Imagine spending half a year automating a process, only to accelerate something that never worked in the first place. The most expensive mistake in digitalisation isn't the wrong technology. It's the wrong process. Before you optimise, one fundamental question needs answering: improve, reinvent, or scrap? And you answer it not by instinct, but with the right metrics.
Eliminate first, then automate
Behind every good optimisation sits a simple sequence that is surprisingly often ignored: ESIA, short for Eliminate, Simplify, Integrate, Automate. Eliminate first, then simplify, then integrate, and only at the very end automate.
The cardinal rule lives in that order: never automate a process you could actually scrap. The most expensive mistake is automating a bad process. You only make waste faster, more stable, and more expensive. A manual detour that costs time ten times a day doesn't get better through automation. It just gets set in stone.
That's why every serious optimisation starts with an uncomfortable question: is this step needed at all? What happens if we switch it off tomorrow? Only once it's clear that a process should stay and holds up in its basic structure does the question of the right tool become worthwhile.
Scrap, reinvent, or improve — the three paths
Not every process deserves the same treatment. Three paths are open, and the art is choosing the right one before money flows.
Improve when the purpose still makes sense and the process basically works. The bottlenecks are measurable, the data is there, and improvement is possible at manageable risk. Example: a quote approval takes too long but is genuinely necessary. High value, good structure, high manual share, this is where automation pays off.
Reinvent when the process has grown historically and is only kept alive through several systems and manual detours. The goal would be reachable in a completely different way with today's means, and automation would only treat symptoms. Example: scheduling runs on Excel, gut feeling, and ERP exports. The goal isn't to "make Excel faster," but a new digital decision process. High value, but structurally broken: rethink end to end instead of patching.
Scrap when the process no longer creates value and only exists because a legacy system forces it. Controls are duplicated, nobody uses the output. Example: a report is produced monthly, but no decision depends on it. No value contribution, so out it goes.
The most honest diagnostic question for the third path is also the most uncomfortable: what happens if we switch this process off tomorrow? If the answer is "nothing bad," you've just found money.
| Starting point | Recommendation |
|---|---|
| Purpose sound, structure holds, high manual share | Improve |
| High value, but grown structurally across systems and detours | Reinvent |
| No measurable value contribution, nobody uses the output | Scrap |
Which KPIs carry the decision
Which of the three paths is right isn't revealed by gut feeling, but by a handful of metrics. They turn an opinion debate into a decision.
- Value contribution × frequency: The biggest lever rarely sits with the most annoying process, but with the most frequent. Cost arises from frequency times time times labour cost times error rate. The small daily friction almost always beats the tiresome quarterly process.
- Error and rework rate: How often does a case need correcting? A high rate points to structural weaknesses, a candidate for reinvention, not mere acceleration.
- Manual share and media breaks: Wherever people type data from one system into the next, a measurable automation lever sits. The higher the manual share on a stable base structure, the clearer the case for improving.
- Cycle time: It reveals bottlenecks and customer impact. Long waits between steps are often more expensive than the processing itself.
The order is decisive: measure first, then decide. Anyone who can't quantify a process's value contribution shouldn't automate it, but understand it first.
How we find the right metrics together
The good news: most of these metrics already exist in your systems. How often a case is corrected, how long it sits, at which step the most queries arise, all of that is in ticket systems, ERP logs, and email threads. It's just rarely read out systematically.
That's exactly where we start. Before we talk technology, we clarify with our clients which process is truly the most expensive and which metric points the way. We help identify the right KPIs, set up an honest baseline measurement, and derive the fundamental decision from it: improve, reinvent, or scrap.
That is the basis of any investment that can later be proven. Because a decision without a metric is a bet. A decision with a metric is a plan.
The next part of this series is about exactly that: how this baseline measurement becomes proof, and how you'll recognise in six months whether the optimisation truly paid off.