In many traditional factories, the accounts live in bookkeeping software, the inventory lives in spreadsheets, and where the material actually sits lives in veteran workers' heads. When an ERP comes up, the first question is usually "buy one or have one built?" and the second is "where do we start?" Drawing on the ERP we have been building for the stone industry since 2023, here's what's worth settling before you begin.
Packaged ERP is built to do common processes very thoroughly. The more generic your processes, the better value it is; the more unusual they are, the more you end up working around the system. A rough guide:
| Situation | Better fit |
|---|---|
| Mostly buying, selling and bookkeeping, with fixed item specifications | Off-the-shelf ERP |
| Mostly generic, with a few special needs (a custom report, one connected device) | Off-the-shelf ERP plus custom integration |
| Materials change form during processing, and pricing and lookups are industry-specific | Custom ERP |
| Storage locations, RFID or shop-floor devices must be managed too, and staff work on phones | Custom ERP, or custom modules integrated with existing systems |
Stone is a classic "unusual" case. A raw block weighing over ten tonnes is sawn into slabs and then processed into finished products, and the same material has different units, specifications and pricing at each stage. There's also industry-specific work such as contract sawing and outsourced processing. Forcing that into a generic ERP means bending the work to fit the software, so we designed the system around how the work is actually done.
A practical test: take the three documents you use most (say, purchase orders, processing orders and invoices) and ask the vendor, "How do I do these three in your system?" If the answer is "you'd do it differently", weigh how much that change will cost you.
ERP has many modules — purchasing, production, inventory, sales, accounts — and switching them all on at once is risky. We recommend getting "what material do we have and where is it" right first: processing, sales and accounts all depend on inventory data, so accurate inventory gives every other module reliable data to build on.
That's how the stone ERP grew:
Starting with inventory has another advantage: the results are easy to see. The stone pilot first proved that stocktaking and finding material could be fast and accurate, and only then was RFID brought into the ERP's location management. Visible results in the first phase make the next modules easier to introduce.
Every record in an ERP is linked by numbers: raw stone IDs, slab IDs, document numbers. Once a numbering scheme is in use it's very hard to change, so it's worth getting right before go-live.
In the stone case, the Stone & Resource Industry R&D Center (SRDC) coordinated an industry-wide barcode format: 17 characters for raw stone and 20 for slabs, covering company code, material code, year-month and serial number. With one format, a slab is tracked by the same ID from cutting through storage, stocktaking and finding it again. See point 3 of how to implement an RFID inventory system for details.
Many plants have stock quantities in their system but not where the stock actually is. Putting locations into an ERP requires agreement first: how areas are divided and how locations are numbered.
When we worked with a primary stone processing plant in eastern Taiwan, SRDC first surveyed the raw stone and stock yards and drew 2D/3D site maps; production and warehouse staff then agreed on location numbering together; only after that did we set up locations and visual dashboards in the ERP. Without agreed numbering, a system just records the chaos more precisely. See storage-location visualization in stone plants.
One of an ERP's biggest benefits is not re-entering the same data in different places. The ID and specifications recorded when raw material arrives carry through processing records and inventory to the invoice.
When evaluating, walk one real order through from start to finish: where is the data first entered? At each step after that, is it carried over automatically or typed again? Wherever data has to be re-entered is usually where records stop matching later.
After go-live, what users do most isn't data entry — it's looking things up. If searching doesn't match how people work, they go back to their spreadsheets.
Stone companies, for example, look up past invoices by contract-sawing invoice number or raw stone ID, so we made both directly searchable. Before implementation, list the three things each role looks up most often each day, and ask the vendor to show how — and in how many clicks — each is done in their system.
Most factory data is created on the shop floor, not in the office. If only office computers can use the ERP, floor data arrives late or not at all.
The stone ERP is web-based and runs in a browser, with phone and tablet versions so staff can look up where material is from anywhere in the plant; stocktakes use a handheld RFID reader app that uploads the latest locations to the ERP as it goes. When evaluating, ask: which tasks must happen on the floor, on what devices, and is the network coverage good enough?
After go-live, new needs keep coming: new processing steps, new reports, new equipment to connect. Look beyond today's features to whether the system can be extended later, for example:
Pre-implementation checklist:
For preparing requirements, quotes and contracts when commissioning development, see commissioning custom software. If your plant's processes don't fit a generic ERP — stone industry or not — see our Industry ERP development service, or tell us how work flows on your floor.