Every ERP holds two kinds of data. Master data describes the things the business deals with again and again: items, customers, suppliers, warehouses, accounts. Transactions record what happened to them: a purchase order, a goods receipt, an invoice, a payment. Master data is set up once and changes slowly; transactions are created every day and, once posted, never change.
In a spreadsheet the two are mixed. A sales sheet repeats the customer's name, address and payment terms on every row, typed slightly differently each time ("PT Sinar Jaya", "Sinar Jaya PT", "PT. Sinar Jaya"). When you try to total sales per customer, you get three customers. An ERP separates them: the customer exists once, and every order refers to it.
The main master data objects
| Object | What it holds | Who usually owns it | Voltrus screen |
|---|---|---|---|
| Item | SKU, name, type, base unit, cost, reorder point, planning settings, lot/serial tracking | Inventory or engineering | Items |
| Unit conversion | 1 box = 12 pcs, 1 roll = 50 m | Inventory | Item units |
| Customer | Name, payment terms, credit limit, price list, tax status, NPWP | Sales and finance | Customers |
| Ship-to address | Delivery addresses per customer | Sales | Ship-to addresses |
| Supplier | Name, terms, lead time, currency, bank details, NPWP, PKP status | Purchasing and finance | Suppliers |
| Item supplier | The supplier's part number and lead time for an item | Purchasing | Item suppliers |
| Price list | Selling or buying prices, quantity tiers, validity dates | Sales / purchasing | Price lists |
| Warehouse and bin | Where stock can sit | Inventory | Warehouses, Bins |
| Bill of materials | What a product is made of | Engineering | Bills of materials |
| GL account | The chart of accounts | Finance | Settings → Ledger |
Each of these is a record you create, check and then use for years. Transactions copy only the reference (and, where the law requires a frozen copy, the value at that moment: a delivery order keeps the address it shipped to even if the customer moves later).
Why master data quality decides everything
Every transaction inherits from master data. A wrong value is not one wrong number; it is wrong on every document that uses it.
- A wrong unit cost on an item values every receipt and every shipment wrongly under moving-average costing, and your gross margin is off for months.
- A wrong payment term on a customer gives every invoice the wrong due date, so your aging report and your collection calls are wrong.
- A missing NPWP or a wrong PKP flag on a supplier means you cannot claim the input PPN on their invoices, or Voltrus refuses to put PPN on the bill at all.
- A wrong base unit (pieces versus boxes) is the worst of all: it multiplies every stock figure by twelve.
SAP consultants say "garbage in, garbage out", and they mean it literally. On most failed ERP projects, the software worked; the master data did not.
Rules for good master data
- One record per real thing. Search before you create. Duplicates split history and break reports. If you find one, stop using the extra record (deactivate it) rather than deleting it once it has transactions.
- Codes that never change. An item's SKU and a customer's name are what people search by. Pick a coding scheme before you import and stick to it. A good SKU is short, unique and meaningless enough that it never needs renaming (
RM-0412ages better thanBLUE-PAINT-5L-NEW). - The base unit is the smallest unit you count. Stock always moves in the base unit; buying in boxes and selling in pieces is handled by unit conversions, not by separate items.
- Fill the fields that drive behaviour. Item type, unit, cost and reorder point; customer terms and credit limit; supplier lead time and tax fields. Descriptive fields can wait; behavioural ones cannot.
- Deactivate, don't delete. Records with history must stay so old documents still make sense. Voltrus only lets you delete a document while it is in its first status or a final one; master records with transactions should be set inactive.
Transactions: what makes them different
Transactions follow the opposite rules. They are created constantly, by many people, and they must be immutable once posted:
- The stock ledger in Voltrus is append-only. A receipt, a shipment, a production output or an adjustment adds a row; nothing is overwritten. Quantity on hand is calculated from those rows.
- Journals are never edited. A mistake is fixed by reversing the journal.
- An issued invoice keeps its amounts. Only notes, the due date and the Coretax faktur fields can change; anything else needs a credit note.
The next article, Documents, statuses and the document flow, explains how transactions move from draft to final.
A worked example
Here is the same week of sales, first the spreadsheet way, then the ERP way.
| Date | Customer (typed) | Item (typed) | Qty | Price |
|---|---|---|---|---|
| 3 Mar | PT Sinar Jaya | Cat Biru 5L | 10 | 250,000 |
| 4 Mar | Sinar Jaya | cat biru 5 liter | 6 | 245,000 |
| 6 Mar | PT. Sinar Jaya | CAT-BIRU-5 | 12 | 250,000 |
A pivot table shows three customers and three products. The price on 4 March was a typo, and nobody notices.
In an ERP, the customer PT Sinar Jaya and the item PNT-005 (Cat Biru 5L) each exist once. Each order line picks them from a list, and the price comes from the customer's price list unless someone types over it. Voltrus records where every line price came from (price_source: manual, quotation, party price list, default price list or last price), so a manually typed 245,000 stands out in review.
In Voltrus
Checklist
- Write down a coding scheme for items, and agree on how customer and supplier names are written (with or without "PT", for example).
- Pick one owner per master data object.
- Clean duplicates in your spreadsheets before importing; it is ten times harder afterwards.
- Check base units and unit conversions for your top 50 items by hand.
- Fill the tax fields (NPWP, PKP, NITKU) for every active customer and supplier.