ERP system development
in Silver Spring

For a company that makes things, the bill of materials is the record the whole ERP leans on. It says which parts, in which quantities, go into one unit of a finished product. Purchasing, costing, stock and the production schedule all read from it. When the bill is wrong, every one of those modules is wrong in a quiet, plausible way.

Most firms arrive with bills scattered across engineering drawings, spreadsheets and the memory of a senior planner. Moving them into the ERP is a project of its own. Plan it as one. Someone has to decide, product by product, which version is current and which spreadsheet is a leftover from a design that was dropped.

Structure is the next decision. A flat bill lists every component under the finished item. A multi-level bill nests sub-assemblies, each with its own bill, which mirrors how the product is really built. Multi-level takes longer to set up. It pays back the first time a sub-assembly is shared by several products, because one change updates all of them.

Units of measure cause more trouble than they should. Steel is bought by the kilogram, stored by the bar and used by the millimetre. Fabric arrives in rolls and is cut in metres. The ERP needs a conversion for each pair, and the conversions need an owner. A missing factor turns a sensible order into a truck of the wrong size.

Material requirements planning, or MRP, is the engine that uses the bill. It takes forecast and confirmed demand, explodes it through the bills, subtracts what is in stock or on order, and proposes purchase and production orders with dates. The logic is old and reliable. The inputs are where it fails.

Three inputs deserve a review before go-live. Lead times per supplier, since the defaults copied from an old system are often guesses. Safety stock, which should reflect how much a shortage actually costs. And scrap or yield factors, so a process that loses a little material on every run is planned with that loss included.

Expect planners to distrust the first MRP runs. That is healthy. Run it in parallel with the current method for a few cycles and compare the suggestions line by line. Each disagreement points to a bad lead time, a wrong bill or a rule the planner applies by habit and nobody wrote down.

The month-end close is where finance finds out whether the ERP was set up for accountants or around them. A good setup lets the close run as a checklist inside the system. A poor one sends the team back to exports and spreadsheets on the last working day, which defeats much of the reason for buying the software.

Start by writing down the current close as it really happens. Every step, who does it, what it depends on, and how long it takes. The list is always longer than expected. Keep it anyway. It becomes the specification for the close tasks, reminders and dependencies the ERP should hold.

Subledgers come first. Receivables, payables, inventory and fixed assets each post to the general ledger, and each needs its own cut-off. Goods received on the last day but invoiced next month need an accrual. The ERP can post it automatically from the receipt, if the receipt is recorded in the system on the day it happens.

Period locks look minor. They are not. Once a month is closed, nobody should be able to post into it without a controlled reopening. Many teams leave periods open for weeks because a late invoice might arrive. Better to allow postings into the next period with a note, and keep the closed month closed.

Reconciliations belong in the system too. Bank feeds matched against the ledger, intercompany balances checked against each other, stock value compared with the inventory subledger. Each reconciliation should have a preparer and a reviewer, with the sign-off stored against the period.

Watch the difference between the subledger and the ledger. When the two agree by design, finance can trust the numbers without rebuilding them. When they drift, the cause is usually a manual journal posted straight to a control account. Block those accounts from manual entry. Then the drift stops.

Measure the close by days, not by effort. The first months will be slower. That is normal. Track which steps overrun, fix the setup behind them, and the close should shrink cycle by cycle. Reports built for management then arrive while the numbers still matter.