Flow Metrics Glossary: 24 Terms, Defined Plainly
Every flow and delivery term you will meet, defined in one paragraph each without the consultancy vocabulary, with links to the full explanation where one exists.
Founder of FlowBoard · Full-stack developer & serial entrepreneur
Introduction
Flow terminology is unusually bad at explaining itself. Half the terms are borrowed from manufacturing, several mean different things depending on who is using them, and a few are used interchangeably by people who would insist they are distinct.
Each entry below is one paragraph, written to be understood without the surrounding framework. Where there is a full article, it is linked. The whole set fits together in the flow metrics guide.
A to F
Aging work in progress
How long each unfinished item has already been open, measured right now. The only flow metric that describes work you can still influence, which makes it the most actionable of the set. See aging work in progress.
Batch size
How much work is grouped together before being released. Large batches inflate measured cycle time, delay feedback, and concentrate risk, because a failure now involves everything in the batch at once.
Blocked time
Time an item spends unable to progress because of something outside the team: an unanswered question, an unavailable environment, a third party. Counting it honestly is what makes flow efficiency meaningful.
Classes of service
Categories of work handled under different policies, such as expedite, standard, and fixed date. Useful at scale, and usually over-engineering for a team under fifteen, where a single priority order does the same job with less to maintain.
Control chart
A chart plotting a metric over time against an expected range, used to distinguish normal variation from a genuine change. In flow contexts largely superseded by the scatterplot, which conveys the same information more directly.
Cumulative flow diagram
A stacked area chart of item counts per stage over time. Band thickness shows work in progress, the horizontal gap approximates cycle time, and the slope of the done band is throughput. See how to read a cumulative flow diagram.
Cycle time
Elapsed time from when work starts on an item to when it is finished. The team-facing measure of speed, reported as a distribution rather than an average. See cycle time vs lead time.
Definition of done
The agreed condition under which an item counts as finished. Matters more than it sounds: an inconsistent definition makes cycle time and throughput incomparable across items and across time. See why the done list matters.
Delivery rate
A synonym for throughput. Sometimes preferred in client-facing contexts because it needs no explanation.
Flow debt
Improvement in one item's cycle time achieved by delaying others, most often by expediting. It looks like progress in the aggregate and is borrowed rather than earned, showing up later as a fat tail in the distribution.
Flow efficiency
Active work time divided by total elapsed time, as a percentage. Most software teams land between five and twenty percent, which reframes where improvement effort should go. See flow efficiency.
Flow load
Another name for work in progress, used when distinguishing items in the system from items being actively worked on.
L to R
Lead time
Elapsed time from when a request arrives to when it is delivered. The customer-facing measure, always longer than cycle time, with the difference being queue time before work began. See cycle time vs lead time.
Little's Law
Average cycle time equals average work in progress divided by average throughput. A mathematical result rather than a technique, and the reason WIP limits reduce delivery time without anyone working differently. See Little's Law for software teams.
Monte Carlo simulation
Forecasting by replaying historical throughput in random order thousands of times and reading the distribution of completion dates. Produces a date with a confidence level rather than a single guess. See Monte Carlo forecasting.
Percentile
The value below which a given proportion of observations fall. The 85th percentile cycle time is the duration 85 percent of items came in under. Preferred to the average for any right-skewed distribution, which cycle time always is. See cycle time scatterplots.
Pull system
A system where new work is started only when capacity frees up, rather than assigned in advance. The mechanism that makes WIP limits enforceable, and the reason a prioritised list works without anyone allocating work.
Queue
Work waiting rather than being worked on. Queues are where most elapsed time goes in a software system and they are usually invisible, because a queue between two stages belongs to neither.
Right-sizing
Asking whether an item is larger than some threshold, typically about a week, and splitting it if so. Replaces effort estimation with a single binary judgement at a fraction of the cost. See no estimates in practice.
S to W
Service level expectation
A statement of the form "85 percent of our work is delivered within eleven days", derived from measured percentiles. A commitment about the system's observed behaviour rather than a prediction about one item. See cycle time scatterplots.
Throughput
The count of items finished per unit of time, usually per week. Requires no estimation and, unlike velocity, survives the removal of sprints. See throughput.
Upstream kanban
Managing requests before they are committed to: capturing, triaging and deferring options rather than accepting everything into the backlog. Where most of the lead time gap usually lives. See ideas to implementation pipeline.
Velocity
Story points completed per sprint. Undefined without sprints, gameable through estimate inflation, and replaced by throughput in any flow-based system. See measuring team productivity.
Work in progress
The count of items started but not finished, right now. The one variable in Little's Law you control directly and immediately, and therefore the fastest lever on delivery time. See WIP limits.
Work item age
See aging work in progress, above. The two terms are used interchangeably.
Where to Start
If you are new to this vocabulary, four terms carry most of the weight: cycle time, throughput, work in progress and work item age. Everything else in this glossary is derived from those four or is a way of displaying them. The flow metrics guide covers them in order.
See Your Real Flow Metrics
Cycle time, throughput and aging work in progress, calculated from the work you already track. No story points required.
Open a Free WorkspaceMore from Flow Metrics: The Complete Guide for Teams Without Sprints
Flow Metrics: The Complete Guide for Teams Without Sprints
Cycle time, lead time, throughput, WIP and flow efficiency explained for small teams. What each one tells you, how to read it, and what to do when it goes wrong.
MetricsCycle Time vs Lead Time: What Each One Actually Tells You
Cycle time measures your team. Lead time measures the customer experience. The gap between them is the number nobody looks at, and usually the one that matters.
MetricsHow to Read a Cumulative Flow Diagram (And What a Broken One Looks Like)
Every guide explains the coloured bands. Almost none show you the four shapes that mean something is wrong. Here are both, with the fix for each.
MetricsThroughput: The Only Delivery Metric That Survives Without Sprints
Velocity is undefined the moment you drop sprints. Throughput keeps working, needs no estimates, and is the raw material for every forecast you will make.