There is a frustration in business that deserves a warning label.
Your team spends weeks investigating a problem, finds the causes, changes the process and celebrates a genuine improvement.
Everyone is delighted. Then, a couple of months later, the old problem returns as though it has been on holiday.
This is where the Lean Six Sigma Control phase earns its keep. It is not the glamorous part of improvement work.
Nobody usually asks for a photograph of the new checklist. But if you want the results to survive staff changes, busy Mondays and the inevitable arrival of somebody saying, “We’ve always done it the other way”, Control matters enormously.
In this article, we’ll look at what the Control phase actually does, how to build a practical control plan, when to use simple monitoring or statistical process control, and what a Green Belt candidate should understand.
You can apply the ideas whether you work in manufacturing, IT, customer service or a small office where the printer appears to have a personal grudge.
Why Good Improvements Quietly Disappear
Improvement teams naturally concentrate on solving the problem in front of them. They map the process, collect data, find likely root causes, test changes and show that performance has improved.
That is valuable work. Unfortunately, it does not guarantee that the new way of working will become the normal way of working.
The new procedure may be hidden in a forgotten folder, or one enthusiastic employee may be the only person who understands the checks. The organisation has not yet learned to maintain the gain.
Lean Six Sigma tackles existing-process problems through DMAIC: Define, Measure, Analyse, Improve and Control.
In plain English, the final phase asks a deceptively simple question: “what must happen on an ordinary Tuesday to stop us drifting back to where we started?”
A project is not truly finished just because its presentation looks wonderful.
Control Is Not Another Name for More Paperwork
A control is anything sensible that helps an improved process keep delivering the intended result. It might be a clear work instruction, an automated validation rule, a short daily check or a named person who investigates an unusual result.
The best control is often the one that makes the right action easier than the wrong one.
A control plan brings these arrangements together. It identifies the important measures, who checks them, how frequently they are reviewed, what signals trouble and what action follows.
Think of it as the operating instructions for keeping the gain, rather than a ceremonial document to file away after the project party.
A single page may be enough for a modest office process. Match controls to the consequences of failure.
A Fictional Example: The Missing Information Problem
Imagine a fictional business called Northbridge Office Supplies. Its customer-service team handles requests for replacement products.
Too many requests arrive without the order number, delivery address or reason for replacement, so staff repeatedly contact customers to collect information they should have received at the start.
For illustration, suppose the team reviews 200 requests in one period and finds that 36 need rework because essential information is missing. That is 18%.
After investigating the causes, the team introduces a clearer online form with required fields and tests the revised process. In another illustrative group of 200 requests, 12 need rework, or 6%.
Those figures are fictional teaching data, not evidence from a real company, and two snapshots alone would not prove that the improvement is stable.
Still, the apparent improvement is worth protecting. What happens when a new employee joins? What if the online form changes? What if customers start using a different route to submit requests?
Build a Control Plan That People Will Actually Use
Start with the process owner. That is the person accountable for how the process performs after the improvement project has finished.
The project team can design a clever solution, but it should not remain the permanent emergency service for a process that belongs to somebody else.
For Northbridge, the customer-service manager agrees to own the process. The team defines one main outcome measure: the percentage of replacement requests requiring rework because information is missing.
It also chooses a useful leading measure: the percentage of requests submitted through the validated online form. The outcome tells us whether the problem is returning; the leading measure may help explain why.
Next comes a clear routine. An administrator records the relevant counts each week. The manager reviews the trend every Monday.
Finally, the team records the response:
- Who investigates?
- How quickly?
- Where are findings noted?
- Who can change the form?
- How will people know that a corrective action worked?
Without these answers, a red number on a dashboard is merely a colourful way of saying that somebody else has a problem.
Make the Better Method the Normal Method
Standard work simply means documenting the current best-known way to perform a task so people can repeat it consistently.
At Northbridge, that could mean a short instruction for handling exceptional requests, an accessible form guide and a five-minute introduction for new starters.
Where possible, mistake-proof the process. In Lean Six Sigma this is often called poka-yoke.
A required order-number field can prevent a common omission before it reaches the customer-service team. It will not solve every problem, but prevention is usually more dependable than hoping somebody remembers to spot the error later.
Keep controls proportionate. Duplicating a reliable automated check can create fresh waste.
Watch the Process, Not Just the Monthly Target
Processes vary over time. A single good result is not proof of permanent success.
