Implementation Guide
Standardized Work.
Standardized work is the current best known way to do one job, written down with the people who do it so it can be trained, audited and improved. It fixes three things: the takt time the job must meet, the work sequence, and the standard work in process. A standard is written to be changed. Without one, every improvement evaporates, every audit is opinion, and every training is folklore.
What standardized work is
Standardized work is the agreed, current best known way to perform one job, captured precisely enough to train from, audit against, and improve upon. In its classic form it fixes exactly three things for one operator: the takt time the job must meet, the sequence of steps, and the standard work in process the job needs in order to flow. Everything else, methods, key points, tolerances, lives in supporting documents underneath.
The definition carries two ideas most plants miss. "Current best known" means the standard describes how the job is actually best done today, not how a process engineer would like it done. And a standard is written to be changed: it exists so the next improvement has a baseline to beat and a place to be recorded. A standard is not a constraint on improvement. It is the memory that keeps improvement from evaporating.
Without standards, three everyday management activities quietly turn into fiction. Improvement cannot be verified, because there is no defined before. Audits become opinion, because there is nothing agreed to audit against. And training becomes folklore: whatever the nearest experienced operator happens to remember, minus whatever they forgot to mention.
The three elements
The example column follows one assembly cell at the 450-person components manufacturer used across these guides: demand of 456 units per day and 54,000 seconds of daily working time. The numbers are illustrative but internally consistent.
| Element | What it fixes | Assembly cell example (illustrative) |
|---|---|---|
| Takt time | The rhythm demand sets: available working time divided by customer demand. Every standard is designed so the job's work content fits inside it. | 54,000 seconds of daily working time for 456 units gives a takt of 118 seconds. Each station's work content must fit under 118 seconds. |
| Work sequence | The order in which one operator performs the steps of the job. Not always the order the material moves, and that difference is deliberate. | Seven steps: pick housing, press bearing, fit shaft, torque both fittings, leak test, label, place on the outbound rail. |
| Standard work in process | The minimum number of pieces inside the process needed to run the sequence smoothly, pieces held in machines included. More is inventory; less is waiting. | Two pieces: one in the press, one at leak test, so the operator never waits for a machine cycle to finish. |
Where it sits in the transformation roadmap
On the manufacturing transformation roadmap, standardized work belongs to the Standard Work & 5S practice in the Run Daily stage. It runs in parallel with lean daily management, and the two hold each other up: standards give the daily system something definite to defend, and the daily boards make deviations from standard visible within a shift instead of within a quarter. Its physical partner is 5S: a workplace organized well enough that following the sequence is the easy option.
- Before Baseline KPIs
- You are here Standard Work & 5S
- In parallel Daily Management
- After Structured Problem Solving
When it fails before it starts
Standardized work is close to free and pays back fast, so the honest warnings are few. But two starting conditions reliably kill the effort, and two intentions reliably corrupt it:
- The process is not minimally stable. If machines break daily and material arrives whenever it arrives, the same job genuinely cannot run the same way twice. Reduce the biggest sources of chaos first, then standardize what remains.
- The plan is to write standards for everything at once. That is a documentation project, measured in pages. The floor watches it happen, then ignores the result.
- The trigger is an upcoming audit or certification. Standards written for a binder read like binders. Write them for the operator, and the auditor gets what they need anyway.
- Nobody will own the updates. An unowned standard is accurate on the day it is printed and fiction within a quarter.
How to create one standard that lives
One standard that is followed, audited and revised is worth more than a hundred that are filed. The sequence below produces one living standard in about a week of elapsed time, most of it waiting for the trial shift. Repeat it job by job, pulled by problems and training needs.
| Step | What happens, and why |
|---|---|
| Pick one job that matters | Repeat quality problems, wide operator-to-operator variation, or a new hire arriving next month are all good reasons. A plant-wide documentation drive is not. Start where a standard changes this quarter's results. |
| Observe the work with the operators | Time several cycles of everyone who does the job, at the process. Note where operators differ and ask why. Almost every difference exists for a reason, and those reasons are where the best method hides. |
| Write the current best known way | Combine the best elements you observed into one sequence the operators agree is the best way known today. Not the ideal way, not the engineering way: the way that works at takt on a normal Tuesday. |
| Trial it for one shift | The person who wrote it stands at the line while one operator runs it. Every hesitation, workaround and missing detail found in that shift is a revision you did not have to discover through a defect. |
| Revise, then train everyone | Fold in what the trial taught, date the chart, and train every operator who does the job, using the job breakdown sheet, at the process, not in a meeting room. |
| Audit it, and name the owner | Put the job on the layered audit rotation and give the standard one owner who accepts or rejects every future change. A standard without an owner is already decaying. |
Who owns the standard
The area's team leader owns the standard and every update to it. Engineering owns the limits inside it: tolerances, torque values, safety rules. The split matters because ownership decides behavior. Standards written by engineering alone get laminated; standards written with the operators get followed, because the people who must live with the sequence chose it. A CI office can facilitate the first few, then it should get out of the way.
