Inventory theft in a shop or warehouse is stopped by eight software controls, in this order: a per-user login on every till, exception reports on voids, refunds and discounts, barcode-mandatory receiving and dispatch, a gate pass that exists only for an approved document, reason-coded stock adjustments with an approver and an owner alert, a damaged-goods register that produces the VAT-4.5 form the VAT Rules require, an in-transit state on transfers between branches, and cycle counts aimed at the few SKUs that carry most of the variance. None of them is a camera. The order matters because the evidence says most employee theft happens at the till, not in the godown.
According to BengalTech's 2026 price list, the till-side controls are in the Multi-Staff Outlet POS tier at Starts at $3,400, the warehouse controls in the Stock Tracker inventory tier at $5,900 – $9,500, and multi-branch transfers in the Multi-Warehouse tier at $12,500 – $21,000, one-time with unlimited users, built by our Dhaka team.
The scale of the problem is measured in the US, not here. The National Retail Federation's 2023 security survey (26 September 2023, 177 retail brands) put average shrink at 1.6 per cent of sales for 2022, up from 1.4 per cent the year before, or US$112.1 billion in total, with internal and external theft 65 per cent of it. No association or ministry publishes a comparable number for Bangladesh, so this guide does not invent one. What it does is take the two ECR Retail Loss studies on how staff actually steal and how stock records actually drift, and map each finding to the control that closes it and the tier of our price list it sits in.
Four numbers that decide the order of the list
- 36 / 29 / 27: the share of shrink US retailers attributed to external theft, employee theft and process failures in 2022, per the Loss Prevention Research Council's release of the NRF survey.
- 69 per cent: the share of dishonest employee events that happened at the till in the ECR offender study (4 retailers, 74 cases, 99 events); 16 per cent were in the warehouse.
- 59.54 per cent: the share of 233,000 audited SKUs at seven European retailers whose physical count did not match the system.
- 2 days: the time Rule 24B of the VAT Rules gives a registered business to apply in Form VAT-4.5 for goods unfit for supply.
The Eight Controls, What Each Stops, and Where It Lives
The theft methods in the second column come from Adrian Beck's ECR Retail Loss study of retail employee theft (19 August 2020), built on interviews with offenders and the security staff who caught them, and from the ECR study on inventory record accuracy (6 July 2020). The Bangladesh column is BengalTech's reading from the shops and warehouses we have scoped, not a survey.
| Control | What it stops | In Bangladesh | Included from |
|---|---|---|---|
| Per-user login and sale attribution | Cash and non-scanning theft at the till, which is where 69 per cent of the dishonest events in the ECR offender study happened, often behind a shared till PIN. | One shared login for the whole counter is the norm in the shops we scope. Every sale, void and refund is tied to the person logged in, and the login times out. | Multi-Staff Outlet POS |
| Exception reports on voids, refunds, discounts and no-sales | Refund fraud, void-after-payment and discount abuse. This is the "exception reporting" the ECR study names as the first control. | The manager password shouted across the floor is replaced by a per-user approval the owner sees by name the next morning. | Multi-Staff Outlet POS |
| Barcode-mandatory receiving and dispatch | Phantom stock. The ECR record-accuracy study found 59.54 per cent of audited SKUs did not match the system; a count typed off the challan is how the mismatch starts. | Supplier cartons are received by scanning, not by the number on the delivery challan, and short deliveries are logged against the supplier. | Stock Tracker |
| Gate pass tied to an approved document | Goods walking out of the warehouse with no paper behind them, or with paper that was never posted. | The guard scans a gate pass that exists only because a transfer order or invoice was approved in the system; an unmatched pass does not open the gate. | Stock Tracker |
| Reason-coded stock adjustments with approval and an owner alert | The "damaged" write-off that hides a theft, which the ECR study documents as deliberately damaging stock to retrieve it from the write-off area, and the process failures behind 27 per cent of shrink in the NRF survey. | Every manual change carries a reason, an approver and a timestamp, and the owner gets the day's adjustments by SKU and by user. | Stock Tracker |
| Damaged-goods register mapped to VAT-4.5 | A VAT exposure, not only a theft. Rule 24B of the VAT Rules requires an application within 2 days for goods unfit for supply, and the input tax on them is reversed. | The damage register produces the application and the adjustment entry, so a write-off cannot happen without the form that the Divisional Officer expects. | Stock Tracker |
| In-transit state on inter-branch transfers | Cartons lost between the Dhaka godown and the Chattogram branch, discovered at the next count. | The transfer deducts at source, sits in transit, and posts at destination only when the branch scans it in; the gap is flagged the same day. | Multi-Warehouse or Multi-Branch POS |
| Cycle counts weighted to the SKUs that carry the variance | The annual audit as the only check. At some retailers in the ECR study, 1.90 to 2.25 per cent of SKUs caused 70 per cent of the discrepancies. | The system schedules short counts of the high-variance and high-value SKUs weekly and leaves the slow shelf for the year-end. | Stock Tracker |
Why the till comes first: the offender study found only three instances of void manipulation but many of cash pocketed, items let through unscanned, staff discounts lent to friends and loyalty points collected on an idle till. A shared PIN makes all of them anonymous. Attribution is cheaper than any camera and it is what makes the exception report mean something.
