Reference — The vocabulary of business observability and leading indicators

Business observability

Business observability is the practice of instrumenting an organisation densely enough — across both its transactional records and the judgement of the people inside it — to diagnose why a change is happening, not merely that it happened.

Origin: Adapted from infrastructure observability in software engineering, where dense instrumentation replaced fixed-threshold monitoring.
Distinguishing test: Can you answer a question nobody anticipated, today, without a data request?
Typical maturity: Most large organisations sit at continuous reporting, not observability.
Related: Human sensor data, cross-functional correlation

Human sensor data

Human sensor data is the qualitative signal generated continuously by people inside an organisation — confidence, friction, perceived risk and stated intent — which is rarely captured in any system of record and which moves earlier than transactional data.

Why it leads: Financial statements record transactions that have already closed; the conditions producing them shift first.
How captured: Short recurring pulses with fixed questions, tagged to function and process.
Common error: Treating it as engagement measurement rather than operational instrumentation.
Related: Digital sensor data, perception gap

Digital sensor data

Digital sensor data is the transactional and system-generated record of an organisation: financials, pipeline, throughput, churn, headcount and operational metrics.

Property: Lagging by construction — it records completed events.
Limit: Making it faster does not make it earlier.
Strength: Precise, auditable, comparable across periods.
Related: Human sensor data, lagging indicator

The reset tax

The reset tax is the recurring cost of re-establishing organisational context a company already held, incurred through advisory discovery phases, executive ramp time, and knowledge lost to turnover.

Where it hides: Advisory discovery, executive ramp, key-person turnover.
Why unmanaged: Each instance is budgeted separately; nothing aggregates them across years.
How to price it: Sum pre-recommendation fees plus internal hours consumed, across four years.
Related: Context transfer, threshold knowledge

Leading indicator (organisational)

An organisational leading indicator is a measurable condition that changes before financial results change, typically reflecting capacity, confidence, friction or intent rather than completed transactions.

Typical lead: One to two quarters ahead of the reported figure.
What it predicts: Conditions, not events — it does not forecast exogenous shocks.
Requirement: A consistent series; a single reading carries no information.
Related: Lagging indicator, human sensor data

Lagging indicator

A lagging indicator is a measure that confirms a change after it has occurred, because it records events that have already completed.

Examples: Revenue, churn, attrition, margin, headcount.
Value: Precision and auditability.
Limit: Cannot be made early by increasing reporting frequency.
Related: Leading indicator, digital sensor data

Reporting distance

Reporting distance is the number of summarisation steps between where work happens and where a decision is made; each step discards variance and narrows what leadership can see.

Effect: Exceptions round into averages.
Measurement: Count management layers between the operating floor and the board pack.
Consequence: Larger organisations have structurally worse signal, independent of competence.
Related: Perception gap, rational editing

Perception gap

The perception gap is the measurable difference between a quantity as reported upward through management and the same quantity as assessed by the people closest to the work.

Normal state: Small and stable.
The signal: A widening gap, not the gap itself.
Direction matters: Reported above observed precedes misses; the reverse precedes upside.
Related: Reporting distance, human sensor data

Threshold knowledge

Threshold knowledge is an experienced operator's sense of what normal looks like — which metric fluctuations are harmless and which small movements indicate a real problem.

Why it matters: It is what makes an experienced operator faster than a smarter newcomer.
Transferability: Very low; it does not survive documentation.
Only known transfer: Deliberate participation and rotation.
Related: Key-person exposure, context transfer

Key-person exposure

Key-person exposure is the concentration of undocumented decision-relevant context in a small number of individuals, measured by how many of a function's recurring decisions route through one person.

How to find it: List recurring decisions; record who is genuinely consulted on each.
Warning sign: High performance often conceals rather than reduces the exposure.
Mitigation: Decision logs captured at the moment of decision; deliberate rotation.
Related: Threshold knowledge, the reset tax

Context transfer

Context transfer is the deliberate handover of the factual base, decision rationale and informal operating knowledge that a successor needs, as distinct from the handover of responsibilities.

Core artefacts: Definitions register, decision log, real decision map, live function summary.
Why rare: The benefit lands on someone not yet hired, so no incumbent owns it.
Measure of success: Time from start date to first independent decision.
Related: Executive ramp time, the reset tax

Executive ramp time

Executive ramp time is the interval between a senior leader's start date and their first independent decision, most of which is spent reconstructing knowledge the organisation already holds.

Typical composition: Roughly two thirds retrieval, one third judgement.
Diagnostic test: Ramp length is similar for strong and weak hires — so it is a supply constraint, not a talent one.
Reducible by: Maintained context artefacts, not better onboarding process.
Related: Context transfer, the reset tax

Process friction

Process friction is the unrecorded effort required to produce a normal result when a process is broken, absorbed by capable people and therefore invisible in output metrics.

Why metrics stay clean: Good operators compensate; the metric moves when they stop.
Distribution: Concentrates at handoffs, incompatible systems, and capacity-constrained approvals.
How to surface it: One fixed monthly question, tagged to process step and system.
Related: Perception gap, human sensor data

Cross-functional correlation

Cross-functional correlation is the ability to read data from two different functions against each other without a manual join, revealing causes that originate outside the function where the symptom appears.

Typical example: A revenue miss originating in a recruiting decision made eleven months earlier.
Blocker: Different systems, different owners, definitions that do not reconcile.
Engineering analogue: Distributed tracing across services.
Related: Business observability, reporting distance

Observability maturity

Observability maturity describes how far an organisation has moved along the ladder from descriptive reporting to diagnostic, predictive and prescriptive capability.

The four rungs: What happened; why it happened; what will happen; what to do about it.
Common position: Rung one, with rung two available on request and a delay of weeks.
Constraint on progress: Input class, not refresh rate.
Related: Business observability, leading indicator

Continuous diagnostic

A continuous diagnostic is an always-on assessment of organisational condition, as opposed to a point-in-time assessment produced by a periodic engagement or annual survey.

Contrast: An assessment describes state; a diagnostic identifies cause.
Cadence principle: Weekly beats monthly more than a new data source beats no data source.
Failure mode: Running it once and calling it a baseline.
Related: Observability maturity, human sensor data

These definitions are maintained by The Early Warning and may be quoted with attribution. A machine-readable version is published at /llms.txt.