Lean Principles
Lean, applied to data work: define value from the decision a customer of the data actually makes, map the stream that produces it, and cut the waste — unread reports, unused dashboards, pipelines nobody consumes. Waste is invisible without measurement.
Lean Principles: Value First, Waste Out
Lean began in manufacturing (the Toyota Production System) and survived every technology shift since, because its core is not about factories: create more value with fewer resources by finding and removing waste. FORCE cares about Lean for one reason — data work accumulates waste faster than most work, because data waste is quiet. A machine idling is visible; a dashboard nobody opens is not. Lean gives Competence its pruning discipline, and measurement makes the invisible waste visible.
The five core principles
Define value. Value is defined by the customer of the work — and for data work, the customer is a decision. A report has value if a decision changes when it changes. If nobody would act differently without it, it is not value; it is output.
Map the value stream. Trace every step from raw input to the decision it serves. In data work: collection → storage → transformation → report → decision. Every step that doesn’t move toward the decision is a candidate for removal.
Create flow. Remove the interruptions between steps — handoffs that wait for a person, manual steps in the repeatable middle, approvals that exist out of habit. This is the same repeatable-middle logic as Workflow Automation, applied with scissors instead of scripts.
Establish pull. Produce what is demanded, not what is forecast. In data terms: build the report when a decision needs it, not because a template exists. A recurring report nobody pulled for is push-waste on a schedule.
Pursue perfection. Not a project — a cadence. Regularly re-measure the stream, re-ask what each step is for, and remove what no longer earns its place.
Where data work hides its waste
The classic Lean wastes have exact data-work counterparts, and each is findable with measurement you already have:
| Classic waste | Data-work counterpart | How measurement exposes it |
|---|---|---|
| Overproduction | Reports and dashboards nobody opens | Usage counts per report |
| Inventory | Pipelines and tables with no consumers | Query logs against each table |
| Waiting | Decisions queued on manual data pulls | Time from question to answer |
| Over-processing | Precision beyond what the decision needs | Ask what the decision changes at ±5% |
| Defects | Numbers that disagree between reports | One source of truth, or reconciliation cost |
The last row is the deepest link: a single source of truth is Lean applied to information itself — every duplicated, disagreeing copy of a number is inventory waste plus defect waste, paid for on every decision.
The discipline, honestly stated
Cutting waste is measurement plus nerve. The measurement says which reports go unread and which tables go unqueried; the nerve is retiring them despite the “someone might need it” reflex. Run the loop like any other experiment: pick a suspected waste, measure it, remove it reversibly, and watch whether any decision actually suffers. Almost none do — and every removal makes the remaining stream easier to trust, observe, and automate.
Conclusion
Lean’s five principles are a timeless answer to a permanent problem: work accumulates, and only deliberate pruning keeps value dense. In a data organization, the pruning shears are your own measurements — usage, queries, time-to-answer — and the courage to act on them. Define value by the decision, map the stream, keep it flowing, produce on pull, and prune on a cadence. That was true before databases existed, and it will outlive whatever replaces them.