A plain explanation of what an ERP system does, what it replaces, and the point at which a growing business genuinely needs one.
ERP (Enterprise Resource Planning) software is a single system that runs the operational and financial processes of a business on one shared database — purchasing, inventory, production, sales, finance and often HR. The defining characteristic is not the feature list: it is that every module reads and writes the same records, so a goods receipt updates stock, the supplier balance and the general ledger in the same transaction.
Most businesses do not start with an ERP. They start with accounting software, a spreadsheet for stock, another for production planning, and a shared folder of quotations. That works until the numbers stop agreeing with each other. The symptom is almost always the same: someone is re-keying data between systems, and the month-end close has become an exercise in reconciliation rather than reporting.
The general ledger, payables and receivables sit at the centre. Every operational transaction eventually posts here, which is why the chart of accounts is designed before anything else is configured.
Goods receipts, issues, transfers and stock valuation. This is where most of the day-to-day transaction volume sits, and where costing errors are most expensive.
Quotation to order to despatch to invoice, with credit limits and pricing applied by the system rather than remembered by a person.
Bills of material, routings, work orders and work-in-progress costing. Manufacturing is the module that most often decides whether an ERP fits.
A report in finance and a report in the warehouse answer from the same tables. That is the whole point of the exercise.
A distributor with forty staff runs accounting in one package, stock in a spreadsheet, and orders in a shared mailbox. Nobody thinks of this as a software problem. It shows up as a person: someone spends roughly a day a week moving numbers between the three.
Count the hops for a single order. A customer emails it; someone types it into the spreadsheet to check stock; someone types it again into the accounting package to raise the invoice; someone updates the spreadsheet after dispatch. One order, three re-keyings, and three opportunities for the number to change. At two hundred orders a month that is six hundred manual transcriptions, and an error rate of even half a percent produces three wrong numbers a month that nobody can trace.
On a single database the same order is entered once. The stock reservation, the dispatch note, the invoice and the ledger posting are views of that one record rather than copies of it. The eight hours a week do not get faster — they stop existing, because the work they consisted of was moving data between systems that now share one.
| Term | What it means |
|---|---|
| Module | A functional area of the system — finance, purchasing, inventory, payroll. Modules share one database, which is what separates an ERP from several products sold together. |
| Master data | The reference records everything else points at: customers, suppliers, items, the chart of accounts. Transactional volume grows; master data should not. |
| Transaction | A single business event the system records — a purchase order, a goods receipt, an invoice. Each one usually carries an accounting consequence. |
| Document flow | The chain linking related records: requisition to order to receipt to invoice to payment. Being able to walk that chain in either direction is most of what an ERP is for. |
| Period close | The monthly process of finishing a period so the numbers stop moving. How long it takes is the most honest single measure of whether an ERP is working. |
A long feature list is easy to produce. What matters is whether the system handles your costing method, your approval chain and your statutory reporting without customisation.
It is the hardest thing to change after go-live and the easiest to get wrong early. Design it before configuration starts.
Extracting and cleaning master data is your team's work and it takes longer than anyone plans for. Start it in the first month, not the last.
Every customisation is a cost at the next upgrade. Push configuration as far as it goes first.
| Metric | What it tells you | What distorts it |
|---|---|---|
| Days to close the month | Whether your data is trustworthy as it is entered, or only after correction. | A close that looks fast because adjustments are posted in the next period instead. |
| Order-to-invoice cycle time | How much manual handoff sits between a customer order and a billable document. | Averaging across channels — a fast web order can hide a slow contract one. |
| Stock record accuracy | Whether the system reflects the building. Below about 95% people stop trusting it and start checking manually. | Counting only fast movers, which are the ones most likely to be right. |
| Manual re-keying points | The count of places a human retypes data that already exists elsewhere. The most direct measure of integration debt. | Nothing — this one is simply worth counting honestly, and it is usually higher than expected. |
We build and support the systems this page describes. Start with a free working proof of concept — no cost, no commitment.