FIFO Costing, the Valuation Audit Trail and Predictive Reorder Levels

Moving average is the default: every receipt blends into one average cost per item, and every issue leaves at that average. In Settings, Inventory Defaults, an organization can switch to FIFO. Under FIFO each receipt becomes a cost layer, and every issue, sale or transfer out uses up the oldest layers first, at what those units actually cost. The item’s unit cost then shows the value of the layers still on the shelf divided by their quantity, so every report that values stock reads the FIFO figure without any other change.

Layers are written under both methods, so the receipt history exists whichever is chosen. Switching never revalues stock: on the day of the switch each item’s open layers are closed into one layer at that day’s unit cost, so the inventory account and the stock valuation agree, and the new method applies from the next movement. Stock that no layer explains, such as opening stock, becomes an opening layer at the current unit cost before the next movement, and layers that add up to more than is on the shelf are trimmed oldest first, so FIFO never costs units that are not there.

In practice: a Nairobi hardware wholesaler buys 100 bags of cement at KES 700 and then 100 at KES 760. Under moving average, a sale of 50 bags costs 730 each. After switching to FIFO, the switch closes both receipts into one layer at the day’s cost; from then on each new delivery is its own layer, and a sale takes the oldest first. The finance lead chose FIFO because the group auditors report on it, and recorded the decision before the switch, because the method should not be changed back and forth to flatter a period.

Reports, Valuation Audit Trail lists every change to an item’s unit cost or buying price: when, from what to what, how much was on hand, the method in force, the reason, a reference and who made it. The trail is written by the item record itself on every save that moves either value, so no screen, import or background job can change a cost without leaving a row. The reasons are stock received, landed cost apportioned, assembly or disassembly, POS return restocked, FIFO layers consumed, edited on the item, bulk import and system recalculation.

Reports, Cost Layers lists every receipt still on the shelf, oldest first, with its unit cost, its landed cost and how much of it is left, and totals the open layers, items, units and value. Under FIFO this is the order stock leaves in. Under moving average the layers are history and the report says they do not set the cost. Filter both by item or reference; the audit trail also filters by reason and date, and exports to CSV and PDF.

In practice: margin on a line of solar lanterns drops overnight. The analyst opens the audit trail filtered to that item, sees a row with the reason edited on the item, the old and new values and the clerk’s name, and a second row the next morning with the reason stock received. The edit was a typo in the buying price; the cost itself moved because of a genuine receipt at a higher price. One conversation, with the evidence on screen.

Reports, Stock Ledger Reconciliation compares three figures for each item. Book is the item’s total. Shelf is the sum of its locations. Ledger is every document that moved it netted together: receiving, check-ins and check-outs, counts, POS, transfers, assembly and returns, each in its own signed column. Matched means all three agree. Book not equal to Shelf points to a write that updated one side and not the other. Unexplained is stock no document accounts for, with the units shown. By default the report shows only the items that do not match.

Reports, Inventory Exceptions is one live list of everything that needs a person: negative shelves, stale stock past your dead-stock period, overstock above a warehouse maximum, expired batch units still on a shelf, stock with no bin or in an inactive bin, and fast movers (the top quarter by 30-day usage) holding less than their lead time. Nothing is stored, so an exception disappears the moment the stock is fixed. Both reports are in the app.

In practice: before a stock count, a supermarket’s inventory controller runs the reconciliation and finds 12 items with unexplained stock. Nine trace back to a till that was offline and synced late; three have no document at all, and those go on the count sheet first. On the exception list, two negative shelves are fixed by receiving a delivery that had been left in draft, and they drop off the list immediately.

On an item’s Stock tab, a warehouse level can be switched from Fixed to Predictive. Predictive levels take a service level, the chance of not running out in an order cycle (95% by default, from 50% to 99.9%), the days of stock each order should cover (30 by default), and optionally Follow last year’s season. Minimum and maximum are then calculated straight away and recalculated every night at 03:20; the fields go read-only and a tooltip shows the working. The Reorder screen and stock balancing read them exactly as they read typed levels.

Demand is that warehouse’s own daily usage over 90 days, with zero days counted so an item that sells in bursts is treated as uncertain. Season compares last year’s rate over the coming lead time with last year’s average, only once there is a year of history, kept between half and double. Lead time is a typed figure for this item and supplier first, then the supplier’s real order-to-receipt days from the ten most recent receipts, then the item’s own lead time. Safety stock combines demand and lead-time variability at the service level; the minimum is the lead-time demand plus safety stock, and the maximum adds the review period’s demand divided by the supplier’s fill rate, never more than doubling for a poor filler.

The Supplier lead times panel on the same tab lists every supplier that has a typed lead time, has supplied the item, or is its preferred supplier, with typed days, observed days and the number of receipts behind them. In practice: a pharmacy chain sees an importer’s observed lead time of 9 days built on two rushed deliveries, knows the real figure is 21 days by sea, types 21, and the item’s predictive levels recalculate on save rather than waiting for the night.