The top 5 mistakes new manufacturers make — and what Japan figured out decades ago
Japan rebuilt itself into a manufacturing superpower with almost no raw materials, expensive energy and cramped floor space — constraints many South African manufacturers would recognise. The methods that came out of that era (most famously the Toyota Production System) aren't academic theory; they're the distilled lessons of decades of factory mistakes. After thirty years of building software for manufacturers, we see new operations make the same five mistakes those methods were invented to prevent.
1. Producing for the shelf, not the order
The mistake: a new factory buys machines, and the machines must "earn their keep" — so they run, whether there's demand or not. Cash turns into stock, stock fills the floor, and six months later the business is profitable on paper and broke in the bank.
The Japanese lesson: Toyota called overproduction the worst of all wastes (muda), because it hides every other problem and buries cash. Just-in-time production means making what's ordered, when it's ordered — utilisation is not the goal, flow is.
Avoid it: measure success by orders shipped and margin per job, never by "the machines were busy". Know your work-in-progress number every week — if it only ever grows, you're funding a warehouse, not a business.
2. Running the factory from the office
The mistake: the owner prices jobs, chases sales and reads reports — and hasn't stood on the floor in weeks. Problems get filtered through three people before they reach the person who can fix them.
The Japanese lesson: genchi genbutsu — "go and see for yourself." Toyota managers are famously expected to stand on the floor and watch the actual process before making any decision about it. Reports summarise; the floor tells the truth.
Avoid it: walk the floor daily, and when something goes wrong, go to where it happened and watch the process before deciding anything. The habit costs twenty minutes a day and prevents the expensive fantasy version of your own factory that exists only in the office.
3. Inspecting quality in at the end
The mistake: quality control = a final inspection before dispatch. By the time a defect is caught, you've paid for the material, the labour and every process step after the one that caused it — and the rework queue quietly eats the month's margin.
The Japanese lesson: jidoka — build quality into the process, and stop the line the moment something is wrong. It feels expensive to stop; it's far cheaper than producing defects at full speed. Quality is checked at each step by the person doing the work, not by a gate at the end.
Avoid it: give every step a simple "is this right?" check before it passes work on, and make it culturally safe — celebrated, even — for a worker to stop the job and flag a problem early. One stopped job beats fifty quiet reworks.
4. Depending on heroes instead of standards
The mistake: the factory runs because Willem knows the settings, remembers the customer's quirks and can coax the old machine. Then Willem is on leave, and output halves. Nothing is written down; every job is made slightly differently.
The Japanese lesson: standardised work — the current best-known way to do each task, documented, visible, and followed by everyone — plus 5S, the discipline of an organised workplace. Standards aren't bureaucracy; they're the baseline that makes problems visible and improvement possible. You can't improve a process that's done differently every time.
Avoid it: write down how each job is actually done — settings, times, materials, checks — starting with your highest-volume work. If it lives only in someone's head, you don't own the process; they do.
5. Improving in leaps instead of steps — and without data
The mistake: nothing changes for a year, then a crisis triggers a big, expensive fix — a new machine, a consultant, a reorganisation — chosen on gut feel, because nobody can say what anything actually costs to make.
The Japanese lesson: kaizen — continuous improvement in small, cheap, permanent steps, made by the people doing the work, based on measured facts. A hundred one-percent improvements beat one heroic project, and they don't bet the company.
Avoid it: capture the basics on every job — hours, materials, rework — and review quoted-versus-actual monthly. Small honest numbers, looked at often, tell you exactly where the next one-percent improvement is. This is where good job card and manufacturing software earns its keep: not as admin, but as the measurement layer kaizen depends on.
The common thread
Every one of these mistakes is really the same mistake: running on optimism instead of visible facts. The Japanese methods all force reality to the surface — on the floor, at each step, in small measured doses — while it's still cheap to act on.
Starting or growing a manufacturing operation and want the measurement side done properly? Tell us how your floor runs — we've been building exactly that for South African manufacturers for 30 years.