The Glossary

The vocabulary of decision intelligence.

Every term, defined properly: what it means, how it misleads, and what decision it should trigger. Each entry links to its canonical page in the Knowledge Centre or a chapter of The Decision Layer.

A

Absorption Debt: The cost a firm carries so that its decisions never have to be reopened, compounding invisibly until the margin report finds it.
AI Audit: An auditor does not ask whether your model is clever. They ask what it did, on what evidence, who decided, and what happened.
AI Governance: An organisation with an AI policy but no inventory, owners or review cadence is not governing AI. It is describing it.
AI Literacy: Not a course in how models work. The ability to know what an output rests on, and when to say no to it.
AI Transparency: If nobody made the decisions a transparency notice describes, the notice is not disclosure. It is fiction with a legal review.
Amnesiac Enterprise: A firm whose memory lives in people rather than systems, and leaves in one pair of shoes when they do.
Annex III: The Act’s most important page is a list. The question it puts to you: is what we do on it, and who decided?
Average Handle Time (AHT): Talk plus hold plus wrap. Why AHT is the right way to price the work and the wrong way to judge it, and what happens when it becomes a target.

B

Backlog: The most reassuring number in a services business, and the least examined. Size says what was sold; quality decides what it becomes.
Base Rate Neglect: Why every project believes it is the exception, and why only firms that kept their outcomes have an outside view to consult.
Bench Cost: The bench is either an investment with an owner and a review date, or a staffing decision nobody revisited. Same cost, different object.
Binding Constraint: Averages hide the one thing that will actually stop you. Convergence surfaces it before you commit.
Binding Date: The date a single constraint stops being the reason you cannot act, and why the latest one sets the plan.
Blended Rate: One rate, one invoice line, one moving mixture underneath. How blended pricing simplifies buying and quietly blurs the economics.
Blind Cockpit: Instruments for everything that can be counted and no reading on the only question that matters: are we going to be alright?
Book-to-Bill Ratio: One number for whether you are selling faster than you deliver. Honest over years, treacherous over any single quarter.
Bounded Rationality: Nobody optimises. Everyone satisfices. Simon showed the fix is better decision structure, not smarter people.
Business Confidence: Nobody types it in. It is computed from evidence, weighted by quality, and it names the dimension that binds.
Business Continuity Plan (BCP): In BPO, resilience is sold, not just kept. A real BCP covers sites, people, technology and data, and it has been tested, not just written.
Business Transformation Outsourcing: BPO executes the work you designed. BTO helps redesign how the work happens. A different philosophy, not a new acronym.
BYO AI: Your provider, your contract, your data terms, the platform's reasoning, unchanged. What BYO AI does and doesn't change.

C

Candidate Option: The option as a first-class, frozen object, so the road not taken stays priced exactly as it was.
Challengeable AI: Any conclusion that cannot answer “why do you believe that?” is an assertion, not intelligence.
Change Control: Every engagement drifts from its contract. Change control decides whether the drift is recorded and priced, or absorbed and discovered at closeout.
Change Order: The formal moment a scope change gets priced. Avoided out of relationship fear, and a leading indicator of margin health when it is not.
Commercial Confidence: A deal can be won, delivered, and still lose money. Commercial Confidence scores the margin that survives true cost to serve, not the revenue booked.
Commitment Intelligence: The discipline of running a commitment through the business as an executable object, not filing it as a record.
Compound Patterns: A contract commitment, a hiring gap and a technology date can each pass review and still, together, describe a promise the firm cannot keep.
Confidence Composition: Composed versus asserted confidence, and why a typed-in number is a feeling wearing a percentage.
Cost-to-Serve: The average cost per client is a fiction. Cost-to-serve is what one client actually costs, and what the renewal decision needs.
Cross-module Intelligence: The expensive problems live between departments. Detection that spans functions finds them before production does.
Customer Effort Score (CES): CSAT scores the mood, NPS scores the relationship, CES scores the machine. Why effort is the most operational of the three, and where it really hides.
Cynefin Framework: Dave Snowden’s sense-making framework, and why treating complex client situations as merely complicated is the costliest habit in services.

