priorities That Really Govern Your Service Queue

Ranked by Rumour: Auditing the Priorities That Really Govern Your Service Queue