Updates are triggered by three events: a kaizen that found a better way, a problem countermeasure that changed the method, and an audit finding that the standard is not followable. The owner's job is to keep the paper matching reality within days of the change, retrain the crew, and re-date the chart. A standard whose date is older than the last change to the process is not a standard. It is a historical document.
Which document does what
Standards most often fail at the line for a boring reason: everything was written into one document. The chart, the method, the training detail and the news of the week have four different readers and four different shelf lives, so they belong in four documents.
| Document | What it is for | Who uses it | Level of detail |
|---|---|---|---|
| Standardized work chart | Fixes takt, sequence and standard WIP for one operator on one page, usually with a small layout sketch and step times. | Team leader and operator at the station; the reference for every audit. | One page. Steps and times, not methods. |
| Work instruction | Describes how each step is performed: quality key points, safety points, tolerances, torque values. | Operators during and after training; auditors checking method. | A few pages per operation. Photos over paragraphs. |
| Job breakdown sheet | The training document: major steps, key points, and the reasons behind the key points. | Trainers during on-the-job instruction. | Detailed on knacks and reasons, silent on anything the trainee can see for themselves. |
| One-point lesson | Teaches one specific point: a new defect to watch for, a machine quirk, a recent change to the method. | The whole team, posted at the station. | One page, one point, five minutes to read, mostly picture. |
The test for each level is time. An auditor has a minute, a trainer has an hour, an operator mid-shift has ten seconds. If a reader cannot find what they need in the time they realistically have, the content is sitting at the wrong level of the hierarchy.
Auditing without policing
A standard nobody checks is a suggestion. But the check is where most programs go wrong, because they audit the operator when they should be auditing the standard. The working rule: an audit checks that the standard is followable, not just followed. Stand at the process, ask the operator to show you the job, and watch whether the sequence on the chart survives contact with a real shift at takt.
A violated standard is data, and it allows exactly two responses. Either the operator was never properly trained to it, in which case train them, or the standard cannot be followed as written, in which case fix the standard. There is no third response called discipline. The first time deviation is punished, deviation goes underground, and from that day the audit trail is green and useless.
Run the checks as layered process audits: short, frequent, question-level checks performed by every layer of management, at the process, asking "show me" rather than "did you". Team leaders check daily, area managers weekly, plant leadership monthly, and every finding gets a disposition: fix, train, or change the standard.
Standards and improvement
The relationship between standards and kaizen is a loop, not a tension. The standard is the baseline that makes improvement measurable, and improvement is what keeps the standard alive. The operating rule is blunt: a kaizen changes the standard, or it did not happen. An improvement that lives only in one shift's habits is not an improvement. It is a variation the night shift never heard about and the next new hire will never learn.
This makes revision history the health record of an area. A chart on revision 7, with dates spread across the year, describes a team that is learning. A chart on revision 1, dated the year the line was installed, describes either a perfect process or, far more likely, nothing at all. Leaders walking the floor learn more from revision dates than from compliance scores.
Standards also feed structured problem solving, the next practice on the roadmap. A problem is a gap between standard and actual, so without the standard half of a plant's problems cannot even be stated, let alone solved.
Common mistakes
What bad looks like
- Standards written by engineering in an office, laminated, and never touched again
- 40-page instructions that nobody opens at the line
- A plant-wide documentation drive measured in pages produced
- Revision dates older than the newest machine in the cell
- The standard pulled out as evidence against an operator after a problem
What good looks like
- One page per job, visible at the station, dated and owned by the team leader
- Written with the operators who do the work, in their words
- Trialed for a shift before release, revised from what the trial taught
- Audited weekly by leaders who ask whether it is followable
- A revision history that names the kaizen behind each change
What happens next
Once standards exist, two capabilities become possible that were not before. Structured problem solving gets a definition of "problem" and a place for every countermeasure to land: the measure of a fix is which standard it changed. And the skills matrix gets something real to qualify people against: performing to standard with support, performing to standard at takt alone, training others to the standard. Without standards, a skills matrix measures attendance.
The unglamorous half of the work is keeping the paper true, and that is where a management system, and TeamGuru, earns its place. Standards live versioned next to the work in Documents, so the chart the operator sees is the current revision and not a photocopy of revision 1, and the layered audit rotation runs as scheduled checklists and audits whose findings become owned, dated actions instead of margin notes.
On the roadmap, the practice that keeps all of this honest over years is the same one that started it: leaders returning to the process on a rhythm, checking that standards are followable, and treating every deviation as the day's most useful piece of data.