D

Daily Decision Briefing: What changed, what binds, what needs deciding today: the morning surface that replaces archaeology.
Data Minimisation: Collect what the purpose needs and nothing more. Every field you never collect is a risk you never carry.
Decision Audit Trail: The reconstructable record of how a decision was actually made, and why it stops being honest the moment it is used for blame.
Decision Debt: Technical debt got a name and a budget. The decisions a firm defers, never writes down and never reviews are the larger debt, still unnamed.
Decision Hygiene: Independent judgements first, structure before debate, intuition last. The unglamorous discipline that outperforms brilliance in repeated decisions.
Decision Intelligence: Not analytics, and not a copilot. The discipline of making decisions you can trace, challenge and learn from.
Decision Intelligence vs AI: Decision Intelligence is not a rebrand of AI. It is the discipline that makes AI safe to decide with: evidence attached, options constrained, outcomes scored.
Decision Intelligence vs Business Intelligence: BI describes the past in dashboards. Decision Intelligence decides the future and remembers why. One reports; the other is accountable.
Decision Latency: The time from signal to decision, and the silent tax it levies on every option with an expiry date.
Decision Layer: The missing layer between systems and strategy where judgement happens: what the book defines and teaches a firm to build.
Decision Log: Five sentences per decision: choice, owner, evidence, alternatives, expected outcome. Written for reuse, not for audits.
Decision Object: When decisions become objects, they gain properties: they can be traced, challenged, audited and scored.
Decision Provenance: Confidence you can audit, not assert: the unbroken chain from facts to recommendation to choice to outcome.
Decision Quality vs Outcome Quality: A sound decision can lose and a reckless one can win. Review the reasoning against what was knowable, not the scoreboard.
Decision Review: The retrospective that examines the reasoning, not the person, against what was knowable at the time.
Decision Rights: Who holds the pen, who is consulted, and who must live with the result. Clarity beats consensus.
Decision Room: Not a meeting, not a dashboard, a governed space where one decision gets made, with the options already priced.
Decision Routing: Findings delivered to the place where a choice gets made and recorded, not to a document that informs.
Decision Window: Every decision has a window in which it is cheap and a point after which it is only expensive. Firms miss the window because nothing announces it.
Decisionless Stack: An enterprise software estate with a system of record for everything except the decisions that run the business.
Delivery Confidence: A green status goes red the week before it fails. Delivery Confidence predicts, composed from workforce, capacity, dependencies and their dates.
Delivery Margin: Engagements are sold at one margin and delivered at another. The gap is not bad luck; it is a ledger of unrecorded decisions.
Demand Signal: External evidence of likely need, treated as decision input rather than an excuse to send an email.
DPIA: Assume the new system harms someone. Work out how, before it exists. That is a DPIA doing its job.

E

Effective Bill Rate: Divide what the work earned by every hour it took. The number that emerges is the firm’s real price, and it rarely matches the card.
Enterprise Intelligence: Not analytics, not a copilot, a reasoning layer that connects every function's evidence into every decision.
Erlang C: The century-old queueing formula behind every staffing plan: what it computes, why thin staffing fails so suddenly, and who owns the assumptions.
Escalation of Commitment: Good money follows bad when reputation, not arithmetic, is doing the deciding.
Event Chain: Attribution as a chain of evidence from touch to revenue, with the broken links admitted instead of papered over.
Evidence Hierarchy: Measured beats modelled beats inferred beats stated. And unmeasured is labelled, loudly, because AI without evidence is just faster guessing.
Evidence State: The label that stops an assumption impersonating a measurement.
Evidence-Based Messaging: The discipline of letting the record write the copy: claims checked against outcomes, not against ambition.
Expected Value Thinking: Weighing outcomes by likelihood rather than vividness, judging the portfolio rather than the single bet, and knowing when the maths stops working.
Explainability: A system that cannot answer “why do you believe that?” with evidence is not explaining. It is narrating.

