CTQ Control Plan: Keep Your Gains After the Project Team Leaves

The CTQ Control Plan sits in the Control phase of RDMAICS. Its job is narrow and useful. It maps each Critical-to-Quality characteristic to its control method, target, and tolerance to ensure CTQs remain in control post-improvement.

You use it near the end of a project, once you know which changes worked. By that point you have run the Improve phase, you have seen the process shift, and you are about to hand the process back to the people who run it every day. That handover is where gains are lost. Not because anyone intends to let them go, but because nobody wrote down what to watch, how often, and who is responsible when the number moves.

The tool forces those answers into a single row per CTQ. One characteristic. One target. One limit. One measurement. One control method. One frequency. One owner. If a row is incomplete, you can see it immediately, and so can everyone else in the review.

The screenshots in this guide show the tool loaded with its built-in sample data. The case is fictional. It is there so you can see the mechanics of the tool without having to supply your own numbers first.

The sample case is an accounts payable process. Four CTQs carry the result of the improvement work. Invoice accuracy, with a target of at least 98 per cent. Processing cycle time, with a target of two days or less. On-time payment, with a target of at least 97 per cent. Duplicate payment rate, with a target of 0.2 per cent or less.

Those four are a reasonable set for a transactional process. Two are quality measures, one is a speed measure, and one is a defect measure. Between them they cover the ways the process can fail a customer: wrong amounts, slow payment, missed due dates, and money paid twice.

Each row in the worksheet needs seven fields. The CTQ name, the target, the tolerance or spec limit, the measurement, the control method, the frequency, and the owner. The sample data fills every one of them.

Note how targets and limits are stated differently. Invoice accuracy has a target of at least 98 per cent and a lower spec limit of 98 per cent. Processing cycle time has a target of two days or less and an upper spec limit of two days. The direction matters. A lower spec limit means more is better. An upper spec limit means less is better. Duplicate payment rate carries an upper spec limit of 0.2 per cent for the same reason.

The measurement column is where many plans go thin. The sample is specific. Invoice accuracy is measured as error rate per cent. Processing cycle time is measured as hours per invoice. On-time payment is measured as the percentage paid on or before the due date. Duplicate payment rate is measured as duplicates per thousand. Each one names a unit, so two people reading the plan will calculate the same number.

Frequency and owner close out the row. Invoice accuracy is checked daily by the AP Manager. Processing cycle time is checked daily by the Green Belt. On-time payment is checked weekly by the AP Manager. Duplicate payment rate is checked weekly by the AP Clerk.

The tool returns a control output summary above the detail. For the sample data it reports four CTQs from the worksheet. Four of four have a target. Four of four have a limit. Three owners are named across the four rows.

The method breakdown reads: p-chart on 2 CTQs, I-MR chart on 1 CTQ, Run chart on 1 CTQ. The tool also states plainly that every CTQ has a target and a limit. That single line is the completeness check. If one row were missing a limit, the count would read three of four and you would know where to look.

Below the summary the tool lists each CTQ as its own card, numbered one to four, showing target, tolerance or spec limit, measurement and control method. The cards are the format you want in a handover pack. Someone picking up CTQ 3 sees a target of at least 97 per cent, a lower spec limit of 97 per cent, a measurement of percentage paid on or before due, and a run chart as the control method.

The tool also offers an AI review that explains the result in plain language, which is useful when you are presenting the plan to people who did not sit through the project.

A completed control plan is not a document for the shelf. It drives a small set of recurring decisions once the project team has moved on.

In the sample data, the owner count of three across four CTQs is itself a decision point. One person holds two rows. That may be correct, because the AP Manager owns both invoice accuracy and on-time payment, which are closely related. Or it may be a load that needs splitting. The tool shows you the count and leaves the judgement to you.

The first mistake is listing too many CTQs. Four is a workable set. Twenty rows will not be checked daily by anyone. If a characteristic does not change the customer experience, it belongs in a different document.

The second mistake is a target without a limit, or a limit without a direction. A target of at least 98 per cent with no spec limit gives the owner nothing to breach. The sample avoids this by pairing each target with an explicit LSL or USL.

The third mistake is naming a function rather than a person or role as owner. The sample names AP Manager, Green Belt and AP Clerk. Those are roles that a specific person fills. Naming the department means nobody is accountable.

The fourth mistake is choosing a control method out of habit. The sample uses a p-chart for the two proportion measures, an I-MR chart for the continuous cycle time measure, and a run chart for the weekly on-time percentage. The method should follow the data type, not the preference of whoever filled in the form.

The plan earns its value in the weeks after the project closes. Two things keep it alive. The first is that the checks actually happen at the stated frequency. Daily means daily, and a daily check that slips to weekly is a quiet change to the plan.

The second is a short review cycle. Put the plan in front of the owners on a set rhythm and ask three questions. Is each CTQ still the right thing to watch. Is each target still correct. Is each owner still in the role. Record any change on the row rather than in a side conversation.

If a CTQ holds inside its limit for a long stretch, that is not a reason to stop measuring. It may be a reason to reduce frequency, or to tighten the target. Both are decisions to make on purpose and to write down.

Be clear about the boundaries. The CTQ Control Plan documents your control approach. It does not collect data, plot the charts, or calculate control limits. The sample row says p-chart, but you still need a p-chart somewhere with real readings in it.

It does not tell you whether your CTQs are the right CTQs. It counts what you entered. In the sample it reports four CTQs, four with a target, four with a limit and three owners. Those counts confirm the plan is complete. They say nothing about whether invoice accuracy at 98 per cent is the level your customers need.

It does not test capability. A target of two days or less with an upper spec limit of two days tells you what you want. It does not tell you whether the process can deliver it. That work belongs to the Measure and Analyse phases.

It also cannot enforce anything. A named owner on a row does not create accountability by itself. The plan is a record of an agreement. Keeping the agreement is a management task, and the tool only makes it easy to see when the agreement has been broken.