Back
Go-live is an event · Performance is a condition

Technology can be live.The improvement can still be missing.

Performance begins when an intended business difference becomes observable. A key performance indicator—KPI—is not the difference itself. It is a decision instrument: a defined signal, read with context, limits and counter-signals.

Build the performance account
KPI instrumentDecision-ready
Outcome KPICustomer need resolved

Not “reply sent.” Not “user logged in.” The indicator must represent the difference the work is meant to make.

DefinitionResolution confirmed within the agreed service boundary
Decision ruleInvestigate when the outcome and operating signals separate
Counter-signalRepeat contact · review burden · unresolved exceptions
One KPI cannot carry the whole claim. The outcome signal needs operational context and a condition for action.
The performance proposition

A KPI begins with a consequence—not a dashboard.

Perspective defines the change worth pursuing. Performance turns that proposition into evidence: what outcome matters, how it will be recognised, what else must be watched, and what decision follows from the result.

Service guidance recommends working out performance measures while a service is being built rather than attaching them afterwards.1 Formal KPI standards in manufacturing likewise specify more than a label: formulas, elements, time behaviour, units and intended users are part of the definition.2 The application here is broader, but the discipline is useful.

Business outcomeWhat should become better?
KPI systemWhat evidence represents it—and what could mislead?
DecisionWhat will continue, change, narrow or stop?
Performance architecture

Build a performance account—not a victory metric.

A favourable number is easy to display. A useful performance account connects the intended outcome to traceable work, counter-signals and a decision. The five “i” concerns prevent the KPI from becoming blind to its own conditions.

01 · Work

Events occur

A need enters, people and systems act, exceptions appear, and an outcome follows—or fails to.

02 · Evidence

Indicators represent

Selected events become KPIs, supporting indicators and counter-signals with explicit definitions.

03 · Decision

Action responds

The organisation continues, investigates, revises, restricts or stops according to the evidence.

KPI specification“Needs resolved reliably”
MeasureConfirmed resolution
DenominatorEligible enquiries received
Time windowDefined reporting period
SegmentationNeed, channel and exception type
SourceTraceable service events
OwnerNamed decision responsibility
iStrategic intelligence

Selects the outcome worth pursuing and the consequence that warrants attention.

iBusiness intelligence

Defines, calculates, segments and compares the KPI and its counter-signals.

iInformation systems

Provides traceable events, definitions, lineage and dependable data movement.

iArtificial intelligence

Adds task-level evaluation, error distribution, deferral and production monitoring.

iUI / UX

Reveals completion, friction, correction, escalation and the lived cost of the workflow.

An editorial architecture for examining performance. It is not a universal measurement standard or a claim that five concerns exhaust the decision.

The KPI remains useful only while its definition, source and decision purpose still correspond to the work.

The five “i” contributions

Each “i” changes what performance means.

The concerns do not form a sequence. They interrogate the same claim from different positions and produce different requirements for measurement and action.

i

Strategic intelligence

Worth

Connect the KPI to a priority, trade-off and accountable decision.

Output · decision relevance
i

Business intelligence

Evidence

Define the measure, baseline, target, segments, trend and uncertainty.

Output · interpretable KPI
i

Information systems

Trace

Establish which records are authoritative and how events move across systems.

Output · data lineage
i

Artificial intelligence

Behaviour

Measure the bounded task in conditions similar to deployment and monitor it in production.3

Output · task assurance
i

UI / UX

Experience

Observe whether people can understand, complete, correct and recover within the journey.

Output · lived performance
Read the levels correctly

Availability, activity and outcome answer different questions.

All three can become indicators. Only the outcome establishes the intended business difference—and even that must be read with quality, effort and consequences. Information-systems research likewise treats success as multidimensional rather than reducible to use alone.4

Availability

Can it operate?

Uptime, access, integration and recoverability describe technical readiness.

Necessary condition—not proof of value.
Activity

Did people or systems use it?

Login, completion and feature events show contact with the capability.

Evidence of use—not necessarily useful change.
Outcome

Did worthwhile work improve?

Assess the result together with quality, effort, exposure and experience.

The claim that the KPI system must help examine.
Performance is a decision discipline

A KPI without a response rule is only observation.

Before launch, define who reads the signal, how often, against what baseline or target, and what happens when the outcome and counter-signals disagree. Targets are choices, not facts; thresholds need consequences and review.

Continue
The outcome holds and counter-signals remain acceptable.
Investigate
The headline improves while effort, exceptions or experience deteriorate.
Revise
The workflow, measure or technical response no longer fits the evidence.
Restrict or stop
Benefit no longer justifies cost, exposure or displaced work.

Preserving an implementation is not the same as preserving performance. The technology remains conditional on the value it was introduced to serve.

A performance account ready for Precision

Make the claim measurable. Keep the decision challengeable.

The performance account now names the intended outcome, the KPI definition, the counter-signals, the evidence source and the response rule. Precision can turn those elements into explicit requirements, limits and acceptance conditions.

The claim
What worthwhile improvement is being asserted?
The instrument
Which KPI and counter-signals can examine it?
The decision rule
What evidence changes the response?
Continue to PrecisionWhat must be specified before the decision can be trusted?

Sources and notes

The customer-resolution KPI is illustrative. The performance architecture and five “i” contributions are editorial syntheses; sources support specific practices, not the combined model.

  1. GOV.UK Service Manual — How to set performance metrics for your service. Supports defining service measures during discovery and development rather than after launch. Government service guidance, not a universal KPI standard.↩
  2. ISO 22400-2:2014 — KPIs for manufacturing operations management. Its abstract identifies formulas, elements, time behaviour, units and user groups as characteristics of specified KPIs. This page borrows the discipline of explicit definition; it does not claim the manufacturing standard governs general service KPIs.↩
  3. NIST AI RMF 1.0 — Measure 2.3 and 2.4. Supports measuring AI performance or assurance criteria in conditions similar to deployment and monitoring behaviour in production. Voluntary guidance, not certification.↩
  4. DeLone & McLean, 2003 — Information Systems Success. Supports examining several dimensions of information-systems success rather than treating use as sufficient. It does not validate this page’s KPI architecture.↩