Rolling inventory: counting is not deciding what to count in 2026

Blocking the warehouse for several days to count all the stock at once is expensive, but splitting the counts without knowing what to count each week doesn't solve anything either.

What a rolling inventory is, and what it is not

A cycle count involves physically counting a portion of the stock at regular intervals, rather than counting everything at once. Each pass focuses on a subset of items: a product family, a warehouse location, or a list of SKUs chosen for a specific reason. Operations continue while the count is being performed because only a fraction of the stock is held up at any given time.

This is not a magic bullet that replaces daily stock control. It is a verification tool, just like the stock management As a whole, it serves to confirm or correct the theoretical quantities recorded in the system. A cycle count is also not a simple mechanical division of the stock into equal parts: dividing the warehouse into twelve parts and counting one per month only makes sense if this division is based on management logic, not on an arbitrary schedule.

It's also important not to confuse cycle counting with a one-off check triggered after an incident. A one-off check addresses a specific doubt about an item after a suspected theft or a reported error. Cycle counting, on the other hand, operates on a recurring schedule, decided in advance, independently of any particular incident. Its value lies precisely in its regularity: it allows for the comparison of the same item on several different dates, something a single check can never do.

distinguish it from the annual inventory

The annual inventory counts the entire stock in a single operation, generally at the end of the fiscal year, to establish a reliable stock value at a given date. This operation often brings commercial or logistics activities to a standstill during the counting process, because stock movements must be frozen to ensure the figures remain consistent. A medium-sized warehouse may therefore see its shipments suspended for several days.

Cycle counting works in reverse. It breaks down the count into regular passes, each limited to a portion of the stock, without ever interrupting the flow of incoming and outgoing goods. A single pass can last an hour, cover twenty items, and take place on a Tuesday morning without anyone else in the company noticing. The drawback is that no single pass provides a complete view of the stock at any given moment: several passes over a period are necessary to cover all items.

The two methods do not serve the same end goal, even though they both involve the physical counting of stock. The following table summarizes the key differences.

The annual inventory fulfills the requirement for accurate accounting at a fixed date. The rolling inventory addresses the need for continuous monitoring, which is more useful for managing day-to-day operations than for closing a fiscal year.

Why splitting the counting isn't enough: you also have to choose what to count.

Splitting a count is only valuable if the division is based on a clear logic. Counting ten items at random each week, without any connection to their actual importance to the business, is like taking an annual inventory spread over twelve months, with the same fundamental flaw: no priority is established between items that deserve frequent attention and those that can wait several months without risk.

The real question posed by cycle counting is therefore not how to organize the counts, but what to count this week rather than another. It's a management decision, not a logistical problem. It requires knowing which items move frequently, which have high financial value, and which items, if counted incorrectly, would have immediate consequences for sales or production.

No tool can independently determine whether an item deserves to be counted weekly rather than annually. This decision rests with the person managing the inventory, because they alone know the commercial context of each item: an upcoming promotion, a supplier with a delayed delivery, or an item that has caused problems in the past. A system can display quantities, calculate discrepancies, and maintain a history. However, it cannot determine, in place of a human, which item is a priority this week.

This is what distinguishes a truly useful cycle count from a simple, repetitive counting exercise done without thought. Dividing the work into small, regular chunks is merely a mechanical process. Selecting the items within each chunk, however, remains a decision that requires knowledge of the business and the inventory.

Three criteria for deciding which items to count often

Three criteria allow us to prioritize items without resorting to a named method or automatic sorting. The first criterion is movement: items that frequently enter and leave the warehouse accumulate discrepancies more quickly, simply because they are handled more often. A reference that moves every day statistically has more opportunities to generate a data entry error than a reference that remains stationary for months. This is directly linked to the stock rotation : the faster a reference circulates, the more it justifies frequent use.

The second criterion is financial value. An item that is expensive per unit or that represents a significant portion of the total inventory value warrants more frequent monitoring, because a discrepancy in that item has a substantial impact on the accounts, even if its quantity is small. Ten missing units of a low-value item make almost no difference to the balance sheet. Ten missing units of an expensive item, however, can represent a significant sum.

