Stock Count, Variance and Shrinkage

Stock goes missing in small amounts: waste nobody logged, wrong portions, deliveries not checked, theft. You only see it by comparing what the books say should be on the shelf with what is really there, and most small businesses count, shrug at the total and never see which items drift.

Add your items with a location, unit and unit cost, or import them from CSV. Print blind count sheets by location so counters write what they see, not what the system says. Start a count session: the opening stock comes from the previous count, you enter receipts and usage since then, and then the counted quantity for each item (typed, or imported from CSV by SKU for counts taken on a phone or scanner). The variance report shows expected against counted, the money difference, shrinkage as a percentage of usage, lines outside your tolerance in percent or dirhams, and the trend across your counts.

What is inside:

• Items, Count sheets, Count, Variance report, Settings and Start here tabs
• Blind and expected-quantity count sheet options, printable per location
• Several count sessions kept side by side, with the previous count carried forward
• CSV import (items, counted quantities), CSV export of the variance report, JSON backup, print
• A worked example of three monthly counts for a small cafe store

Who it is for: cafes, restaurants, groceries and shops with up to a few hundred items who count monthly or weekly.

What it is not: a variance is a signal, not proof of loss, because unrecorded usage, wrong units and a miscount look the same, so recount flagged lines before acting. It does not post to your books and is not a stock valuation for tax or audit.

How it differs: the Stock Count Sheets and Reorder Planner and the Retail Daily Cash-Up and Stock-Take Pack in the catalogue are Excel files. This app keeps a history of count sessions, carries the opening stock forward, takes counted quantities from CSV and shows the trend and the money lost by item and location.

Tested: expected, variance, shrinkage percent and valuation were compared with a separate Python implementation on randomised count histories, 6,902 values, with no differences. 16 scripted browser checks cover the example, sessions, CSV import, backup restore, print and the 375 and 1280 pixel layouts, with no console errors. Browser tests ran in Chromium and WebKit; Firefox, a real phone, a real printer and a real scanner were not available, so they were not tested.