How to choose an ERP for your factory

Step 1: Map your process before you look at software

The most reliable requirement document for an ERP is not a feature list. It is a walk through your plant with a notebook. For each department, write down every register, document and approval that exists today, who keeps it and who reads it. A typical mid-sized plant ends up with thirty to sixty items.

This map does three things. It tells vendors what you actually need; it shows which pains matter most; and it becomes the checklist for the demo and, later, for the implementation.

Step 2: Separate must-haves from nice-to-haves

From the map, mark each item as a must-have (the business cannot run without it), an improvement (it would save time or money) or later. In manufacturing, must-haves are usually about stock, production and statutory compliance. Dashboards, mobile reports and automation are improvements, valuable but not decisive.

Typical must-haves for an Indian manufacturer:

• Stock by lot, heat or batch, with rack locations
• Multi-stage production with rejection and rework
• Job work with challans, receipts and shortage tracking
• GST, e-invoice, e-way bill and ITC-04 from the same entries
• Your print formats for invoices, challans and labels
• Units and conversions your trade uses
• Role-based access so the gate sees only the gate
• Works with your internet and power conditions

Step 3: Test for industry fit

Industry fit is the single best predictor of a smooth implementation. A system that already knows your trade needs configuration; one that does not needs development. A quick way to test fit is a vocabulary test: ask the vendor to show, without preparation, how their system handles five terms from your floor.

If the answer to most of them is "we can configure that", treat it as a development project with the timeline and cost that implies.

Step 4: Insist on a demo with your own data

A demo on the vendor's sample company proves that the software works for the sample company. Before the demo, send one week of real documents: a few orders, purchase orders, challans, inspection sheets and invoices. Then use a script.

1. Enter one of your real orders and plan its material.
2. Receive material at the gate and into the store, by lot.
3. Run it through two production stages with some rejection.
4. Send part of it to a job worker and receive it back short.
5. Fail one lot in inspection and try to invoice it.
6. Dispatch the rest with your invoice format, e-invoice and e-way bill.
7. Show the cost of that order and the stock left behind.

Watch who drives. If the vendor's engineer needs to leave the script often or promises "this will be there after customisation", note it. eManage demos are run this way, on your documents, by the engineer who would implement the system.

Step 5: Check who implements and who supports

Most ERP problems are implementation problems. The software matters, but the team that studies your floor, sets up masters, migrates data and trains people matters more. Ask directly:

• Will the implementation be done by the vendor's own staff or by a reseller?
• Will the engineers be on site, and for how many days?
• Have they implemented this edition in a plant like yours? Can you speak to that plant?
• Who answers support calls after go-live, in which language, and within what time?
• How are statutory changes (GST, e-invoice rules) delivered, and at what cost?

Step 6: Decide on hosting and data

Cloud or on-premise is a practical decision, not a philosophical one. Consider your internet reliability, power backup, number of locations, need for remote access and who will manage backups. Our comparison of cloud vs on-premise ERP goes through each factor. Whichever you choose, confirm three things in writing: that the data belongs to you, how you can export it, and how backups are taken and tested.

Step 7: Compare full cost, not licence price

Quotes are hard to compare because vendors package them differently. Normalise them into the same five-year view:

More detail, including what drives each line up or down, in how much does ERP cost?

Step 8: Talk to a plant like yours

Ask the vendor for two references in your trade and region, and call them yourself. The useful questions are specific: how long the go-live really took, what was harder than expected, how quickly support responds, whether the floor still uses the system or has gone back to registers, and what they would do differently. A visit to a running plant is worth more than any presentation.

A simple scorecard

Score each shortlisted system from 1 to 5 on these lines and weight them by what matters to you. The weights below are a reasonable starting point for a manufacturer.

Price carries less weight than you might expect. In our experience the cheapest system to buy is rarely the cheapest to own.

Common mistakes to avoid

• Buying on brand. A famous name does not guarantee fit for a dyeing house or a forge.
• Choosing by demo polish. Dashboards impress; challans and lots decide whether the plant can run.
• Skipping the process study. Without it, requirements appear during implementation, when they are most expensive.
• Underestimating data cleaning. Duplicate parties and inconsistent item names slow every go-live.
• No internal owner. Someone in the plant must own the project and the masters.
• Too much customisation. If you change the software to match every old habit, you rebuild the old problems.

Questions buyers ask