Warehouse inventory automation: why stock in 1C doesn't match the shelf

Warehouse stock accounting means keeping a record in the accounting system of how much of each product sits in each storage location, so that the figure in the software matches what is on the shelf. Automating it turns every movement of goods into a document that is created at the moment of the operation and confirmed by scanning a barcode. That covers receiving, transfers, picking, shipping, returns and write-offs. Stock in 1C, the accounting and ERP platform most Russian companies run on, drifts away from the shelf because of specific steps: receiving without scanning, item mix-ups, reservations nobody releases, and backdated returns and write-offs. Barcodes and handheld data terminals close some of these steps; the rest is fixed with accounting rules and by linking warehouse operations to 1C.

Below we break down the levels of stock automation and where exactly the numbers diverge. We show separately how to use a measurement to find the step where the error starts, and which of the problems you find barcodes and handheld terminals fix and which they don't. Next comes inventory counting without stopping the warehouse: when to count everything at once and when to count by zone. Then we look at when handheld terminals or a bot on top of 1C are enough for accounting and how to calculate what discrepancies cost you. At the end: what to prepare before you start, and why an off-the-shelf product covers only part of the job. What a WMS is and when a warehouse has outgrown 1C is covered in WMS: what it is and when 1C is enough.

What is warehouse stock automation?

Stock automation means every movement of goods is recorded by a document created at the moment of the operation, not by a delivery note the storekeeper types into the system in the evening. An employee scans the product and the location, and the system updates the stock itself and shows it to sales right away. There are several levels of automation, and each closes its own share of discrepancies.

Discrepancies are far more common than people tend to assume. Nicole DeHoratius and Ananth Raman studied nearly 370,000 inventory records in 37 stores of one US retailer and found 65% of them inaccurate (Management Science, April 2008, vol. 54, no. 4). The study was done in retail, but the mechanics are the same as in a warehouse: the system record is changed by documents, while the shelf is changed by hand.

A recent measurement shows the same picture. In the ECR Retail Loss report (dated December 2025, published February 9, 2026), professors Christoph Glock, Aris Syntetos and Yacine Rekik interviewed 25 retail practitioners and experts from different countries. According to their findings, more than 60% of inventory records contain inaccuracies, and only about 35% are fully correct (ECR Retail Loss, “Inventory accuracy as a capability”).

Why doesn't stock in 1C match what's on the shelf?

Stock diverges at the steps where goods move physically while the document is created later, from memory, or not at all. A typical warehouse has six or seven such steps, and each leaves its own trace in the records. The error is rarely random: the same places produce discrepancies month after month.

The ECR report quotes one retailer: when they dug into it, they found about twenty different causes of inaccurate stock records. DeHoratius and Raman have an observation that is useful for warehouses. Differences between product categories accounted for 26.4% of the variation in inaccuracy, while differences between stores accounted for only 2.7%. So the first place to look is product types and operations, not “which warehouse is the bad one.”

Three rows of the table deserve a separate discussion, because a scanner doesn't fix them. A reservation holds goods for an order, and if cancelling the order isn't linked to releasing the reservation, the manager sees an empty warehouse in front of a full shelf. A return physically arrives before the document: the goods are put back in place, and posting the return is forgotten or put off until a decision on the defect is made. A backdated write-off changes stock for past days. Everything sold and reserved on those days was counted against the wrong figure.

The order-based scheme is a common source of phantom discrepancies. In 1C configurations for trade, a warehouse can work with warehouse orders: goods are first received into the warehouse with a receipt order, and the purchase receipt document is processed by accounting or the buyer. Until the documents match, the goods have two stock figures: the warehouse one and the accounting one. The warehouse and sales argue, even though each is looking at its own correct number.

How do you find the step where the numbers diverge?

The step is found from facts; gut feeling won't help here. You take a sample of items, recount them by hand, and for each discrepancy use the document history to trace when the recorded stock split from the shelf. After three or four dozen items have been analyzed, the causes start to repeat.

How to run the measurement:

1. Sample. Thirty to fifty items from different groups: fast and slow movers, piece goods and weighed goods, the ones sales complain about. If you take only the problem items, the picture will look worse than reality.
2. Recount. Two employees count the sample at the start of the shift, before the first shipments, and note the storage location. That way the recount doesn't get mixed up with goods in motion.
3. Comparison. The recount is checked against stock in 1C at the same moment, with free stock and reservations looked at separately.
4. Document history. For each item with a discrepancy, review the movements over the last month: receipts, shipments, returns, transfers, write-offs. Look for the document after which the figure stopped making sense.
5. Cause. Next to each discrepancy, record the step from the table above. If the cause can't be established, that is a finding too: the movement happened without a document.

The result of the measurement is a table of “step — number of discrepancies — value.” It shows where to start. If half the errors happen at receiving, you buy scanners for receiving first. If half are in reservations, the scanners can wait: first you fix the reservation release rule and its link to orders.

