Saudi FMCG manufacturers: the production visibility gap and what ZATCA changed

Saudi manufacturers are in an unusual position. The commercial side of the business has been digitised by regulation, faster and more thoroughly than almost anywhere else in the region. The production side, in most mid-size operations, has not moved at all.

That asymmetry is the opportunity, and it is specific to this market.

What e-invoicing already forced

ZATCA’s phased e-invoicing rollout did something no internal initiative had managed: it made structured, machine-readable transaction records mandatory. Invoices are generated from a system, cleared, and archived in a defined format. There is no version of the business where sales data lives in a drawer.

The consequence is that most manufacturers here now have clean revenue data. They know exactly what they sold, to whom, at what price, with a timestamp.

They still do not know what it cost to make.

The gap this creates

Clean revenue against estimated cost produces a margin figure that is precise on one side and a guess on the other. That is arguably worse than both sides being rough, because it looks authoritative.

The estimate is nearly always built the same way: raw material cost plus a fixed waste percentage plus allocated overhead. The waste percentage was measured once, some years ago, and has been applied ever since. It does not move with supplier variability, humidity, line speed, operator experience, or product mix, all of which move constantly.

What is missing from the calculation entirely:

Rework. Product that failed and went back through, consuming machine time and often material, but recorded as a normal run because it eventually sold at full price.

Unplanned downtime. Not breakdowns. Those get noticed. Changeover overruns, material staging delays, short stoppages resolved without a record. In the operations we have instrumented, this consistently runs at fifteen to twenty percent of available capacity and is almost never in anyone’s cost model.

Yield variance by SKU. The blended figure conceals which products are actually expensive to make. In an eleven-SKU operation there are usually two or three where real cost exceeds the estimate by enough to change whether they should be in the range.

Why this matters more here than elsewhere

Two Saudi-specific pressures make the gap expensive rather than merely untidy.

Localisation and investment scrutiny. Manufacturers pursuing local content certification, government contracts, or outside investment are increasingly asked to evidence unit economics rather than assert them. An estimated cost per unit does not survive that conversation.

Input volatility. Import-dependent raw material costs have moved sharply and unpredictably. A fixed-percentage cost model absorbs that movement invisibly, which means pricing decisions lag the actual position by a full accounting cycle.

What closing it involves

The mechanism is capture at the line, and the constraint is time: if recording an event takes more than about ten seconds, it will be reconstructed from memory at the end of the shift and the data will be confidently wrong.

Practically that means fixed terminals at each line rather than a shared office PC, reason codes as physical buttons rather than dropdowns, no per-transaction login, and Arabic and English interfaces at parity, not Arabic as a translation layer over an English system, which operators reject quickly and quietly.

Material issued, output recorded, rejects with a reason, downtime with a cause, changeover start and end. Five events. Everything downstream, from true cost per unit and yield variance to utilisation with attributable causes and margin by SKU, is arithmetic on those five.

What the first month usually surfaces

Two findings, almost every time.

The first is a product being sold below its real cost, protected by a blended margin that looked acceptable. It is frequently a product sales pushes hardest, because it is easy to sell, which is often precisely why it is priced where it is.

The second is changeover time, which turns out to be roughly double what everyone believed. Reducing it costs nothing but sequencing, and it typically returns more capacity than the additional line the operation was considering buying.

Neither requires new machinery. Both require being able to see the line that is already there.