The Standardize-Then-Improve Cycle

Why this matters

Most shops try to improve a process that was never pinned down in the first place, and then wonder why the improvement does not stick. If three techs do a job three different ways, "we improved it" means one of them changed something and the other two never heard. You cannot improve what you have not first agreed on. Standardizing is not the enemy of improvement, it is step one of it. This is the loop that turns a good month into a permanent gain instead of a story you tell once.

The loop in one line

Standardize the current best way, run it until you learn something, improve the standard, then re-standardize. Repeat. The point of writing down "how we do it now" is not to freeze the shop forever; it is to create a baseline you can measure a change against. Without a standard, every change is a guess with no control group.

Step one: standardize the current best way

Pick the best version you have today, not a perfect version you wish you had. Write it down plainly enough that a competent new hire could follow it: the steps, the order, the point where judgment is required. This is the baseline. It does two jobs at once. It makes the work repeatable, and it makes the work visible, so the next problem shows up against a clear background instead of getting lost in "well, everybody does it differently anyway."

A standard nobody wrote down is not a standard, it is a habit that lives in one person's head and leaves when they do.

Step two: run it and watch for signal

Now the standard earns its keep. Every callback, complaint, near-miss, and new-hire question is a test of the process against reality. A near-miss is something that almost went wrong but did not, a free warning you did not have to pay for. When the field keeps bumping into the same rough edge, that is not noise, it is the process telling you where it is weak. Because you have a written baseline, you can see exactly which step failed instead of arguing about whose fault it was.

Step three: improve the standard, not just the moment

When you find the weak step, fix the standard, not only today's job. Get to root cause, the underlying system flaw rather than the surface symptom or the person who happened to be holding it when it broke. Ask why the step failed, and keep asking. If the answer is "the tech forgot," that is rarely the real cause; ask why the process let a forgettable step be forgettable. The fix belongs in the written way of working so the next person inherits it automatically.

Step four: re-standardize and make it the new floor

An improvement that lives in one conversation evaporates. Update the written standard, tell everyone who runs it, and confirm they actually switched. Now the improved version is the new baseline, and the loop starts again from a higher floor. This is what "getting better" actually looks like in a shop: not a heroic overhaul, but a floor that quietly rises a step at a time and never slides back.

Why the order cannot flip

Owners often want to improve first and standardize later. It does not work, and here is the mechanism. If the process is not fixed, you cannot tell whether the thing you changed caused the result or whether it was just a different tech, a different day, a different customer. You have no control. Standardize first and a single change stands out cleanly against the baseline, so you actually learn something you can keep.

The trap: standardize once, then walk away

The opposite failure is treating the written standard as sacred. A standard that never changes is a standard slowly going stale, because the work, the tools, and the customers keep moving. The written way of working should show edits. If your procedures have not changed in a long time, either you have stopped learning or you have stopped writing down what you learned. Both are the same problem wearing different clothes.

The mental model to keep

Standardize to create a floor. Improve to raise it. Re-standardize to keep it from sliding back. A shop that only standardizes gets rigid; a shop that only improvises never compounds. The gain comes from the loop, not from either half alone.

References

• See related: Standardize This Process or Leave Room to Adapt (decision tree)
• See related: Why Fixing One Thing at a Time Beats a Big Overhaul
• U.S. Small Business Administration (SBA), operations and process-improvement guidance
• Trade-standard practice for documented work procedures