KPIs and KRIs: what an indicator is for, and where it stops working
A KPI tracks progress, a KRI warns of risk; both fail without a purpose and a limit.

An indicator is a proxy that somebody chose for a purpose. NIST SP 800-55 Vol. 1 supplies the vocabulary: a measure is a quantity taken from data, a metric sets that quantity against a target, and the two indicator types answer different questions — a key performance indicator is “A measure of progress toward intended results” (line 1246), a key risk indicator is “A metric used to measure risk” (line 1248). Both are metrics, though not every metric is one: “Key risk indicators (KRIs) and key performance indicators (KPIs) are examples of metrics, though not all metrics fall into these categories” (lines 694–695).
Their purpose is set by risk appetite and tolerance, not by the measurement function; their limits follow from the same sentence. An indicator that does not state what it is for, where its data comes from and which boundary it guards is a number, not a warning.
The vocabulary: measures, metrics, and the two Ks
SP 800-55 Vol. 1 defines the words so they stop being interchangeable. “Measurement” is the process of obtaining values; “Measures” are “Quantifiable and objective values that result from measurement”; “Metrics” are “Measures and assessment results designed to track progress, facilitate decision-making, and improve performance with respect to a set target” (lines 287, 289, 291–292). A metric has a target built into it; a measure is what the data source delivers.
measure a quantity taken from data number of failed backup jobs this week
metric that quantity against a set target failed jobs per week against a target of zero
KPI progress toward an intended result restores that succeeded — the result
KRI a signal of exposure to risk failed backups trending up — the risk
leading before the incident predictive; not yet proven
lagging after the event or trend an outcome; already happened
No one of these stands alone: “A single measure alone may not provide sufficient data to make risk-based decisions” (lines 387–388). Measurement earns its keep when the set is chosen well, because such measurements “can provide information on the implementation, effectiveness, efficiency, and business impacts of controls” (lines 621–622) — four different questions no single indicator answers at once.
Purpose: warning before a stated boundary
The indicator’s job is defined by whoever owns the risk appetite. NISTIR 8286r1 says senior leaders “should clearly express guidance regarding risk appetite and risk tolerance, and those tolerance statements should have clear and measurable boundaries where possible to define. Key performance indicators and key risk indicators should be created to warn that these tolerance boundaries are being approached and reported accordingly” (lines 1030–1033). The word “warn” is the design requirement: an indicator exists to give time to act before the boundary is crossed, which is why a KRI with no tolerance attached cannot do its job.
This is a governance layer, not a dashboard. NISTIR 8286Cr1 describes the loop the indicators feed: “By monitoring the results of CSRM activities at each hierarchical level, senior leaders can adjust various governance components (e.g., policy, procedures, workforce skills) to achieve risk objectives” (lines 336–338). CSF 2.0 states it as an outcome — “Organizational cybersecurity risk management performance is evaluated and reviewed for adjustments needed” (GV.OV-03, lines 822–823) — and SP 800-137 describes the top tier of metrics as those “designed to deliver information necessary to make risk management decisions in support of governance” (lines 640–641). Indicators here are chosen for a decision-maker, not for the team that produces them.
The communication problem: the consumer sets the requirement
The hardest step is not collection. SP 800-55 Vol. 2 names it plainly: “A major challenge concerning information security measurement is translating technically measurable performance data into a form that is contextually relevant to the consumer of the report” (lines 614–616). The same volume hands the requirement to the reader of the report: “Senior leadership will consider their organization’s reporting requirements to identify the most useful reporting style” (lines 620–621).
Vol. 1 adds two consequences: “The purpose and output of measurements must be unambiguous and easily understood” (lines 669–670), and “If a measure is reported to stakeholders without proper context, there can be unintended consequences or indecision” (lines 830–831). The same figure reads as healthy or alarming depending on its target, its tolerance and the decision it supports.
The limits: a proxy, and what it cannot promise
An indicator is a proxy, which fixes what it can and cannot do. Because it is chosen, it must state its purpose, its source data and its tolerance; without the first two a reader cannot tell what was measured, and without the third no value can be read as near or far from a boundary. Direction matters too. A leading indicator is “A predictive metric that tracks events or behaviors that precede incidents” — a forecast, not a proof — while a lagging indicator is “A metric that tracks the outcome of events or trends” (lines 1250–1251, 1253), and so always arrives after the event it describes.
The most common failure is measuring activity instead of the intended result. A count of tickets closed, patches applied or scans run records that work happened; it does not show that risk moved. SP 800-55 Vol. 1 treats implementation measures as a floor — useful as “a record of what exists and what needs improvement”, after which “the emphasis and resources of the measurement program broaden to include effectiveness, efficiency, and impact measures” (lines 989–991). A programme that never broadens keeps a KPI that rewards the wrong behaviour, and “Poorly selected quantitative metrics can undermine the overall quality of reporting and erode confidence in the work product” (lines 680–682). The neighbouring sheets in this batch take coverage and timeliness, residual-risk indicators and composite scores; this sheet owns only what a KPI and a KRI are for.
Level and prerequisites
L4 — framework and assurance: the vocabulary, governance purpose and limits of the measurement layer, above the records indicators are computed from. Prerequisites: the L1 vocabulary of risk and control; the L2 risk register and control objectives; the L3 sheets on risk-based prioritisation and the risk–control–owner–evidence chain.
Where to go next
- Knowledge — the Knowledge register this sheet belongs to.
- Cybersecurity Governance — the macro-area.
- Risk management — the node this sheet sits in.
References
- NIST — SP 800-55 Vol. 1, Measurement Guide for Information Security: Identifying and Selecting Measures — the measurement vocabulary, the KPI/KRI/leading/lagging definitions and the single-measure limit.
- NIST — SP 800-55 Vol. 2, Measurement Guide for Information Security: Developing a Measurement Program — the translation challenge and the consumer’s role.
- NIST — IR 8286r1, Integrating Cybersecurity and Enterprise Risk Management (ERM) — tolerance boundaries and the warning purpose of KPIs and KRIs.
- NIST — IR 8286Cr1, Staging Cybersecurity Risks for ERM and Governance Oversight — monitoring results and adjusting governance components.
- NIST — SP 800-137, Information Security Continuous Monitoring (ISCM) — Tier 1 metrics for governance decisions.
- NIST — CSF 2.0 (NIST CSWP 29) — GV.OV-03.