Why the Excel Sheet Is the Weakest Control
A stock sheet has no user, no timestamp and no approval. A missing carton is a deleted cell, and the deletion is invisible until the annual count, which in most shops we scope is months after the person responsible has left. The record-accuracy study is the quiet half of the same problem: across 233,000 SKUs, 59.54 per cent were wrong, and correcting the records alone lifted sales by between 3.83 and 8.38 per cent, 5.98 per cent on average, because the shelf that the system believed was stocked was empty. That is money lost without anyone stealing anything.
The fix is not a better spreadsheet. It is a ledger where every movement is a posted document with a user on it, which is also what Rule 40(1)(bbb) of the VAT Rules has required for the purchase and sales books since 1 July 2025, and what Rule 95 means when it says electronic records must be preserved with proper security. For the full record-keeping rule set, see our NBR VAT compliance guide for POS and ERP.
The Rule That Turns a Fake Write-Off Into a VAT Problem
The most common way a theft is hidden in the ledger is a “damaged” adjustment. The VAT and SD Rules 2016 make that expensive to fake. Rule 24B says that when goods stored for supply become unfit for supply through damage, destruction by accident or any other reason, the registered person shall apply in Form VAT-4.5 to the Divisional Officer within 2 days. The officer enquires, values the goods, approves disposal within 30 days, and the business then reverses the input tax credit on those goods in its next return. Rule 24A does the same for purchased inputs that turn out to be unusable, through Form VAT-4.4.
So the damaged-goods register in the system is a compliance document, not only a control. Each entry carries the SKU, batch, quantity, a photograph, an approver and the input tax on the goods, and it produces the VAT-4.5 and the adjustment entry. A supervisor who used to write “damaged” in the margin now has to start a form addressed to the VAT office, which is the point. What the Rules do not fix, the exception report does: a write-off pattern by one user, one shift or one SKU stands out in a week.
What the Software Does Not Do
- Cameras at the tills that let the owner tie a till discrepancy to a person; the ECR study calls this the evidential link.
- Random staff searches that are genuinely random and cover every grade of staff, including managers.
- Locked cages for the high-value SKUs, with a log of who opened them and when.
- Telling staff what the reports can see. The study found that publicising the data-mining capability deters on its own.
These four come from the same ECR study and they are physical or managerial. We list them because a client who buys the software and skips them will still lose stock at the cage and in the car park.
What the Controls Cost
The tiers below are the ones on our POS service page and inventory software service page. Prices are one-time with unlimited users; hosting and an optional support contract are the ongoing costs. A single-counter shop gets stock tracking and receipts at the entry POS tier, but the attribution and exception reports that stop till theft start at the multi-staff tier.
| Tier | Controls it carries | Price |
|---|---|---|
| Multi-Staff Outlet POS | Per-user logins and sale attribution, exception reports on voids, refunds, discounts and no-sales, barcode checkout. | Starts at $3,400 |
| Multi-Branch POS | Everything above plus one stock ledger across branches, central prices, role-based access and in-transit transfers between outlets. | Starts at $5,900 |
| Stock Tracker (inventory) | Barcode-mandatory receiving and dispatch, gate pass, reason-coded adjustments with approval and alerts, damaged-goods register with VAT-4.5, targeted cycle counts. 6–10 weeks. | $5,900 – $9,500 |
| Multi-Warehouse (inventory) | Everything above plus in-transit transfers between warehouses, batch and expiry, landed cost. 14–20 weeks. | $12,500 – $21,000 |
For a super shop the same controls sit in the super shop POS guide; for a distributor with depots, the distribution ERP guide covers the transfer and landed-cost side; and the POS price guide explains what moves a shop between the tiers.
Sources
External references behind the figures and claims on this page. Rate bands and vendor pricing move — check the source before quoting a number.
- Shrink accounted for over $112 billion in industry losses in 2022, according to NRF report
National Retail Federation
National Retail Security Survey 2023 (26 September 2023, with the Loss Prevention Research Council): average shrink 1.6 per cent of sales in FY2022, up from 1.4 per cent in FY2021, US$112.1 billion in losses; internal and external theft nearly two-thirds (65 per cent) of shrink; 177 retail brands surveyed.
