Garage Management: Parts, Open Work, and the Vehicle Record in One Chain

A part in a garage changes hands four times: reserved against a work order, issued physically from the shelf, fitted to a vehicle, and billed on an invoice line. If the system records only the first and the last, the two middle handovers become a stock variance found at the annual count. The chain is the product, not the screen.

This is a technical description of data and workflow, not professional advice on vehicle repair or on regulatory compliance.

The chain: where a part changes hands

Every handover gets a record, a timestamp and an owner. Without those there is no way to explain a variance:

• Reservation. The technician adds a part to a work order. On-shelf quantity does not move; available quantity drops.
• Issue. The part physically leaves the store for the lift. Now on-shelf drops, and the system records who took it and against which work order.
• Fitting or return. Either the part is fitted and bound to the vehicle, or it goes back to the shelf as a separate return record. A part issued, never fitted and never returned is exactly the gap the annual count finds.
• Billing. The invoice line is generated from the issue record rather than retyped. One identifier travels from work order to invoice, on the same principle as one identifier across quote, order and invoice.

The rule is short: quantity never moves without a record, and a record is never created without an owner.

Three stock numbers that are easy to confuse

The first four rows are numbers; the fifth is a relationship between SKUs. Without an alternates table, every shortage looks worse than it is.

Why does the shelf disagree with the system?

Because some movements happen by hand and never through a screen. These are the common sources, roughly in order of size:

• An unrecorded issue. A technician took a part from an open shelf to avoid stopping work, and the record was created that evening or never.
• An unrecorded return. The part did not fit, went back on the shelf, and the system still counts it as fitted.
• Partial goods receipt. Six of eight units arrived, eight were booked in, and the variance was born on the delivery note.
• The same item under two SKUs. A manufacturer part number and a local code were opened as two records, and each shows half the quantity.
• Counting during working hours. The shelf was counted while issues continued, so the count describes two different moments.
• Consumables. Oil, filters and fasteners are consumed in variable amounts and are usually drawn from a batch rather than tracked per unit.

What does an open work order actually cost you?

An open work order holds three kinds of resource at once, none of which is available to any other job: parts already issued, technician hours already logged, and physical space on a lift or in the yard. That makes work-order ageing an operational report rather than a financial one:

• Open up to two days. Normal. The part is on order or the work is in progress.
• Open three to seven days. Needs an owner and a reason code: waiting for a part, waiting for customer approval, waiting for a third party.
• Open beyond seven days. Needs an explicit decision. Some of the parts issued against it are already missing from the shelf without anyone noticing.

These statuses are worth something only as a closed value set. A free-text notes field produces dozens of phrasings of the same reason, and then nothing can be counted.

The vehicle record: what a garage stores and what it merely looks up

A garage's vehicle record is not a copy of an official register. It is service history the garage created, plus fields pulled from an external source and flagged as such:

• Fields the garage creates. The chassis number as read from the car, arrival date, odometer reading, the customer's complaint in their own words, work performed, parts fitted and warranty terms.
• Fields that are looked up. Manufacturer, model, model year and fuel type. These are supporting data, and if the external source returns nothing the record still opens. How to build that kind of check without treating a register as absolute truth is covered in validating public vehicle data.

The Ministry of Transport's open register of licensed garages and testing stations shows why lookups need care: it holds roughly 13,900 rows, but a row is a garage-and-trade pair. Garage number 16, for example, appears in four separate rows with the same name and address because four trades are registered to it. Counting rows counts trades, not garages.

What breaks in practice

• Free-text part descriptions. A technician types a description instead of picking an item, and the part appears in no consumption report.
• Plate number as the key. A vehicle changes plate and its history splits into two records. The chassis number is the stable key.
• Overwritten purchase cost. Goods receipt updates the cost on the item itself, and all historical profitability shifts retroactively.
• Linking RECALL by model alone. The public manufacturer recall dataset is keyed on manufacturer, model and a build-date window, so a match against one car is a flag to check, nothing more.
• A stale supplier price file. A price list nobody refreshed produces quotes at last year's numbers, which is why pulling supplier catalogues and prices needs a validity date on every row.
• A retyped invoice. Anyone rekeying lines instead of deriving them from the issue record creates a permanent gap between what left the shelf and what was charged.

Sources

• data.gov.il - licensed garages and testing stations dataset
• data.gov.il - manufacturer RECALL notices dataset