Manufacturing ERP connects the parts of a factory that generic business software leaves separate: the bill of materials, work orders on the floor, raw material stock, finished goods, costing and the accounts. Invento implements ERPNext and Odoo for manufacturers in Bangladesh, from selection through data migration and go-live to support afterwards. This page sets out what a manufacturing implementation actually involves, including the parts most often underestimated.
What manufacturing needs that generic ERP does not cover
Most ERP platforms handle accounting, purchasing and stock out of the box. Production is where they diverge, and where an implementation succeeds or stalls. A manufacturer needs five things a trading business does not.
- Bill of materials, with versions. Not a single parts list, but a structure that survives engineering changes, sub-assemblies and substitutions without rewriting history.
- Routing and work orders. Which operation happens at which workstation, in what order, taking how long. Without routing you have a stock system, not a production system.
- Batch and serial traceability. Given a finished unit, which raw material lot went into it, and given a raw material lot, which finished units carry it. This is a recall requirement, not a reporting nicety.
- Capacity and material planning. Knowing whether a confirmed order can actually be produced by its promised date, given what is on the floor and what is on order.
- Real production costing. Material plus labour plus overhead plus scrap, per work order rather than per month, so margin is known per product instead of guessed at in aggregate.
If a platform demo skips these and spends its time on invoicing, you are being shown the wrong system.
The Bangladesh-specific parts
Four requirements come up on almost every manufacturing implementation here, and none of them ships as standard on any international ERP platform. They are localisation work, and they are the most commonly underestimated line in a Bangladeshi ERP budget.
| Requirement | What it means in practice |
|---|---|
| Mushak 6.3 VAT invoicing | The National Board of Revenue requires businesses to issue a Mushak-6.3 VAT invoice on sales and obtain one on purchases (NBR VAT compliance guide). No standard ERP invoice matches that form, so it is built. |
| VAT registers and returns | The wider Mushak series covering purchase and sales registers and return preparation, which has to reconcile to the same ledger the accounts use. |
| Bonded warehouse handling | For export-oriented manufacturers, duty-free imported inputs have to be tracked against entitlement and consumption separately from ordinary stock. |
| Bangla interface and multi-currency | Bangla for floor and store users who will not adopt an English-only system, and multi-currency for exporters invoicing in USD or EUR while reporting in BDT. |
If you make garments, the compliance layer is different in kind
Ready-made garments account for the majority of Bangladesh’s manufacturing exports, and an export garment factory carries obligations that no international ERP platform has ever heard of. These are not harder versions of ordinary requirements. They are a separate layer, and they decide whether a system is usable on the floor or abandoned within a year.
- Utilisation Declaration (UD). BGMEA and BKMEA issue UDs on the National Board of Revenue’s behalf — BKMEA under NBR notification SRO 3(36)-CUS-4/87(Part-3)-373, in force since 1998 (BKMEA). The UD is what permits duty-free import of inputs under bond, and it is tied to export processing, GSP eligibility and cash incentive claims. An ERP has to track actual consumption against the declared entitlement, per style and per order — not as a report produced afterwards, but as the stock moves.
- Back-to-back L/C reconciliation. Inputs are financed against the master export L/C. The system has to hold the import L/C, the UD and the export L/C against the same order, because that is the chain customs follows.
- Wastage against the declared percentage. Actual consumption versus declared wastage is precisely the gap that gets queried. If the system cannot show it per order, somebody is rebuilding it in Excel at audit time.
- A bond register separate from ordinary stock. Duty-free material under bond is not the same inventory as duty-paid material, and treating them as one pool is the most common way a garments implementation fails its first audit.
- Costing by style and order, not by SKU. A garment factory prices a style for a buyer against a specific order quantity. Costing built around a stock-keeping unit answers a question nobody asked.
- Industrial engineering data. SMV and SAM, line efficiency, target versus achieved by hour. This is how a garment factory actually measures production, and it is not a standard module on any general ERP platform.
