Terms and acronyms used in this book, in alphabetical order. Each entry gives a short plain definition and points to the home topic, where the term is explained and used. Where this glossary and a home topic differ, the home topic is the authority.

A

  • Alert quality: how often alerts are actionable, timely, and worth waking someone for, as opposed to noisy or duplicated (see topic 5.5, Monitoring and alert quality metrics).
  • Automation bias: the tendency to accept machine output without enough scrutiny (see topic 10.1, AI in service management measurement).
  • Availability: the share of time, or of requests, in which a service does what its users need (see topic 5.3, Availability metrics).

B

  • Backlog: the ordered list of work waiting to be done, whose age and health can be measured (see topic 2.6, Backlog and refinement metrics).
  • Baseline: a recorded view of how things stand before a change, used for later comparison (see topic 10.1, AI in service management measurement).
  • Blameless review: a review of an incident that looks at systems and decisions without assigning fault to individuals (see topic 4.2, Postmortem and learning metrics).

C

  • Change enablement: the ITIL practice of making changes safely and quickly, with approval proportional to risk (see topic 4.3, Change enablement metrics).
  • Change failure rate: the share of changes that cause degraded service or need remediation (see topic 4.4, Change failure and rework).
  • Charter (metrics): a short written statement of what measures are used for and never used for, signed by leaders (see topic 11.3, Rolling out metrics without breeding fear).
  • CMDB: configuration management database, the store of configuration items and their relationships (see topic 7.4, Configuration and CMDB quality metrics).
  • Continual improvement: the ongoing, small, evidence-based improvement of services and ways of working (see topic 9.5, Continual improvement metrics).
  • Continuous delivery: the practice of keeping software and services always in a releasable state and delivering value in small, frequent, low-risk steps (see topic 2.8, DORA metrics).
  • Cycle time: the elapsed time from when work starts to when it is finished (see topic 2.2, Lead time, cycle time, and flow efficiency).

D

  • Dashboard: a display of a small set of measures for a named audience, with trend and context (see topic 11.1, Designing a service metrics dashboard).
  • Deflection: a contact avoided because the person found an answer or completed a task by self-service, to be read with verified resolution and return contacts (see topic 3.7, Self-service and automation metrics).
  • Deployment frequency: how often a team puts changes into production (see topic 2.8, DORA metrics).
  • DORA metrics: the four delivery performance measures popularized by the DevOps Research and Assessment program: deployment frequency, lead time for changes, change failure rate, and time to restore service (see topic 2.8, DORA metrics).
  • Drift: gradual change in the data or behavior of a model or system away from what was tested (see topic 10.5, AI governance and risk metrics).

E

  • Error budget: the amount of unreliability a service may spend within its objective over a period (see topic 5.2, SLIs, SLOs, and error budgets).
  • Experience level agreement (XLA): an agreement framed around the outcomes and experience users need, as a complement to SLAs (see topic 6.2, Experience level agreements).

F

  • Flow efficiency: the share of elapsed time in which work is actively being worked on, as opposed to waiting (see topic 2.2, Lead time, cycle time, and flow efficiency).
  • FinOps: the practice of managing cloud and technology spend through shared accountability between engineering, finance, and the business (see topic 7.6, FinOps and cloud cost metrics).
  • First contact resolution: the share of contacts resolved at the first contact, read with return contacts (see topic 3.1, Service desk metrics).

G

  • Gaming vector: a way in which a metric can be made to look better without the underlying outcome improving (see topic 1.2, Goodhart's law and the psychology of metrics).
  • Goodhart's law: when a measure becomes a target, it ceases to be a good measure (see topic 1.2, Goodhart's law and the psychology of metrics).
  • Guardrail metric: a measure paired with another to detect distortion in the first, or a practice with the same purpose (see topic 1.3, Outcomes over output: choosing what to measure).

I

  • Incident: an unplanned interruption to a service or reduction in its quality (see topic 3.2, Incident volume and priority).
  • ITIL: a body of guidance for IT service management, whose fourth edition frames work as a service value system (see topic 1.1, Why measure agile ITIL).

K

  • Known error: a problem with a documented root cause or workaround, tracked until fixed (see topic 4.6, Technical debt and known errors).
  • Knowledge management: the practice of capturing, sharing, and improving what people know, measured by use and usefulness (see topic 3.6, Knowledge management metrics).

