This is a composite scenario built from how we deliver this kind of work. The client is fictional and the figures are targets, not measured results. We will replace it with a named engagement when a client agrees to be quoted.
A consumer electronics distributor with three warehouses, eleven delivery vans and about 9,000 SKUs across Bangladesh
- System and shelf disagreed every month. The count took three days across three warehouses, vans carried stock the system thought was in the warehouse, and unexplained variance ran at about 2.4 percent of stock value.
- Balances were typed into the system by hand, so a wrong number could be overwritten without a trace. Nobody could say which movement had caused a difference.
How we approached it.
A ledger, not a balance
Every receipt, issue, transfer, van load and return became an event with an actor and a time. The balance is a sum, never a typed number.
Scan, do not type
An offline-first Android app with barcode and QR scanning for warehouse staff and van drivers, syncing when the network returns.
Blind cycle counts
Counters do not see the system figure. A variance above the threshold creates an adjustment event that waits for a supervisor.
Vans as locations
Each van is a stock location. Loading and delivery are transfers, so van stock is always visible.
Alerts and reorder
Low-stock alerts per warehouse and a reorder list from sales velocity.
Try it yourself.
A simplified illustration with made-up data. It plays once when it scrolls into view, and you can pause it or step through.
Step 1 / 7 Two lists that should agree: the distributor sheet and the ERP. 37 differences, none explained.
What we built.
- Movement ledger with derived balances and full history
- Offline-first scanning app for warehouses and vans
- Blind cycle counts with approval thresholds
- Variance and write-off reports that link to the movements behind them
- Low-stock alerts and reorder suggestions
- Integration with the accounting system
Outcomes (targets).
These are the results we would aim for. They are not measured results from a named client.
- Target3 days to 6 hoursmonthly countthree warehouses
- Target2.4% to 0.3%unexplained variance by valueafter two cycles
- Target100%adjustments with a named approvernone without a reason
- Target11 / 11vans tracked as stock locationsload and delivery as transfers
Timeline and stack.
12 weeks, warehouse by warehouse
What we learned.
- A typed balance has no history. A ledger always does.
- Offline-first is not optional in a warehouse or a van.
- Thresholds and approvals turn counting from an argument into a record.
How we would deliver it.
Want this
for your business?
Message an engineer on WhatsApp or send a brief. We agree scope and price before work begins and build in small working steps you can test.