Reference · Glossary

The language we use with clients.

Short, opinionated definitions of the process, operating-model and AI-first terms that come up in every engagement. Built to be linked, scanned and disagreed with.

59 terms

A

Agentic AI: AI systems that plan and carry out multi-step tasks with tools and data, instead of only answering a single prompt. In a process context an agent is a new kind of actor that needs a defined scope, decision rights and an audit trail. Without those boundaries, an agent is an unmanaged employee.
AI-native process: A process designed from the start for humans, automation and AI agents to operate as one system not a legacy process with an LLM bolted on. Decisions, data and handoffs are explicitly assigned to the actor best placed to make them.

B

Bottleneck: The step that limits the throughput of the whole process, no matter how fast the other steps run. Improvements anywhere else only build queues in front of it. Finding and relieving the bottleneck is the highest-return move in almost every operational redesign.
BPMN: Business Process Model and Notation is the standard visual language for describing process flows, events, gateways and roles. Its value is a shared notation that business and IT read the same way. Its risk is precision theatre: a perfect diagram of a process nobody follows.
Business process: A repeating set of activities that turns an input into a customer or business outcome. Defined by its trigger, its steps, the roles that own those steps, and the result it is accountable for not by the system that happens to run it.

C

Category management: Organising spend into categories with an owner, a strategy and a plan for each, instead of reacting purchase by purchase. It makes demand visible, so the conversation moves from price negotiation to what the business actually needs. Categories without a named owner and a mandate stay slideware.
Change management: The deliberate work of moving an organisation from the current way of operating to a new one communication, training, role redefinition and reinforcement. Process redesign without change management ends up on paper, not in behaviour.
Contract management: Keeping contracts findable, current and actually enforced across their full life cycle, from drafting to renewal or exit. Savings agreed in negotiation only materialise when prices, terms and volumes are applied in the buying systems. Unmanaged contracts are the most common source of value that was promised and never landed.
Cycle time: The time it takes to complete one pass of a process step or of the whole process, measured from start to finish. It is the most honest operational metric because customers experience it directly. Averages hide the problem, so look at the distribution and the slowest tenth of cases.

D

Data governance: The rules, roles and routines that keep data definitions, ownership and quality under control. Effective governance names an owner for each critical data object and measures quality where the data is created. Without it, analytics and AI amplify the noise already in the process.
Digital twin: A live, data-fed model of a physical operation (a factory, a warehouse, a supply network) used to simulate decisions before they are committed in reality. In process work, the twin is the bridge from reactive planning to intelligent operations.
DMAIC: The improvement cycle of Define, Measure, Analyse, Improve and Control used in Six Sigma projects. Its discipline is that no change is made before the current performance is measured. The Control phase is the one most often skipped, which is why gains fade.

E

ERP (Enterprise Resource Planning): The system of record that ties finance, supply chain, manufacturing and HR data together. ERP rollouts succeed or fail on process work that happens before the system is configured not on the system itself.

F

First-pass yield: The share of cases that pass through a process correctly the first time, without rework, correction or escalation. It is the clearest measure of built-in quality in administrative and production processes alike. A low first-pass yield means capacity is being spent twice on the same work.

G

Governance model: The structure of forums, decision rights and escalation paths that keeps an operating model running. A workable model states who decides what, on what data and how often. Most governance failures are not missing meetings but missing decision rights.

H

Handoff: The point where work moves from one role, team or system to another. Handoffs are where information is lost, waiting time accumulates and accountability blurs. Counting the handoffs in a process is often the quickest estimate of how much improvement is available.
Hire-to-Retire (H2R): The employee value stream from recruitment and onboarding through development and role changes to exit. It spans HR, managers, IT and payroll, which is why access rights and equipment are the classic failure points. Good design here shows up as a new hire being productive in days rather than weeks.
Human in the loop: A design where a person reviews, approves or corrects an automated decision before it takes effect. It is the standard way to use automation and AI where errors are costly or regulated. The review point should be placed where judgement is genuinely needed, not on every step by default.
Hyperautomation: The combined use of several automation technologies, such as workflow engines, software robots, integrations and AI, on the same end-to-end process. The idea is to automate the flow rather than isolated tasks. It only pays off when the process has been simplified first.

