GUIDE

Inventory you can trust, item by item.

A field handbook for keeping system stock equal to physical stock: the usual causes of drift, how to set up items and warehouses, how to run a cycle-counting program, and how to measure accuracy honestly.

SUNDIAL / GUIDE

The inventory accuracy handbook.

Guide 04

What this guide covers.

If the system says you have stock and the shelf is empty, every downstream decision is wrong: promises to customers, reorder points, production plans and the balance sheet. This handbook covers how to get the numbers right and keep them right.

By the Sundial team. Free to use, copy and adapt for your own business.

Try the demo   Talk to us

1. What accuracy means

A stock record is accurate only when all four of these match the physical reality:

  • Quantity: the count on the shelf.
  • Location: the warehouse and bin where it actually is.
  • Identity: the right batch or serial number.
  • Value: the cost the books carry for it.

2. Why stock drifts

  • Goods received without a receipt in the system, or received against the wrong purchase order.
  • Transfers between warehouses moved physically but not recorded.
  • Material issued to production without a stock entry, or more used than the bill of materials says.
  • Customer returns put back on the shelf without being booked.
  • Unit of measure mix-ups: received in cases, sold as each.
  • Negative stock allowed, so sales are recorded before the goods are.
  • Damage, expiry and samples written off on paper but not in the system.

3. Set the system up to prevent errors

  • Give every item a single code, a stock unit and correct conversion factors.
  • Model warehouses the way stock is stored: separate warehouses for stores, work in progress, finished goods, rejected and in transit.
  • Turn on batch tracking for anything with an expiry or recall risk, and serial numbers for anything under warranty.
  • Do not allow negative stock.
  • Choose a valuation method (FIFO or moving average) and keep it consistent.
  • Require a document for every movement: receipt, transfer, issue, delivery, return, adjustment.

4. Build a cycle-counting program

Counting everything once a year is slow, disruptive and finds errors months after they happened. Count a little every week instead.

  • Classify items by value and movement. A common starting point: A items (high value or fast moving) counted monthly, B items quarterly, C items twice a year.
  • Add trigger counts: count an item when it hits zero, when a pick fails, or after a large adjustment.
  • Spread counts so each week has a similar workload.
  • Set a tolerance per class, tighter for A items, so small differences on cheap items do not consume the team.

5. Run a count properly

  • Generate the count list from the system, without showing the expected quantity (a blind count).
  • Stop movements for the items being counted, or record the exact cut-off time.
  • Count by location, and record batch or serial numbers as well as quantity.
  • Recount anything outside tolerance, using a different person.
  • Find the cause of every confirmed variance before posting it.
  • Post the adjustment with a reason code, so you can see patterns over time.

6. Measure accuracy honestly

Measure accuracy as the share of counted lines that were within tolerance, not as the net value difference. Net value hides errors: a surplus on one item cancels a shortage on another.

Example (sample figures for an illustrative company, Kestrel Workspace): 200 item-location lines counted this month; 188 within tolerance; line accuracy 188 ÷ 200 = 94%. Net value difference was only $3,200, but the gross value of all differences was $61,400. The second figure is the one to act on.

Track accuracy by warehouse and by reason code. The reason codes tell you which process to fix.

7. Fix the flows, not only the numbers

  • Receive only against a purchase order, and inspect before stock becomes available.
  • Pick from a pick list, and confirm what was picked.
  • Book production consumption from job cards as work happens.
  • Book returns into a quarantine warehouse until inspected.
  • Record landed cost (freight, duty, insurance) so values are right, not only quantities.

8. Then tune reorder levels

Reorder levels set on inaccurate stock either tie up cash or cause stock-outs. Once accuracy is stable for two or three months, review reorder levels and safety stock for A items against recent demand and supplier lead times.

Common pitfalls

  • Posting count adjustments without finding the cause, so the same error returns next month.
  • Counting with the expected quantity visible.
  • Using one warehouse for everything, including rejected and in-transit stock.
  • Judging accuracy by net value difference.
  • Letting the annual stock-take replace continuous counting.

Where Sundial helps

Sundial’s Inventory & Warehousing module covers multi-warehouse stock, batches and serial numbers, pick lists, packing, delivery trips and landed cost, and every stock movement posts to the ledger as it happens.

Explore Inventory & Warehousing   Cycle counts use case

THE NEXT AGE OF ERP

Run the whole business
in one calm system.

Try the demoTalk to us