F

First Contact Resolution (FCR): The one number that predicts satisfaction, cost and repeat volume at once. Who decides what “resolved” means, and why FCR fights with handle time.
Forecast Accuracy: A daily total can be exactly right while every interval was wrong. Why forecast error compounds, and what honest scoring looks like.

G

Gross Margin per Engagement: The portfolio average is a comfortable fiction. Profit and loss happen one engagement at a time.

H

High-Risk AI System: High risk is not a ban. It is the law’s way of saying this decision matters enough to be governed.
Human Oversight: The law does not ask for a human near the output. It asks for a human able to change it.
Human-in-the-Loop: A human clicking approve on things they cannot inspect is not a loop. It is theatre with an audit trail.

I

Institutional Memory: Memory that lives in people is not institutional memory at all. It is an exposure the organisation has not priced.
Integration Illusion: The belief that connecting systems produces decisions. Integration makes the nouns consistent; it does nothing for the verbs.
Intraday Management: Reforecasts, skill moves, break shuffles and voluntary time off: the fastest decision loop in the business, and what it inherits from planning.

L

Land and Expand: The small first engagement is a trust decision. Expansion is what happens when delivery keeps the promise the landing made.
Lawful Basis: Six legal reasons to process personal data. Choosing one is a decision, and consent is rarely the right one at work.
Legitimate Interests Assessment: Purpose, necessity, balance. Three questions that turn “legitimate interests” from a hunch into a defensible decision.
Living ICP: The ideal customer profile as a maintained belief, updated by scored outcomes rather than laminated after a workshop.

M

Margin Leak: The quiet, repeated escape of margin through unowned decisions at the seams between functions.
Marketing Readiness: The question that belongs before the budget: should this demand exist at all, and can we prove we should be the ones to create it?
Master Services Agreement (MSA): The MSA holds the relationship’s terms; the SOW holds the work. Trouble starts when SOWs quietly begin rewriting the container.
Model Drift: The model you validated is a photograph. The model you are running is whatever the world has made of it since.

N

Nearshore, Offshore, Onshore: The labels sound like geography. The decision they name is about total cost, control and time, and it is never made only once.
Net Revenue Retention: The growth you already own. NRR nets churn, shrink and expansion inside the existing client base into the rate at which a people business compounds.

O

OODA Loop: Speed matters, but orientation decides. Boyd’s loop is won in the sense-making, not the acting.
Optionality: Options are bought, held at a cost, and quietly expiring. The discipline is knowing which ones you hold, what they cost, and when they die.
Organisational Judgement: Individual judgement is fast, personal and fragile. Organisational judgement survives the departure of everyone who built it.
Outcome-Based BPO: Paying for results instead of effort is the most honest commercial model on paper, and the easiest one to fake in practice.
Outcome-Based Pricing: Paying for results sounds like perfect alignment. Without baselines and attribution rules it is a dispute with a delayed start date.
Outcomes Ledger: The record of decided versus happened, scored. Without it, every recommendation starts from zero.

P

Pattern Library: What an organisation is allowed to believe about its own history: patterns mined from scored outcomes, never from the loudest story.
People Business: Some companies sell software. A large, under-served category sells people doing skilled work. Their margin is not manufactured, it is decided.
People-Business Loop: The single loop a people business runs on: work is won, delivered, understood, and fed back into the next sale.
Per-Seat vs Per-Transaction Pricing: Paying for capacity and paying for volume are different bets. Each model shapes behaviour, and each hides a different truth about cost.
Pipeline Coverage: Three times coverage feels safe. But the multiple measures hope, not evidence, and 3x of bad pipeline hides the gap until it is too late to close.
Pre-mortem: Assume the commitment failed. Write the story of why. Then decide whether to make it.
Project Accounting: The ledger knows the firm made money. It cannot say which engagement made it, and which quietly gave it back.
The Pyramid Model: Rates get the credit, but the shape earns the margin. The pyramid is the economics and the career model of a services firm in one structure.