A common mistake is to start with hardware because it is easier to buy. Terminals speed up warehouse work, but if the errors originate in documents, a month later the discrepancies will be in the same places. How to clean up the master data before connecting scanners is covered below, in the section on preparation.

What do barcodes and handheld terminals fix, and what don't they?

Barcodes and handheld data terminals fix errors that happen at the moment of a physical action: the wrong product, the wrong quantity, the wrong bin. Errors in how documents are processed, such as stuck reservations, returns without a status and write-offs with past dates, are untouched by scanning. Those need rules in the accounting system.

A scanner works only with clean master data. If a product has no barcode, two items share one code, or one item has three cards in the product catalog, the terminal will make mistakes just as confidently as a person. That is why duplicate products, units of measure and package barcodes are fixed before launch.

The second limit is discipline. Scanning works when people always scan. A single “I'll mark it later” during a busy shift brings the discrepancies back, and a month later sales stop trusting the stock figure again. So the procedure states that no movement is posted without a scan, and that rule has an owner.

How do you run an inventory count without stopping the warehouse?

The warehouse keeps running when a full count is replaced by cycle counting throughout the year: every day or week, a small zone or product group is counted while shipments continue. A full count remains for annual reporting and for cases when there are so many discrepancies that counting in parts makes no sense.

The mandatory inventory count is set by FSBU 28/2023 “Inventory Count,” a Russian federal accounting standard approved by Order No. 4n of Russia's Ministry of Finance dated January 13, 2023. Paragraph 15 lists preparing the annual financial statements among the mandatory cases. Paragraph 16 allows inventory for them to be counted no earlier than October 1 of the reporting year (FSBU 28/2023, website of Russia's Ministry of Finance). The timing of counts is set by the company's head. Cycle counting during the year does not replace the annual count, but it means the warehouse reaches December without surprises.

A cycle counting schedule is built around the value and movement of goods:

1. Fast-moving and expensive items. Counted most often, for example once every week or two: an error on them costs more and hits sales faster.
2. Mid-range items. Once a month.
3. Slow and cheap items. Once a quarter.
4. Unscheduled counts triggered by events. An item went negative, sales couldn't find the product, a bin turned out to be empty during picking. A signal like that sends the item for a recount the same day.
5. Items from the measurement. Groups where the measurement showed the most discrepancies are counted more often until the cause is removed.

Counting during working hours requires one rule: while a zone is being counted, no movements are posted for it, or they are recorded separately. Otherwise the count gets mixed with shipping, and the adjustment introduces a new error instead of fixing the old one.

When are handheld terminals or a bot on top of 1C enough, and when do you need a WMS?

Handheld terminals or a bot on top of 1C are enough when there is one warehouse, a modest number of operations per shift, and the discrepancies originate at receiving, returns, reservations and recounts. A WMS is needed when the bottleneck is the movement of goods itself: many people, thousands of bins, batches and expiry dates, picking routes, several warehouses.

A messenger bot suits simple operations. The employee picks an operation, scans with the phone camera or enters the code and quantity, and the bot checks the data and creates a document in 1C. For receiving, returns and recounts in a small warehouse or retail outlet, that is often enough. Our solutions catalog includes a stock-tracking bot that records sales, returns and stock right at the point of sale via Telegram.

The signs of a warehouse that needs a full WMS, and the cases where it is overkill, are covered in WMS: what it is and when 1C is enough. The only conclusion here: the decision is made after the measurement. If errors originate in document processing, a WMS will carry them over into the new system.

How do you calculate what discrepancies cost you?

The cost of discrepancies is calculated under three headings. The first is lost sales: goods exist in the records but not on the shelf. The second is unnecessary purchases of goods that are in stock but not visible in the records. The third is staff time spent searching, recounting and reconciling. The total is compared with the automation budget.

A hypothetical example:

1. Lost sales. Twenty times a month a manager confirms an order, and at picking the goods aren't there. If half of those customers go to a competitor and the average order is forty thousand rubles, that is about four hundred thousand rubles of revenue a month.
2. Unnecessary purchases. Goods worth three hundred thousand rubles sit on the shelf but are recorded under someone else's SKU or reserved for a cancelled order, and they get bought again. The money is frozen in inventory.
3. Time. The storekeeper and the manager together spend an hour a day on “do we have it in the warehouse” calls and hunting for goods. That is more than twenty hours a month.
4. Inventory counts. A day of full counting when the warehouse doesn't ship, plus reviewing hundreds of discrepancies afterwards.

Plug in your own numbers: you know your average order and purchase amounts, and the number of failed picks becomes visible as soon as the warehouse starts recording them. Normal values differ for every warehouse, so it is worth comparing yourself with yourself before and after changes.

The effect of correcting stock records was measured in a field experiment cited in the ECR Retail Loss report. Stores where stock was audited and inaccurate records were corrected were compared with control stores. Sales rose by 4 to 11% (ECR Retail Loss, “Inventory accuracy as a capability”). The mechanics are clear: a system that can see the goods reorders them on time and lets you sell them. How to put stock and losses on a single screen for the head of the business is covered in “The owner's dashboard”.