- 2023 NRSS release, 26 September 2023
Loss Prevention Research Council
The FY2022 cause breakdown of the same survey: 36 per cent external theft, 29 per cent employee theft, 27 per cent process and control failures and errors, 6 per cent unknown, 1 per cent other.
- Reduce employee theft in retail: what offenders exploit (Adrian Beck, University of Leicester)
ECR Retail Loss
Study of 4 retail companies, 74 cases and 99 dishonest events (19 August 2020): 69 per cent of incidents at the till and 16 per cent in the warehouse or back of store; methods including cash pocketed, items let through unscanned, staff-discount and loyalty-point abuse, shared till PINs, deliberately damaging stock to retrieve it from the write-off area, forged refund receipts and collusion with delivery staff; recommended controls led by exception reporting, CCTV at tills, genuinely random staff searches, security-cage procedures and publicising the data-mining capability.
- Measuring the sales impact of improving inventory records (Rekik, Syntetos and Glock)
ECR Retail Loss
Seven European retailers, about 233,000 SKUs in roughly 100 stores (6 July 2020): 59.54 per cent of audited SKUs did not match the system record; correcting the records lifted sales by 3.83 to 8.38 per cent, 5.98 per cent on average; at some retailers 1.90 to 2.25 per cent of SKUs caused 70 per cent of the discrepancies.
- Authentic English text of the Value Added Tax and Supplementary Duty Rules, 2016 (as amended to the 10 November 2025 gazette)
National Board of Revenue, Bangladesh
Rule 24B: goods stored for supply that become unfit through damage, destruction by accident or any other reason require an application in Form VAT-4.5 to the Divisional Officer within 2 days, an enquiry and valuation, approval of disposal within 30 days, and an increasing adjustment of the input tax credit in the next return; Rule 24A: unusable purchased inputs, Form VAT-4.4; Rule 40(1)(bbb), effective 1 July 2025: VAT-6.1, VAT-6.2 and VAT-6.2.1 preserved in ERP software; Rule 95: records preserved by fiscal year and electronic information preserved with proper security.
Frequently Asked Questions
How much stock does a retailer lose to theft and error?
The most recent National Retail Federation security survey with a cause breakdown, published 26 September 2023, put average shrink at 1.6 per cent of sales for 2022, up from 1.4 per cent, or US$112.1 billion across US retail, based on 177 retail brands. There is no equivalent published survey for Bangladesh; BengalTech does not publish a local figure because we do not have one we can source.
Is it staff or customers who steal?
Both, and a large share is neither. In the NRF survey, retailers attributed 36 per cent of shrink to external theft, 29 per cent to employee theft, 27 per cent to process and control failures and 6 per cent to unknown causes. The ECR Retail Loss offender study of 4 retailers found 69 per cent of dishonest employee events happened at the till and 16 per cent in the warehouse or back of store, which is why per-user tills and exception reports come before cameras in the list above.
How does a digital gate pass stop stock from walking out the door?
Goods cannot leave the warehouse without a gate pass, and a gate pass can only be generated from an approved transfer order or sales invoice already posted in the system. The guard scans the barcode; an invalid or already-used pass is refused. The manual chit that a supervisor signs on the way out, the loophole in most paper and Excel workflows, no longer exists.
What happens when someone marks stock as damaged or adjusts a quantity by hand?
The adjustment carries a reason code, needs an approver, and is logged with user, time and quantity; the owner receives the day's adjustments by SKU and user. For goods genuinely unfit for supply, Rule 24B of the VAT and SD Rules 2016 requires an application in Form VAT-4.5 to the Divisional Officer within 2 days, an enquiry, approval within 30 days, and reversal of the input tax credit in the next return. The register produces that form, so a write-off without the paperwork is not possible.
Can the system catch stock that goes missing between branches?
Yes. A transfer deducts stock at the sending branch and holds it in an in-transit state until the receiving branch scans it in. A carton that never arrives shows as a transit variance the same day, against the vehicle and the two users who dispatched and received, instead of surfacing at the next physical count.
What do these controls cost?
According to BengalTech's 2026 price list, the till-side controls are in the Multi-Staff Outlet POS tier at Starts at $3,400 and the multi-branch sync in the Multi-Branch POS tier at Starts at $5,900. The warehouse controls are in the Stock Tracker inventory tier at $5,900 – $9,500 (6–10 weeks) and the in-transit transfers in the Multi-Warehouse tier at $12,500 – $21,000 (14–20 weeks), one-time with unlimited users, built by our Dhaka team.
Tell us where the stock goes missing
A free scoping session maps your tills, godowns and branches to the controls above and gives you a fixed quote, with an honest answer if a per-user login on the till is all you need this year.