Q

Quality Assurance Score: Sampled, scored and calibrated, yet still not the customer’s verdict. What a QA score can tell you, and where it quietly stops telling the truth.

R

Rate Card: Every firm has a rate card. The question is whether the card governs the pricing or merely decorates it.
Re-derivation Tax: The question your firm answered a year ago is being answered again this week, from scratch, by people who never knew it was asked.
Readiness Audit: One honest afternoon that shows where a firm loses its decisions: five areas, one worksheet, no six-week assessment.
Realisation Rate: The rate card states what an hour is worth. Realisation records what the firm, under pressure, actually dared to charge for it.
Revenue Leakage: The money a firm earned but never invoiced. Invisible in aggregate reporting, and every leak hides a moment where someone chose, or failed to.
Revenue Recognition: Revenue is earned as the work is delivered, not when the invoice goes out. Blur that line and delivery problems come back as restatements.
Reversible vs Irreversible Decisions: Two-way doors deserve speed; one-way doors deserve weight. Most firms get both backwards.

S

Sales Velocity: Four levers, one number. Clarifying as a frame, deceptive as an average, corrosive as a target.
Scenario Run: You can rerun a decision with today’s facts. You can never rewrite yesterday’s run.
Schedule Adherence: The most personal metric in the contact centre. What adherence measures, how it differs from conformance, and what over-tightening it costs in people.
Scope Creep: Not a delivery failure but hundreds of pricing decisions made by default. The honest yes: price it, trade it, or gift it knowingly.
Seat Utilisation: A full building is not a productive one. Why the facilities metric gets mistaken for output, and how per-seat economics warp delivery decisions.
Service Level: X% answered in Y seconds. The tail of callers the headline never mentions, and why the operational metric is not the contract that borrows its name.
Service Level Agreement (SLA): The contractual floor of a service, not the truth of it. Green SLAs and a red relationship can coexist: the watermelon effect.
Span of Control: The ratio nobody decides: what a leader’s week actually buys, and what widening the span quietly spends on quality and attrition.
Statement of Work (SOW): The written record of the promises the delivery team inherits: scope, assumptions, acceptance, rates and change control.
Sunk Cost Fallacy: What was spent is a fact about the past. What to do next is a claim about the future.
System of Record vs System of Memory: Enterprise software remembers what happened with total fidelity, and why it happened not at all. That is the missing system.

T

T&M vs Fixed Price: Two pricing structures, one estimate. The difference is who pays when the estimate turns out to be wrong.
Theory of Constraints: Every system has one constraint that sets its pace. Find it, exploit it, and stop polishing everything else.
Trust Economy: The market condition in which trust, earned by a demonstrable run of good decisions, is the scarcest asset a firm owns.

U

Utilisation Rate: Billable over what, exactly? Why the denominator matters, and why a very high utilisation number is a warning dressed as a win.

V

Voice of the Customer (VoC): Everyone collects feedback. A real VoC programme converts anecdotes into evidence, reads trajectories instead of averages, and changes decisions.

W

Win Confidence: Not a rep’s gut feel at quarter-end. A transparent score of whether a deal will actually close, composed from the evidence the deal has produced.
Work in Progress (WIP): Between the hour worked and the invoice raised sits an asset that quietly decays. WIP ages faster than anyone bills it.
WYSIATI: The deck is not the world. Kahneman named the mind’s habit of judging only from what is in view.

Z

Zero Data Retention: The model provider forgets your prompts. Your application does not. Know exactly where the promise ends.

The glossary grows deliberately: a term earns its entry when it can be defined honestly, taught properly, and traced to how the platform or the industry actually behaves.

Explore the Knowledge Centre