Deming and the quality culture: measure to improve

The Mistake Almost Every Small Business Makes with Quality

Most small businesses check quality at the end. After the product has shipped, after the customer has complained, after the error has already cost money.

That is not quality management. That is damage control.

The difference is not semantic. It is operational. And W. Edwards Deming explained it with a clarity that remains relevant more than seventy years later: quality is not inspected at the end — it is built into the process. To build it, you have to measure. To measure, you have to understand variability. And to understand variability, you first have to stop believing that processes are static.

In this article I translate Deming's ideas into a language that is useful for a small business in Latin America. No unnecessary jargon. Focused on what you can do this week.

Who Deming Was and Why You Should Care

Deming worked in the U.S. war industry during the 1940s. His mission: improve production quality without increasing costs. What he discovered was not a manufacturing trick. It was a universal principle.

Quality does not depend on working harder. It depends on understanding the system.

After the war, Japan hired him to help rebuild its industry. The results are history: the Japanese industrial transformation of the 1950s and 1960s has Deming as one of its principal architects. Companies like Toyota did not emerge by accident. They emerged from a culture that learned to measure, to analyze, and to improve systematically.

For a small business in Costa Rica, Mexico, or Colombia, the message is the same: you do not need more resources to improve. You need more information about what is already happening.

The Principle Deming Put at the Center: Variability

This is the concept that is hardest to accept.

We tend to think that a well-defined process produces identical results. It does not. An order that normally takes three days may take two or four. An invoice that is normally error-free may have one in every twenty. A response time that averages eight minutes may range between five and fifteen.

That variation is normal. It is inherent in any human or mechanical system.

What is not normal is ignoring it.

Deming distinguished between two types of variation:

The most common mistake: treating common cause variation as if it were special cause variation. Reacting to normal fluctuation as if it were an emergency. That does not improve the process. It destabilizes it.

To know which type of variation you have, you need data. To have data, you need to measure.

PDCA: The Cycle That Turns Data into Improvement

The PDCA cycle is not complicated. It is discipline.

Plan: Define what you want to improve, what the specific problem is, what your hypothesis is about its cause, and how you will measure it. Without this phase, any action is improvisation.

Do: Implement the change at a small scale. Do not reform the entire operation at once. Test in one process, one shift, one week.

Check: Compare actual results against what you expected. Did the indicator improve? By how much? Did any side effects appear that you did not anticipate?

Act: If it worked, standardize the change. If it did not, learn why and start the cycle again with adjusted hypotheses.

What makes PDCA powerful is not its complexity — it is its absence. There is no room for unverified intuition. No room for "that is how we have always done it." Every cycle demands evidence.

How to Measure Without Being a Statistician: The Invoice Example

This is where many small businesses get stuck. "Measuring processes" sounds like an industrial laboratory. It is not.

Here is a concrete example from the world of finance.

Imagine you have 500 invoices issued during the quarter and you want to estimate how many contain some type of error before they reach the customer. You do not have time to review all 500. You do not need to.

With a sample of 50 invoices selected in a random and representative way — what in statistics is called stratified random sampling — you can estimate the error rate for the total with reasonable confidence. The standard error of the mean tells you how precise your estimate is from that sample.

What do you need for this in practice?

• Define what counts as an error (clear criteria, not subjective).
• Select the sample randomly (do not pick the ones you "think are fine").
• Record the results in a simple spreadsheet.
• Calculate the proportion of errors in the sample.
• Repeat the process each quarter to track the trend.

That is statistical process control without complex formulas. The principle is the same one Deming applied in wartime factories: you do not need to review everything if you measure a representative portion correctly.

Three Quality Culture Practices for a Real Small Business

Deming wrote for large organizations. But his principles scale down. These three practices are direct adaptations for a small business with a lean team and limited resources.

1. Choose Three Process Indicators and Measure Them Every Week

Not ten. Three.

Each indicator should answer a concrete operational question: how long does delivery take? what is our error rate in this process? how many customers make a repeat purchase within 90 days?

Record in a simple table. Compare week over week. Do not react to one bad week — observe the trend.

2. Distinguish Between Noise and Signal Before Acting

When a number drops in a given week, the right question is not "what did we do wrong?" The right question is: is this drop within the normal range of variation, or is it outside it?

If your average delivery time is three days with typical variation of one day up or down, a result of four days is noise. Do nothing. If the result is seven days, that is a signal. Investigate the cause.

Acting on noise does not improve the process. It makes it more unstable.

3. Hold a Monthly Root Cause Review

Once a month, take the indicator that varied the most and ask: why did this happen? Not the first answer — the answer behind the answer. Toyota's "5 Whys" technique — asking why five consecutive times — is simple and effective.

An order arrived late. Why? Because the supplier delivered late. Why? Because we did not confirm the order far enough in advance. Why? Because there is no defined confirmation process. There is the root cause. There is the improvement.

What Is Not Measured Cannot Be Improved

This phrase attributed to Deming is the summary of everything.

Not because measuring is an end in itself. But because without data you cannot distinguish between real improvement and the illusion of improvement. Without data you cannot know whether a change worked or whether results improved for other reasons. Without data, your management rests on intuition — and intuition scales very poorly.

A quality culture does not require an expensive consultancy or an ISO certification. It requires three things:

1. The decision to measure: choose what matters and record it consistently.
2. The discipline to interpret: read the data in context before reacting.
3. The habit of improving: use each PDCA cycle to learn something that changes the process.

Deming demonstrated this in wartime industry and in Japan's reconstruction. The principle does not change for a small business in Latin America. The scale changes — the logic does not.

Ready to implement a continuous improvement system in your business without overcomplicating operations? Schedule a diagnostic session and let's identify which processes to measure first.