For a straightforward service process, a run chart showing weekly results in time order may be a useful starting point. It can reveal trends, sudden changes and the timing of interventions.
More formal statistical process control, or SPC, uses appropriately selected control charts and statistically determined control limits to distinguish routine variation from signals that the process may have changed.
There is an important distinction here. A customer target, such as “fewer than 8% of requests require rework”, is not automatically a control limit.
Targets describe what the business wants; control limits are calculated from process behaviour using suitable methods and data.
Do not draw arbitrary lines on a chart, call them statistical limits and hope the examiner is feeling generous.
If the process is high volume or consequential, a trained practitioner can select the correct chart for the data and sampling method.
For a small team, begin with reliable definitions, consistent data collection and a named reviewer. Sophistication is useful only when it improves decisions.
What to Do When Performance Starts to Slip
Suppose Northbridge’s weekly rework percentage begins to rise.
The manager should not immediately accuse the team of being careless or change the entire process.
First, confirm the data: are we counting requests consistently, and has the mix of work changed?
Then investigate what actually happened. Did a software update remove a required field? Are customers using an email address that bypasses the form? Has a new employee missed the training?
Once the cause is understood, take an appropriate corrective action and check whether it works.
If the form has been altered, restore the validation and test it.
If the work has shifted to another channel, decide whether that channel needs a comparable safeguard. Record the action so the next person does not have to rediscover the same problem.
Avoid constantly changing the rules in response to ordinary fluctuations. Investigate meaningful signals and respond proportionately.
The Handover That Makes Control Real
Before closing an improvement project, arrange a short, practical handover with the process owner and the people who do the work:
- Demonstrate the revised method.
- Show where the measures come from.
- Agree the review routine and the response to a warning signal.
- Make sure any automated checks are maintained by somebody who will still be there after the project team moves on.
A useful test is to imagine that the improvement lead takes a fortnight’s holiday.
Can the team explain the new process, find the latest performance figures and respond to an emerging problem without calling that person? If not, the handover probably needs another pass.
Schedule a follow-up review to check that the controls work under normal operating conditions.
Three Mistakes Worth Avoiding
The first mistake is measuring everything. A dashboard containing twenty-seven indicators can look impressive, but it may leave nobody clear about which two or three actually matter.
Choose measures that connect directly to the problem and the customer’s experience, then expand only when the extra information helps a decision.
The second mistake is writing controls without consulting the people who use them.
A procedure invented in a meeting room can collapse on contact with the real working day. Ask staff to test the new routine, explain the exceptions and point out where the process still creates unnecessary effort.
The third mistake is declaring victory before ownership has transferred.
The project manager may be excellent at getting a change introduced, but the process owner must be equipped to maintain it.
Otherwise the new system becomes one more clever initiative that works beautifully until its champion leaves the room.
What Lean Six Sigma Green Belt Candidates Should Remember
For Green Belt study, understand Control as part of the whole DMAIC story, not as an isolated definition to memorise:
- Define establishes the problem
- Measure gives you credible evidence
- Analyse investigates causes
- Improve tests solutions
- Control protects the result.
A strong examination answer should connect the proposed control to the actual failure mode and the way the process operates.
Be comfortable explaining the purpose of a control plan, standard work, mistake-proofing, monitoring and an appropriate reaction plan.
Understand the difference between common-cause and special-cause variation, and between specification or business targets and statistically calculated control limits.
Projex Academy’s Lean Six Sigma Green Belt information includes Control, SPC, control charts and performance measurement among the relevant learning areas.
Choose the tool that provides trustworthy evidence and enables the right person to act
The Lesson to take Back to Work
A successful improvement should not depend on the permanent presence of the person who invented it.
The process needs a clear owner, a sensible way of working, a small number of meaningful measures and an agreed response when something changes.
That is the real achievement of the Lean Six Sigma Control phase. Not another folder of documents, and certainly not a collection of charts that nobody understands.
It is making the better way of working ordinary enough to survive a busy Tuesday.
If you would like to develop these skills in a structured way, the Projex Academy Lean Six Sigma Green Belt learning material covers the DMAIC journey and practical improvement techniques.
Explore the training, see whether it suits your goals, and start with a process you would genuinely like to improve. The best place to learn Control is often the very next problem that threatens to come back.
Relevant training: Lean Six Sigma Green Belt Masterclass and Guide to Implementing Lean Six Sigma.