I

Idea-to-Launch (I2L): The value stream from a product idea through concept, development and industrialisation to market launch. It is the hardest stream to standardise because the work is creative and the decisions are irreversible. The lever is usually clearer decision gates, not more project reporting.
Integration: The technical connection that lets two systems exchange data without a person copying it between them. Integrations remove manual steps but also freeze the process logic embedded in them. Design the process first, then decide what the integration should carry.

K

Kaizen: Continuous, incremental improvement driven by the people who actually do the work. Effective when the underlying process is sound; insufficient when the process itself needs redesign.

L

Large language model: A model trained on large volumes of text that generates and interprets language, and increasingly other formats as well. In process work it is useful for reading, drafting, classifying and summarising, and unreliable as a source of truth. Its output needs the same controls as any other step in the process.
Lead time: The total elapsed time a customer waits from request to delivery, including all queues and waiting. Lead time is almost always far longer than the sum of the actual work steps. Shortening it usually means removing waiting, not making people work faster.
Lean: A management system originating at Toyota that defines value from the customer's perspective and eliminates everything that does not produce it. In process work, the contribution is rigorous waste identification not the slogans.

M

Master data: The shared reference data that processes depend on: customers, suppliers, materials, prices and organisational structures. When master data is inconsistent, every downstream process inherits the error and covers it with manual work. Master data quality is a process question long before it is a system question.

O

Operating model: The combination of processes, roles, decision rights, performance metrics and governance that turns strategy into day-to-day execution. The translation layer between what the company says it does and what it actually does.
Order-to-Cash (O2C): The value stream from a customer placing an order through fulfilment, invoicing and cash collection. The single most common process where ERP and process design collide.
OTIF (On-Time-In-Full): The share of orders delivered to the customer on the promised date, complete and in the agreed quantity. A single number that exposes the health of the whole Order-to-Cash chain.

P

Plan-to-Produce (P2P): The value stream from demand plan through scheduling, materials and production to finished goods. It is where sales promises meet factory capacity, and where planning assumptions get tested weekly. Most schedule instability traces back to a planning process, not to the shop floor.
Process architecture: The hierarchical map of how an organisation's processes relate strategic-level value streams at the top, operational-level steps at the bottom. A good architecture lets anyone find where their work lives and how it connects to adjacent functions.
Process KPI: A small set of metrics that show whether a process is delivering its intended outcome typically cycle time, first-pass yield, on-time-in-full and cost-to-serve. Good KPIs are owned by the process owner, not by a reporting team.
Process map: A visual description of how work actually moves from trigger to outcome, including steps, decisions, systems and handoffs. A useful map records reality first and the target state second. Maps drawn only from workshops tend to describe the process people wish they had.
Process maturity model: A scale that describes how far a process has moved from ad hoc work to defined, measured and continuously improved operation. It is useful as a shared language for prioritising, and useless as a target in itself. Maturity levels matter only when tied to a business outcome such as cost, speed or quality.
Process mining: Reconstructing the actual flow of a process from event logs in the underlying systems, rather than from interviews and assumptions. Useful for diagnosing where reality diverges from the documented process.
Process owner: The named person accountable for a process end-to-end: its design, its performance metrics, its improvement backlog and the decisions that cross functional borders. Without one, every process is everyone's job and therefore no one's.
Process sprint: A 4–6 week fixed-scope engagement that diagnoses one process, quantifies the loss and ships a redesign the team can act on Monday. The opposite of a 90-slide assessment.
Procurement: The end-to-end function that buys the goods and services the business runs on, from demand intake to contract, order and payment. Good procurement is measured on total value and risk, not on unit price alone. When it is treated as a back office, spend leaks quietly through the operating units.
Purchase-to-Pay (P2P): The value stream from identifying a need through ordering, receiving, invoice matching and payment. It touches procurement, operations and finance, so its failures usually sit in the handoffs. Typical symptoms are maverick buying, unmatched invoices and late payments.