Where do you start?

Start with a measurement and a description of how goods actually move through the warehouse. They show which steps a scanner closes and which need rules and a link to 1C. The company does most of the preparation itself, and this list makes it immediately clear how much work lies ahead.

1. Measuring discrepancies. A sample of items, a recount and a review against the document history, as described above. Without it, any decision is discussed on gut feeling.
2. Describing how goods move. Receiving, put-away, picking, shipping, returns, write-offs, transfers: who does it, with which document and at what moment. Write down how things work now, including operations agreed on verbally.
3. Clean master data. A product catalog without duplicates, units of measure, barcodes on products and packages. If the warehouse uses bins, you need a bin layout with codes.
4. Rules for what a scanner doesn't close. Reservation lifetime and who extends it, return statuses, a ban on posting documents in a closed period, and the procedure for adjusting stock after a recount.
5. Access to the accounting system for data exchange. A separate 1C user account with the right permissions and a copy of the database for testing the exchange before launch. How links between systems work is covered in our piece on system integration.
6. Access rights. Who can post write-offs and adjustments, who sees purchase prices, who can release other people's reservations. A stock adjustment without a responsible person turns a recount into a way to hide a shortage.
7. An owner after launch, and support. An employee who looks after the counting schedule, master data and discrepancy reviews, and a support contract for the 1C exchange. The product range, suppliers and warehouse layout change, and without an owner the rules will drift away from practice within six months.

The company handles the first three points on its own, and they alone remove some of the discrepancies. The remaining four show what drives the project price: how many operations need to be linked to 1C, what state the master data is in, and how many exceptions the process has.

Off-the-shelf product or a system built around your process?

Off-the-shelf solutions handle the standard part well: receiving and picking with a scanner, recounts from a terminal, label printing. Discrepancies in reservations, returns, write-offs and the link between the warehouse and sales are different at every company, because each has its own procedures, master data, access rights and exceptions.

At the scanner level, warehouses barely differ: goods have to be received, put away and picked without errors. It makes sense to take this part ready-made. For a small warehouse with tidy master data, linking handheld terminals to 1C is often enough for a long time.

Where companies differ is in the rules around stock. One holds a reservation for three days, another until payment, a third until the commercial director decides. One returns defective goods to the supplier, another marks them down and sells them. One runs the warehouse on the order-based scheme, another without it, and each has its own rule for when stock counts as available for sale. In an off-the-shelf product, these rules have to be bent to fit someone else's data model: with customizations that break on updates, or with manual reconciliations that fall on staff. We covered the choice between a ready-made solution and a system built around your process separately in custom system or off-the-shelf.

Two projects show how this looks in practice. In a project for a wholesale and manufacturing company, orders came in by email and messengers, managers checked stock by phoning the warehouse, and reservations got lost. That led to item mix-ups and failed shipments. After launch, an order sees real stock immediately and places a reservation, data exchange with 1C runs both ways, and shipments stopped failing because of mix-ups. In a project for a retail chain, stock tracking covering receipts, sales, returns and stock levels runs automatically, and sales and returns are recorded by a Telegram bot.

INCUBE AI builds stock accounting as custom development: receiving, recounts and returns from a terminal or through a bot, reservations under your rules, two-way exchange with 1C, and a discrepancy report with named owners. For standard areas there is a solutions catalog, and each solution is adapted to the client's way of working. We work under contract, data is stored in Russia, and the system remains under support after launch. If your stock doesn't match the shelf, tell us about your task: we will review a sample of your items, find the steps where discrepancies start, and estimate the scope of work from them. How sales see stock and reservations on the way from inquiry to payment is covered in “Sales automation”.

Sources and links

• DeHoratius N., Raman A. Inventory Record Inaccuracy: An Empirical Analysis. Management Science, April 2008, vol. 54, no. 4, pp. 627–641 — nearly 370,000 inventory records in 37 stores of one retailer, 65% of which were inaccurate. Differences between product categories account for 26.4% of the variation in inaccuracy, differences between stores for 2.7%.
• ECR Retail Loss. Inventory accuracy as a capability: building maturity and customer value — by Christoph Glock, Aris Syntetos and Yacine Rekik. The report is dated December 2025 and was published February 9, 2026. Twenty-five interviews with retail practitioners and experts: more than 60% of inventory records contain inaccuracies, about 35% are fully correct. One retailer found about twenty causes of inaccuracy. In a field experiment, correcting stock records raised sales by 4 to 11%.
• FSBU 28/2023 “Inventory Count,” website of Russia's Ministry of Finance — approved by Order No. 4n of Russia's Ministry of Finance dated January 13, 2023. Paragraph 15: a mandatory inventory count when preparing annual financial statements. Paragraph 16: inventory for the annual statements may be counted no earlier than October 1 of the reporting year.