The invoice they didn't see coming
A garment factory — 180 staff, three production lines — went live on Odoo in October. The implementation vendor charged BDT 8 lakh. By March, the finance director was telling me the ERP had cost them closer to BDT 28 lakh.
The vendor's BDT 8 lakh invoice was real. The remaining BDT 20 lakh was invisible — it was never billed, never budgeted, and never appeared in any line item. It appeared instead as: two months of parallel operations (the old Tally system and Odoo running simultaneously, both needing maintenance and reconciliation), four weeks of overtime for the accounts team to reconcile diverging inventory values, a three-week production halt when the BOM configuration produced incorrect material requirements, and the cost of a second consultant brought in to fix what the first one had missed.
I tell this story not to alarm, but to reframe the question. The right question is not "how much does ERP cost?" — it is "how much does a poorly-scoped ERP cost, and how do I avoid paying that?"
The vendor invoice is the down payment. The real cost comes in the quarters after go-live.
Three go-live failure patterns
Across five manufacturing deployments, the failures that generated unexpected cost clustered into three patterns:
Data migration failure
Opening balances are wrong. Item master has duplicate SKUs. BOM structures weren't reviewed before import. On day one of live operations, stock valuations don't reconcile — and the accounts team runs the old system in parallel while the issue is diagnosed and corrected. This takes weeks, not hours.
Training gap go-live
Users were trained once, during configuration. By go-live, the system has changed. The training was screen-based, not task-based. No floor-level operator can complete their actual Tuesday job in the system without asking for help. Productivity drops 40–60% for the first month.
Scope creep discovery
Three weeks after go-live, a production supervisor discovers that the system doesn't handle the split-lot tracking they use every morning. It wasn't scoped. The vendor says it's a customisation. There's no budget for it. Operations reverts to manual tracking while the commercial discussion happens.
The parallel-operation trap
Every business runs parallel operations after a go-live. That is normal and prudent — you want a fallback if something is catastrophically wrong in the first week. The trap is when "parallel" becomes the permanent state.
Parallel operations in manufacturing typically cost:
- The salary time of the person maintaining the old system (often 30–50% of one accountant's working week)
- Month-end reconciliation between the two systems — typically 3–5 days of accountant time per month
- The cognitive overhead of management running on two data sources that say different things
- Delayed decision-making because nobody knows which system to believe
In a 150-person factory, two months of parallel operations costs roughly BDT 3–6 lakh in productive time before the accounting starts. I have seen parallel operations persist for six months. At that point, the business has quietly paid for the ERP twice.
How to budget for the real cost
A realistic budget for a mid-size Bangladesh manufacturing ERP deployment looks like this:
| Cost item | What most clients budget | What it actually costs |
|---|---|---|
| Software license (Odoo Enterprise, 20 users) | BDT 4–6L/year | BDT 4–6L/year |
| Vendor implementation fee | BDT 5–15L | BDT 5–15L |
| Data migration (audit + clean + import) | BDT 0 (included in above) | BDT 1–4L (dedicated resource + time) |
| Change management & training | BDT 0 (one training day included) | BDT 1–3L (floor walks, refreshers, champions) |
| Parallel operations (if go-live fails) | BDT 0 (not anticipated) | BDT 3–10L (2–4 months of dual-system operation) |
| Rework & fix engagements | BDT 0 (not anticipated) | BDT 2–8L (common in under-scoped projects) |
| 90-day hypercare support | BDT 0 (assumed included) | BDT 1–3L (explicit post-go-live commitment) |
| Total (well-scoped project) | — | BDT 12–31L |
A well-scoped ERP project costs more upfront than a poorly-scoped one. It also costs significantly less in total. The difference is all the work that gets priced into the contract instead of absorbed by your operations team over six months.
What a clean go-live looks like
In five go-lives, two were clean. Both had identical characteristics: a dedicated internal champion who knew the business and the system, a data migration that finished two weeks before go-live (not two days), a UAT period that lasted three weeks with real users running real Monday-to-Friday operations, and an explicit 90-day hypercare clause in the vendor contract.
The hypercare clause is the one I tell every client to fight for. It changes the vendor's incentive structure. When a vendor knows they are on-site for 90 days after go-live at no extra charge, they configure more carefully during implementation. Every shortcut they take in discovery becomes a cost they pay later.
The change management side of a clean go-live — the floor walks, the named champions, the repeated training — is covered in detail in Change management isn't a memo. The root causes of why go-lives fail structurally are in Why ERP implementations fail in Bangladesh. Read both before signing a contract.
The question is not "can we afford a good implementation?" The question is "can we afford to run a bad one twice?"
FAQ
What does a failed ERP go-live actually cost a Bangladesh manufacturer?
A failed or delayed go-live typically costs 2–4× the original project budget once you factor in parallel operations, overtime for data rework, consultant extensions, and the productivity dip during adoption. On a BDT 15 lakh project, hidden go-live failures can add BDT 20–40 lakh in total cost — most of it never appearing on an invoice.
What are the hidden costs of an ERP go-live in manufacturing?
The most underestimated lines are: parallel operation staffing (running Tally and Odoo simultaneously), data rework after a bad migration, user overtime during adoption, consultant extensions, and senior management time diverted to firefighting. None of these usually appear in the initial vendor quote — see the budget table above for realistic ranges.
How can Bangladesh manufacturers reduce ERP implementation risk?
Three practices consistently reduce go-live risk: thorough discovery and process mapping before configuration, structured UAT with actual end users, and a phased go-live by department rather than a big-bang cutover. Budget 20–30% contingency on top of the quoted implementation cost.
Don't want to read the full breakdown? The free Odoo Cost Estimator gives you a Bangladesh-specific cost range — including the hidden lines covered above — in 60 seconds.
If you're in the scoping phase for a manufacturing ERP, I offer a structured discovery engagement that produces a scoped cost estimate before any implementation starts. Let's talk →