R

RACI: A matrix that assigns Responsible, Accountable, Consulted and Informed roles to each step of a process. Useful as a sanity check on decision rights; dangerous when treated as a substitute for actual process design.
Record-to-Report (R2R): The finance value stream from booking transactions through reconciliation and close to statutory and management reporting. Its performance shows up as closing days and the number of manual adjustments. Long closes are almost always an upstream data and process problem, not an accounting one.
Root cause analysis: A structured search for the underlying reason a problem occurs, rather than the symptom that was noticed first. Common tools are the five whys and the cause and effect diagram. The test of a real root cause is that removing it prevents the problem from returning.
RPA (Robotic Process Automation): Software bots that mimic clicks and keystrokes inside existing UIs to automate repetitive work. RPA delivers fast wins on stable processes; on broken ones it locks the dysfunction in place.

S

S&OE (Sales & Operations Execution): The weekly or daily loop that closes the gap between the agreed plan and what the factory, warehouse and customer service actually do right now. It handles the short-horizon exceptions that S&OP is too slow to catch. Without it, the monthly plan quietly dies in the first week.
S&OP (Sales & Operations Planning): A monthly leadership cycle that balances demand and supply across sales, operations and finance, and lands on one plan with one set of numbers. It works only when the cadence carries real decision rights, not just a review deck. Most S&OP failures are governance failures, not forecasting failures.
Shared services: An operating model where repeatable work from several units is consolidated into one service organisation with defined service levels. It creates scale and standardisation, and it moves the difficulty into the interface between the centre and the business. Service level agreements are only as good as the process behind them.
SIPOC: A one-page frame that lists Suppliers, Inputs, Process, Outputs and Customers for a single process. It is the fastest way to agree on scope before anyone starts drawing detailed flows. Most scope disputes in process work disappear once a SIPOC is on the wall.
Six Sigma: A quality method that reduces variation in a process using statistical analysis and structured problem solving. It is strongest where volumes are high and data is reliable. Applied to low-volume or poorly defined processes, the analysis costs more than the variation it removes.
Sourcing: The upstream work of defining the requirement, finding suppliers, running the competition and awarding the business. It sets the ceiling for everything that follows, because a poorly specified requirement cannot be recovered downstream. Strategic sourcing repeats this on a cycle rather than treating each buy as a one-off.
Standard operating procedure: A written description of the agreed way to perform a task, precise enough that two people produce the same result. Standards are the baseline improvement is measured against, not bureaucracy for its own sake. An unused standard is a sign that the work and the document have drifted apart.
Supplier collaboration: Working with key suppliers on shared forecasts, capacity, quality and development rather than exchanging orders and complaints. It requires shared data and a joint cadence, not goodwill statements. Done properly it shortens lead times and removes buffer stock on both sides.
Supplier Relationship Management (SRM): The governance of supplier performance, risk and development after the contract is signed, segmented by how critical each supplier is. It covers scorecards, review cadence and joint improvement targets. Most organisations invest heavily in selecting suppliers and almost nothing in managing them.
Swimlane diagram: A process diagram laid out so each role or function has its own horizontal lane and handoffs cross those lanes visibly. The fastest way to surface where work is dropped between teams.

T

Target operating model: A description of how the organisation intends to work once a transformation is complete: processes, roles, decision rights, data and systems. It is useful when it is specific enough to argue with and to plan against. A target model that fits on one slide is usually a wish list, not a model.

V

Value stream: The end-to-end sequence of activities across functions and systems that delivers a specific outcome to a customer. Order-to-Cash, Hire-to-Retire and Idea-to-Launch are classic examples.
Value stream mapping: A Lean technique that maps every step of a value stream together with its lead time, processing time and quality losses. The point is the ratio between the two times: how much of the elapsed time is actual work. In office processes that ratio is often under five percent.

W

Waste (muda): Any activity that consumes time, money or attention without adding value for the customer. Classic categories include waiting, rework, unnecessary movement, overprocessing and excess inventory. In knowledge work the biggest categories are waiting and rework, not effort.
Work in progress (WIP): The amount of work that has been started but not yet finished at any given moment. High work in progress lengthens lead time, hides defects and makes priorities impossible to see. Limiting it is often faster and cheaper than adding people.
Workflow orchestration: The layer that sequences tasks across people, systems and automations, and keeps track of state, exceptions and deadlines. Orchestration is what turns separate automations into one coherent process. Its main benefit is visibility: every case has a known status and owner.