What that means for platform choice, said plainly
Invento implements ERPNext and Odoo. Both are general manufacturing platforms, and none of the six requirements above ships as standard on either. They are built. That is normal and it is entirely doable — but it is scope, it costs money, and you should hear it before you sign rather than in month three.
It also means a dedicated garments product may genuinely suit you better. If your factory is a straightforward CMT operation running a conventional UD and bond process, and you want it working in weeks rather than months, a purpose-built apparel system will get you there faster. A general platform earns its place when you need the ERP to cover more than the factory — group companies, trading arms, retail, non-garment lines — or when your process differs enough from the standard that a fixed product would have to be bent anyway.
We would rather tell you that at the first meeting than win the project and discover it at user acceptance testing.
How a manufacturing implementation runs
The sequence below is the one we work to. It is deliberately front-loaded: the expensive failures in ERP happen because process and data questions were left until the system was already configured.
- Process discovery. Walking the floor and the office and writing down how things are actually done, not how the org chart says they are. Undocumented exceptions are where implementations break.
- Data assessment. Item master, bills of material, supplier and customer records, opening stock. Migrating an inaccurate stock list produces an inaccurate system, so this is assessed before anything is configured.
- Configuration and localisation. Chart of accounts, warehouses, routings, costing method, and the VAT work described above.
- Conference room pilot. Running a real order end to end — purchase through production through despatch through invoice — with your own data, before anyone is trained.
- User acceptance testing and training. Floor users and store users separately from finance, because they need different things and learn at different speeds.
- Cutover and go-live. Stock count, opening balances, and a defined fallback if something goes wrong on day one.
- Support and iteration. The first month after go-live produces more change requests than the whole build. Plan for it rather than treating it as failure.
Questions worth answering before you start
- How many people will log in? Count floor and store users, not just office staff. This single number changes which platforms are affordable, because some price per user and some do not.
- Do you make to stock, to order, or both? Both is normal and it is also the configuration most often got wrong, because the two need different planning behaviour.
- How accurate is your current stock figure? If the warehouse and the accounts disagree today, they will disagree in the new system until the underlying count is fixed.
- What has to integrate? Weighbridge, barcode scanners, payment gateways, existing accounting. List them before you price anything, because API access is gated behind higher tiers on some platforms.
- Who owns this internally? An implementation without a decision-maker on the client side stalls at the first process question. This is the strongest single predictor of whether a project lands.
Common questions
How long does a manufacturing ERP implementation take?
It depends far more on your process complexity and data quality than on the platform. A single-site manufacturer with clean data and standard processes is a different project from a multi-site operation with custom workflows and stock that has never been fully counted. Ask any partner to size it against your own process list rather than quoting a generic timeline.
Does ERP handle Bangladesh VAT out of the box?
No. Mushak 6.3 invoicing and the related VAT registers require localisation on every ERP platform sold here. Ask to see it working on a live system rather than described in a proposal.
Which platform should a manufacturer choose?
That depends on user count, integration needs and where the data has to live, and it is worth working through properly rather than taking a recommendation. Invento implements both ERPNext and Odoo, so we have no reason to steer you toward either.
Can ERP work alongside systems we already run?
Usually, through integration rather than replacement. Check API terms before choosing a platform, because some vendors restrict API access to higher subscription tiers, which changes the cost comparison.
Talk to us about your factory
The useful first conversation is not a product demo. It is twenty minutes on what you make, how many people touch the system, and where things currently go wrong — enough to tell you honestly whether ERP is the right answer and roughly what it would involve. Book a free consultation, or see our ERP implementation and support page for the wider service.
Background reading: why manufacturers move to ERP covers the operational problems that drive the decision, the platforms manufacturers most often shortlist covers selection, and ERP for food manufacturing covers traceability and shelf-life requirements specific to food and beverage production.