The wrong reason to buy an ERP
"We are growing and should be professional" is not a reason. It is an aspiration that ERP vendors are exceptionally good at monetising.
The right reasons are specific operational failures that cost measurable money and cannot be solved with better process alone.
The triggers that justify it
- Inventory you cannot trust. Physical counts diverge from records often enough that production plans are built on guesses.
- Departments with conflicting truths. Sales promises a date, production has different capacity data, and accounts have a third version of what shipped.
- Order status requires phone calls. Nobody can answer where an order is without asking three people.
- Planning lives in one person's head. When that person takes leave, scheduling degrades.
- Month-end close takes weeks because data must be reconciled manually across systems.
- Compliance and traceability requirements you cannot satisfy from current records.
Three or more of these, sustained, is a real trigger.
What an ERP will not fix
- It will not create data discipline. If stock movements go unrecorded today, they will go unrecorded in the ERP, and the reports will be wrong with more authority.
- It will not resolve disagreements about process. Software forces a decision; it does not make it for you.
- It will not compensate for unclear ownership. Every master data set needs a named owner.
- It will not be cheap. Licence cost is typically the smaller half of total cost.
Phasing: the single most important decision
Big-bang implementations fail disproportionately. Phase by operational pain instead.
A common effective order:
- Inventory and procurement. Accurate stock is the foundation everything else depends on.
- Production planning and shop floor. Once stock is trustworthy, scheduling can be systematised.
- Sales and order management. Connect demand to the now-reliable supply picture.
- Finance integration. Automate what was being reconciled manually.
- Quality, maintenance, and analytics. The refinements that only make sense on clean data.
Each phase should deliver a visible operational win before the next begins. If phase one does not improve daily life, stop and fix the approach rather than continuing.
Data preparation is the real project
Before any module goes live:
- Clean the item master. Remove duplicates, standardise naming and units.
- Establish bills of material accurately. Wrong BOMs produce wrong plans forever.
- Agree units of measure and conversion rules.
- Assign a named owner to each master data set.
- Take an honest physical stock count as an opening balance.
Most implementations that fail were failing here, months before the software was blamed.
Change management is not optional
Budget for it explicitly. Practical elements:
- Involve shop floor and store staff in design, not just management.
- Train on the actual daily tasks people perform, not on the software's feature set.
- Expect a temporary productivity dip and plan around it rather than being surprised.
- Identify respected internal champions in each department.
- Make it clear the system replaces the old method rather than adding to it. Parallel running beyond a short period guarantees the new system loses.
Build versus buy versus configure
Standard ERP products cover most manufacturing needs. Custom development is justified when a genuinely differentiating process exists that no product supports — and that is rarer than most owners believe.
A pragmatic middle path: use a standard product for the core, and build small custom applications around it for the two or three processes that are genuinely unusual, connected through the ERP's integration layer.
Never customise the core so heavily that upgrades become impossible. That decision looks efficient in year one and traps you by year four.
Signals the project is going wrong
- The scope has grown twice and the first go-live date has moved twice.
- Users describe the system as "the consultant's project".
- Data cleaning is being deferred to after go-live.
- Nobody internally can explain how a core process will work in the new system.
- Training is scheduled for the week of launch.
Any two of these warrant pausing rather than pushing through.
The realistic outcome
A well-phased ERP gives you inventory you can trust, plans that survive contact with reality, order visibility without phone calls, and a month-end close that takes days instead of weeks.
That is a genuine transformation of how a manufacturing business is run. It is also months of unglamorous data and process work, and no software purchase substitutes for it.
