A control plan is a written document that lists, for every step of a process, what to check, how to measure it, how often to sample, and what to do when a result falls outside limits. A solid control plan example manufacturing teams can borrow is a table with one row per characteristic and one column for each of those answers. Operators use it on the floor, quality engineers use it to audit, and customers use it to judge whether you can hold your process steady.
This article shows you why the document matters, how to build one in eight steps, and a worked example for a machined shaft with real numbers, sampling math, and a reaction plan you can copy. It closes with how to keep the plan current and the mistakes that make plans useless.
Control plans are the main output of the Control phase in DMAIC, the last stage of a Lean Six Sigma project. Lean Six Sigma Experts has helped organizations lock in process gains since 2011, and a vague control plan is a quick way to lose them.
Why a control plan matters in manufacturing
A control plan matters because it turns what your best operator knows into a standard that everyone follows. It keeps your process stable, and it keeps improvement gains from fading.
It keeps the process the same on every shift
Without a plan, the second shift measures at a different frequency than the first, adjusts at a different point, and scraps at a different threshold. Variation between people and shifts costs as much as variation in the machine. A written plan fixes who checks what, with which gauge, and how often, so the answer does not depend on who is working.
It protects the gains from improvement projects
Improvement projects fade when nobody owns the new settings. That is why DMAIC ends with a Control phase, and the control plan is its main deliverable. Say a project cuts scrap from 4% to 1%. The plan records which settings and checks produced that result, so the next engineer cannot quietly undo it.
Customers also expect the document. Automotive suppliers working under IATF 16949 are required to maintain control plans, and many other customers ask for one during supplier audits.
A control plan is how an improvement becomes the standard way of working instead of a one-time result.
How to build a control plan for your process
To build a control plan, list every process step, choose the characteristics that matter, and answer the same six questions for each one. Then test the draft on the floor before you release it.
The six core elements
Every row of a manufacturing control plan should answer these:
- Process step: the operation, such as receiving, turning, deburring, or final inspection.
- Characteristic and specification: what you control and its limit, such as a diameter of 20.00 ± 0.02 mm.
- Measurement method: the gauge or technique, such as a micrometer, a profilometer, or a visual check against a boundary sample.
- Sample size and frequency: how many parts, and how often.
- Control method: how you keep the characteristic in limits, such as an SPC chart, a fixture, or a checklist.
- Reaction plan: what happens when a result is out of limits.
Some templates add columns for the responsible person, the machine, or the drawing revision. Add them if your team needs them, but keep all six of the core columns.
Eight steps to a first draft
- Draw the process flow from receiving to shipping.
- Pull critical characteristics from the customer drawing, your PFMEA, and past defect data.
- Write each specification and tolerance from the drawing, not from memory.
- Pick a measurement method, then validate it with a gage R&R study. Under 10% of tolerance is generally considered good, 10% to 30% is marginal, and over 30% is not acceptable.
- Set sample size and frequency by risk and capability. Sample more at startup and after changes, and relax only when the data shows stability.
- Choose a control method, such as SPC charts, mistake-proofing, or a checklist.
- Write a reaction plan with a clear stop, contain, fix, and notify sequence.
- Trial the plan with operators, revise it, train everyone, and release it under revision control.
Validate the gauge before you trust the data, because a control plan built on a noisy measurement only documents guesses.
A control plan example for manufacturing
The example below covers a CNC-turned steel shaft. The values are illustrative, so replace them with the numbers from your own drawing and customer requirements.
Example: a turned shaft
| Step | Characteristic and spec | Measurement method | Sample size and frequency | Control method | Reaction plan |
|---|---|---|---|---|---|
| 10 Receive bar stock | Material grade matches drawing | Mill certificate review | Every lot | Receiving checklist | Hold lot, notify quality, return to supplier if certificate is missing |
| 20 CNC turning | Diameter 20.00 ± 0.02 mm | Digital micrometer | 5 consecutive parts, hourly | X-bar and R chart | Stop, contain parts since last good check, adjust tool offset, re-measure |
| 20 CNC turning | Surface finish Ra 1.6 µm max | Portable profilometer | 1 part at shift start and after each tool change | Tool-life counter, check sheet | Replace insert, recheck 3 parts, sort suspect parts |
| 30 Deburr and wash | No burrs or sharp edges | Visual check against boundary sample | Every part | Operator self-check | Rework part. Three consecutive failures stop the line and call the supervisor |
| 40 Final inspection | Length 150 ± 0.1 mm | Calipers | 5 parts per lot | Final inspection record | Hold lot, sort 100%, notify process engineer |
How to read the example
The sampling math shows how the plan works. If the lathe makes 60 parts an hour, five measured parts per hour is 40 parts checked per shift. If the check at 10:00 fails, the worst-case containment is 60 parts, everything made since the 9:00 check. A shorter interval shrinks that pile, and the cost of extra measuring is the trade-off.
The reaction plan matters more than most people think. "Notify supervisor" is not a reaction plan. A good one names the stop, the containment scope, the fix, and the restart check. Use this structure for each characteristic:
Characteristic: [name and spec]
If out of limits:
1. Stop the process and tag the machine.
2. Contain every part made since the last good check.
3. Correct the cause: [adjustment or action].
4. Re-measure [number] parts before restarting.
5. Notify [role] and record the event on [form].
A reaction plan that names the stop, the containment, and the restart check removes guesswork at the moment an operator needs it most.
Keeping a manufacturing control plan accurate
A control plan stays useful only if you revise it whenever the process, the product, or the data changes. Treat it as a living document, tied to your PFMEA.
Prototype, pre-launch, and production versions
The APQP framework used in automotive manufacturing defines three control plan phases:
- Prototype: covers the dimensional measurements and material and performance tests during prototype build.
- Pre-launch: covers the controls after prototype and before full production, usually with heavier inspection.
- Production: covers full documentation of product and process controls during volume production.
Even outside automotive, the same logic works. Start with tight sampling and loosen it as capability data proves the process is stable. Update the plan after any change in tooling, material, supplier, or machine, and whenever the PFMEA changes.
Mistakes that make a plan useless
- Listing every drawing dimension: the plan buries the critical few. Focus on characteristics that affect fit, function, or safety.
- Writing vague reactions: operators improvise, and results vary by shift.
- Freezing sample rates: if a process runs clean for months, revisit the frequency in both directions.
- Skipping gauge validation: the data looks precise but misleads.
- Filing it in a binder: a plan that is not posted at the workstation and in daily use does not control anything.
A control plan nobody opens during production is paperwork, not control.
Start with one process
A control plan is a table that answers six questions for every critical characteristic: what step, what spec, how measured, how often, which control, and what reaction. Build it from your process flow and PFMEA, validate your gauges, and trial it with operators before release. Then revise it as the process changes.
Pick one high-risk process and draft a plan using the shaft example as your model. Our Green Belt Certification Training covers the Control phase through hands-on exercises if your team wants structured practice. If you want help writing control plans that hold up on the floor and in customer audits, contact Lean Six Sigma Experts to talk through your process.