Step 8 of 8
REFY · then refine
Yield — prove the return
Y is the step that makes the method checkable. Without it, everything before it is a well-argued opinion. The most important rule of this step is therefore a restriction: measure with the same arithmetic used in L and O, not with a better one.
Re-measurement also covers what did not happen. A report showing only the improvement is not a measurement but a presentation — and it deprives the next cycle of its basis.
- Answers the question
- What did the intervention achieve — and where is the bottleneck now?
- What the step produces
- A before-and-after in euros and lead time from the same arithmetic as L and O, the deviation from the estimate in W, and the newly named bottleneck.
How the step is carried out
The same measurement, not a new one
Same step sequence, same loaded rate, same volume basis, same number of simulation runs. Improve the method in between and you are measuring the improvement of the method rather than the process.
Wait until the new state is routine
Measure immediately after the changeover and you measure the changeover: questions, adjustment, double work. A process needs a few dozen cases to show its normal behaviour. Measuring earlier produces a figure of no use to anyone that has to be explained away later.
Report two quantities, not one
Annual cost and lead time as a distribution. A process can become cheaper and less reliable at the same time — mean lead time falls while P90 rises. A single figure hides exactly that case.
Name the new bottleneck
After every effective intervention the bottleneck sits somewhere else; the second-slowest step is now the slowest. That belongs in the result even when it is unwelcome — it is the entry point to the next cycle.
Hold the difference against W
What was estimated in W against what came out. The deviation is the most valuable information in the whole cycle: it says whether your own estimating holds up, and it corrects the next W.
Finished when
- Before and after come from the same arithmetic.
- The difference is stated as a range, not a point value.
- Cost and lead time are reported separately.
- The new bottleneck is named.
- The deviation from the estimate in W is explained rather than rounded away.
The typical mistake
Reporting only the success
A re-measurement that shows nothing but the improvement is not a measurement. It has to include what got worse, what stayed the same and where the bottleneck moved to. Leave that out and you have bought a pleasant closing report and will pay for it in the next cycle, which then starts from an assumption instead of a finding — and the assumption will be the one that happened to look good.
What you can compute yourself
The process cost calculator is used a second time here, with the same inputs except the changed durations. That is the whole trick of re-measurement: it is the same calculation, not a better one.
FlowVisual is the instrument this method uses where a questionnaire stops: draw the process, run it hundreds of times, read which step blocks. It is a separate product on its own domain — Flowrefy owns the method, FlowVisual owns the instrument.
The step, worked through
The sample analyses set before and after side by side, together with a “limits” section stating what the measurement does not show. The two belong together: the second column without the limits would be an advertisement.
The analyses are constructed models, not client projects. Their figures show what a measurement would look like — they are not results achieved for anyone.
- Sample analysisAN-2026-08
The sales lead who shows up as two boxes
A packaging printer measures the path from incoming order to order confirmation. The bottleneck is the credit check. The second finding is a role two steps share, and it only becomes an action once the first one is dealt with.
- Sample analysisAN-2026-07
The loudest process in the building is not the most expensive
An engineering consultancy wants to digitalise leave requests. The measurement agrees with the complaint and still not with the investment. The process is slow, but cheap.
- Sample analysisAN-2026-06
The €500 approval limit nobody has touched since
A plastics processor has every purchase above €500 signed by the managing director. The limit is old, the prices are not. Four out of five orders pass through a gate that opens twice a week.
Further reading
Your spreadsheet does the math right and is still wrong
Six steps, three days each on average, eighteen days lead time. That sum is correct. And it is exactly why you hit the date less than half the time. Why the sum of the averages is not the lead time. Simulated, not asserted.
15 Process Automation Examples That Save Time Immediately
These 15 process automation examples show you exactly how to reduce manual work. From simple to complex - with concrete implementation tips.
Questions
Long enough for the process to be routine again — usually a few dozen cases or one to two months, whichever comes first. For low-volume processes that takes longer; then you measure later rather than less precisely.
Then that goes in the result, along with a hypothesis about why: the wrong bottleneck, too small a lever, or an effect absorbed by another step. A cycle that shows the assumption in W was wrong is uncomfortable — and cheaper than the next intervention built on the same wrong assumption.
Because the field has changed. The process that was the most expensive may no longer be after an effective intervention, and a different candidate from the list now sits at the top. Building the next lever in the same process would skip the selection L exists to make, and put you back on instinct.
If you want to know what your process costs: measure it.
Start with the free diagnosis or download FlowVisual. If you want to speak to someone afterwards, we're reachable.
Remote · fixed price · result in euros