How to Choose an ERP for a Garments Factory

Choosing an ERP for a garments factory takes seven steps: map your current process, write the scope down, shortlist on process fit, run a scripted demo, check references of your own process type, compare year-three cost, and agree the migration method. Skipping step two is why most selections end in a change request.

What are the seven steps, and how long do they take?

Six to ten weeks from first internal meeting to signature, for a single-unit factory. The sequence matters more than the pace. Every step exists to move a decision out of a sales conversation, which is where ERP selections reliably go wrong.

Note who owns step 6. If the finance director first sees the numbers after a vendor has been chosen, you have run the process backwards.

Step 1: Map what you already do in Excel

Start with the spreadsheets, not with the requirement. List every workbook that something important depends on: the costing sheet, the consumption calculation, the order tracker, the bond reconciliation, the attendance file. That list is your actual requirement document.

Most factories find between eight and twenty of these. Each is a process the business runs on that nobody owns formally. Note who maintains each, what breaks if they are on leave, and whether anyone else can read the formulas. That last question is usually the uncomfortable one.

Do this before any vendor conversation. A requirement list assembled during a demo is really a list of features you were shown.

Step 2: Write the scope before you talk to anyone

Turn the spreadsheet list into a module list and a one-page written requirement. Name the modules you need in year one, the process types you run, the number of units to consolidate, and the statutory outputs you must produce. Then send the same document to every vendor.

This is the step buyers skip and the one that pays. Quotes against one written scope are comparable. Quotes against a conversation are not, and the variance between them is mostly scope rather than price.

Be explicit about the statutory items, because they are rarely in a standard proposal: the bonded warehouse register, Utilisation Declaration reconciliation, the Mushok VAT forms, and master LC to back-to-back LC on one order record.

Step 3: Shortlist on process fit, not on feature count

Three vendors, chosen on whether they have implemented your process type at your scale. Not eight, because you will not run a real evaluation eight times, and not one, because a single vendor leaves you no negotiating room.

Process fit means something specific in garments. Knit is line-balanced, sweater is programme-driven and piece-rate, woven adds subcontract fabric that stays on your bond, and a buying house coordinates production it does not control. A vendor strong in one is not automatically competent in another.

Our garments ERP selection guide by factory size and process type sets out what to require for each, and the vendor diligence question cover how to assess the company rather than the product.

Step 4: Run a scripted demo

Send your own script in advance and make every vendor follow it. A standard demo is rehearsed on clean data and shows you the software on a good day. You need to see a bad day.

Six things to script: create a live style with two buyers and different size ratios, change a colour after cutting has started, show a partial shipment and the bond position after it, trace one named worker from attendance through to payslip, post a store issue against an order with no fabric received, and ask who in the room will be at your factory in month three.

Score each vendor in the room. Memory of a demo is unreliable within a week and you will run three of these.

Step 5: Check two references of your own process type

Ask for two clients of your process type and roughly your size, then ask them four questions the vendor cannot coach: what slipped, what did you argue about, what did it cost beyond the quote, and would you buy again.

A reference of a different process type answers a different question. A knit reference tells you nothing about whether the sweater module works. If a vendor can only supply references outside your process, that is itself the finding.

Ask about elapsed time against the original plan. Almost every implementation runs long. What matters is by how much, why, and who absorbed it.

Step 6: Compare year-three cost, not licence

Ask each vendor for one itemised number: total cost at the end of year three, covering licence, implementation, migration, training, annual support, and a named customisation allowance for the changes you have not thought of yet.

Licence is typically 25 to 40 percent of a three-year total, so a vendor 30 percent cheaper on licence is offering roughly 8 to 12 percent off the project, assuming equal scope. Our ERP cost guide for Bangladesh works the full model.

Add a line no vendor will quote: your own people's hours at loaded cost. It is usually comparable to the migration line.

Step 7: Agree the migration method and the cut-off date

Before signature, get the migration method in writing: extraction, cleansing, who cleans what, a parallel run period, and a dated cut-off after which the old system is read-only. Migration is where implementations slip, and an undefined method is the largest schedule risk you can still remove for free.

Pin down the written definition of configuration versus customisation at the same time. You will be arguing about that boundary in month five, and the time to define it is while you still have a choice of vendor.

For how order, cutting, work in progress, subcontract and bond sit against a single order record, see our garments ERP software page.

Frequently asked questions

Which ERP software is best for a garments factory? None universally. Knit, woven, sweater and buying house operations need materially different systems, so the right answer depends on your process type first and your size second. Build a requirement list from your own spreadsheets, then evaluate vendors against it rather than against each other.

How long does it take to choose an ERP? Six to ten weeks from first internal meeting to signature for a single-unit factory, assuming someone owns the project. Multi-unit groups take longer because scope has to be agreed across units. The selection is not the long part; implementation is three to twelve months after that.

Who should lead the ERP selection? A champion with decision authority, not just availability. The most expensive selections are the ones where the assigned person has to refer every question to a director's meeting. Involve the finance director at step six at the latest, and ideally from step two.

Should we shortlist one vendor or several? Three. One leaves you no negotiating room and nothing to compare answers against. More than three means you will not run a real evaluation on any of them. Three is enough to see genuine differences in how the same requirement gets interpreted.

What is the most common mistake in ERP selection? Not writing the scope down before talking to vendors. Without one written requirement sent to everyone, quotes are not comparable, the gap reappears as a change request after signature, and it is priced without competitive pressure because you have already committed.

Sources

• Customs Bond Commissionerate, Dhaka, on bonded warehouse licensing and register requirements. http://www.dhakabond.gov.bd
• National Board of Revenue, Value Added Tax and Supplementary Duty Act 2012 and rules made under it, on prescribed Mushok forms. https://nbr.gov.bd
• Bangladesh Garment Manufacturers and Exporters Association, on Utilisation Declaration issuance. https://www.bgmea.com.bd
• Export Promotion Bureau, Bangladesh, apparel export statistics. http://epb.gov.bd
• BDStall, published ERP price listings for the Bangladesh market. https://www.bdstall.com