Step 7 of 8
REFY · then refine
Fit — exactly one lever, cleanly wired
F is the step many projects mistake for the whole thing. It is the seventh of eight, and everything before it is the reason it can stay small: the bottleneck is named, the ballast is gone, ownership is settled. What remains is a change in one place.
Limiting it to one lever is not modesty, it is a precondition for Y. Two simultaneous changes cannot be separated in the re-measurement; afterwards nobody knows which one worked, and the next cycle starts from a guess.
- Answers the question
- Which single technical lever gets built, and how is it wired in?
- What the step produces
- One implemented change at one point, connected to the systems already running, with documented behaviour on failure.
How the step is carried out
Take the lever from W, not the obvious one
W made a decision, against cost and with a payback period. Building something else here because it presented itself along the way bypasses that decision and invalidates the calculation.
Connect rather than replace
The spreadsheet stays the spreadsheet, the ERP stays the ERP. A system replacement is a project of its own, with its own business case, its own risk and a duration that breaks this cycle. It belongs as a candidate in a new F, not in this one.
Act at the break point
The lever goes where O named the bottleneck — typically a media break, a waiting loop or a manual transfer. An intervention anywhere else accelerates a step that was not the constraint and leaves lead time unchanged.
Build the failure path
What happens when the connection is up but the far end does not answer? An intervention without a fallback moves the bottleneck into operations: the process is faster in the normal case and blocked in the exceptional one, and the exceptional case turns up more often than the plan allowed for.
Exactly one lever
If a second idea appears during implementation, it gets written down and assessed in the next cycle. Not built.
Finished when
- The lever runs in daily operation, not only in a test.
- There is a documented path for what happens when the connection fails, and somebody who knows it.
- No existing system had to be replaced.
- It is exactly one change. If it is two, the re-measurement in Y can no longer be attributed.
- The old route is either switched off or carries a date for switching off — two routes running in parallel is not a changeover, it is duplicated work.
The typical mistake
Three levers at once
It is tempting while you are already inside the process, and it feels efficient. The price only lands in Y: the process is faster and nobody can say from what. The next cycle then reopens the same debate as the first, with more opinions and a larger stack of tools for somebody to maintain.
What you can compute yourself
The automation check settles before construction whether the lever will hold technically: data quality, exceptions, interfaces, volume. It does not replace the implementation, but it prevents implementing at a point that has no interface.
The step, worked through
The intervention in each sample analysis is deliberately kept small and named — one change, not a bundle. That is what makes it possible to see which part of the improvement comes from where.
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
15 Workflow Automation Examples That Save Hours
Concrete workflow automation examples from various industries. With step-by-step instructions and tool recommendations.
n8n vs Make: The Ultimate Comparison [Pros & Cons]
n8n or Make (Integromat)? We compare both workflow automation tools: features, pricing, data privacy & use cases. Find the right tool for you.
Questions
Because a replacement is a project of its own: its own business case, its own risk, a duration that breaks the cycle, and a data migration that has never once matched its estimate. This step is about the single bottleneck O named. If the replacement is worth it, it enters a new F as a candidate and gets costed against the others there.
Then it is not a lever but a development project, and the calculation from W no longer holds. The right move is back to W with the real costs — not forward on the hope that it will work out.
No. A lever can be a changed sequence, a form with mandatory fields, or an automatic notification replacing a chase. What matters is that it acts at the point O named and that its effect is measurable in Y.
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