L

  • Lead time: the elapsed time from a request or commitment to its delivery (see topic 2.2, Lead time, cycle time, and flow efficiency).
  • Little's law: the relationship that average time in a stable system equals average work in process divided by average throughput (see topic 2.3, Work in process and queueing).

M

  • Major incident: an incident with significant impact that needs a coordinated response (see topic 3.4, Major incident metrics).
  • Maturity model: a description of stages through which a practice develops, used as a conversation tool and never as a rating (see topic 11.4, Maturity model for metrics programs).
  • Mean time to restore: the average time to restore service after an incident, usually read with percentiles because averages hide long cases (see topic 3.3, Incident response and resolution time).
  • Metric card: a short record of a measure's definition, formula, source, owner, limits, gaming vector, and guardrail (see topic 12.4, Templates).

N

  • Net Promoter Score: a survey-based measure of the willingness to recommend, to be read with other experience measures (see topic 6.1, Customer satisfaction metrics).

O

  • OKR: objectives and key results, a framework for linking goals to measurable outcomes (see topic 1.7, Measurement frameworks).
  • OLA: operational level agreement, an internal agreement between teams that supports a service level (see topic 5.1, SLA, OLA, and XLA metrics).
  • Outcome: a change in the situation of people or the organization that matters, as opposed to the output produced (see topic 1.3, Outcomes over output: choosing what to measure).

P

  • Percentile: the value below which a given share of observations fall, such as the 85th percentile of a time measure (see topic 1.6, Statistical literacy for service metrics).
  • Postmortem: a written review of an incident for learning (see topic 4.2, Postmortem and learning metrics).
  • Problem: the underlying cause of one or more incidents (see topic 4.1, Problem management metrics).
  • Prompt injection: an attack in which instructions hidden in input manipulate a language model (see topic 10.5, AI governance and risk metrics).
  • Psychological safety: a shared belief that a team is safe for speaking up about mistakes and problems (see topic 11.3, Rolling out metrics without breeding fear).

R

  • Release and deployment: the practices of making new or changed services available and moving changes into production (see topic 4.5, Release and deployment metrics).
  • Return on investment: the net benefit of an investment relative to its cost, stated with its assumptions (see topic 7.7, Return on investment and business case).
  • Risk register: a record of identified risks with owners, ratings, and actions (see topic 8.2, Risk and control metrics).
  • Run chart: a chart of a measure over time with a median line (see topic 1.6, Statistical literacy for service metrics).

S

  • Self-service: the ability for users to find answers or complete requests without contacting a person (see topic 3.7, Self-service and automation metrics).
  • Service level agreement (SLA): an agreement between a provider and a customer on the level of service (see topic 5.1, SLA, OLA, and XLA metrics).
  • Service level indicator (SLI): a measured aspect of service quality, such as the share of successful requests (see topic 5.2, SLIs, SLOs, and error budgets).
  • Service level objective (SLO): a target for an SLI over a period (see topic 5.2, SLIs, SLOs, and error budgets).
  • Service request: a formal request from a user for something to be provided, such as access or information (see topic 3.5, Service request fulfillment metrics).
  • Shadow IT: technology used without the knowledge or approval of the responsible function, including AI tools adopted by staff (see topic 10.5, AI governance and risk metrics).
  • Sustainable pace: a rate of work that people can maintain indefinitely without harm to their health or the quality of work (see topic 9.2, Satisfaction and well-being metrics).

T

  • Technical debt: the accumulated cost of shortcuts and deferred maintenance that slows future work (see topic 4.6, Technical debt and known errors).
  • Throughput: the number of items of work finished in a period (see topic 2.1, Flow measures).
  • Toil: repetitive, manual, automatable operational work that grows with the service and has no lasting value (see topic 5.7, Toil and on-call metrics).
  • Total cost of ownership: the full cost of acquiring, running, and leaving something over its life (see topic 11.2, Tooling landscape).

U

  • Unit economics: the cost and value of a service per unit of use, such as per active user or per transaction (see topic 7.5, Service cost and unit economics).
  • Unplanned work: work that arrives without being planned, such as fixing incidents, which competes with planned work (see topic 2.9, Operational work and unplanned work).

V

  • Value stream: the sequence of steps from a request to the value delivered, including waits (see topic 2.4, Value stream metrics).
  • Verified resolution: a resolution confirmed by the user or by evidence that the problem did not return (see topic 10.3, AI agent and chatbot metrics).

W

  • Work in process (WIP): the amount of work started and not yet finished (see topic 2.3, Work in process and queueing).