The third criterion is operational risk, that is, the consequence of a discrepancy not detected in time. An item nearing its reorder point, or whose stockout would halt significant production or sales, justifies an early reorder even if its unit value is low. This is where the concept of minimum stock : an item whose stock is close to this threshold must be checked more often than an item that has a comfortable margin.

These three criteria are not mutually exclusive: an item can have a high turnover, a high value, and a significant risk of stockout, making it an absolute priority for counting. Other items do not meet any of these criteria and can remain outside of frequent circulation without causing any problems.

This table highlights an important distinction: counting by turnover and counting by risk of stockout do not apply to the same items. A high-turnover item may never come close to being out of stock because it is regularly replenished in large quantities. Conversely, a low-turnover item can be at risk of being out of stock if its supplier becomes unreliable. Both approaches must coexist in the order scheduling, without replacing one another.

Frequency of passage: why it shouldn't be the same for all articles

Applying a single frequency to all inventory ignores the criteria presented above. A company that counts each item once a month, without distinction, treats a strategic item with daily turnover the same as a secondary item that has been sitting on the shelf for six months. The time spent on this second item is wasted time that could be used to more closely monitor the first.

An appropriate inspection frequency stems directly from the three prioritization criteria. High-turnover, high-value, or out-of-stock items warrant more frequent inspections, sometimes weekly. Stable, inexpensive items with no availability concerns can be inspected quarterly or even semi-annually. This hierarchy is not fixed: an item can change category depending on the season, a sales promotion, or the emergence of a problem. sold out recurring on a previously quiet reference.

The time available for the counts must also be taken into account. An overly ambitious scan, attempting to cover too many items at once, will ultimately be rushed or postponed. It's better to aim for a realistic number of references in each scan, even if it means multiple scans of priority items, rather than striving for exhaustiveness at the expense of accuracy.

Therefore, frequency is never a fixed value decided once and for all. It is revised as knowledge of the inventory improves, as new discrepancies appear, or as business activity evolves. A schedule of inspections that never changes is a sign that it was probably never based on a genuine analysis of priorities.

An isolated deviation versus a recurring deviation: how to read the history of a reference

A discrepancy between the theoretical quantity and the counted quantity does not have the same significance depending on whether it occurs once or repeatedly. A single discrepancy, corrected during the count, can have many benign explanations: a typing error, a package counted twice by mistake, a delivery recorded one day too late. These errors happen, they can be corrected, and they do not warrant any particular alarm if they do not recur.

The situation changes drastically when the same item shows a discrepancy on the next pass, and then again on the following one. This is no longer a one-off error; it's a warning sign. This sign could point to a mislabeled item, confused with another on each receipt. It could reveal repeated, discreet but regular theft. It could also indicate a systematic error in the receiving or shipping process, generating the same type of discrepancy with each cycle.

The only way to make this distinction is to review the history of a reference over several passages, not just to correct each discrepancy as it appears. Correcting a discrepancy without comparing it to previous ones is like treating a symptom without ever looking for the cause. This is precisely where the rolling inventory becomes essential: its regularity allows for this comparison over time, something a single annual inventory could never achieve on the same scale.

In this example, the first pass shows no discrepancies: nothing to report. But from the second pass onward, a negative discrepancy appears and recurs, with a slightly increasing magnitude each time. A single one of these discrepancies, taken in isolation, could have been classified as a simple counting error. Placed side by side, they reveal a pattern that warrants investigation: a problem of unreported breakage, recurring theft, or a stock removal error that is repeated with each sales cycle.

This information must be included on a counting sheet

A counting sheet serves as a working tool during the inspection and then as a record for subsequent analysis. To be useful, it must contain a minimum of information; without it, comparison between inspections becomes impossible. The article reference is fundamental: without a unique and reliable identifier, no comparison over time is meaningful.

The theoretical quantity, as it appears in the system at the time of counting, must always be entered on the record before counting begins. It serves as an immediate point of comparison with the quantity actually observed in the field. The counted quantity, entered during the count, completes this pair of data. The discrepancy, calculated as the difference between the two, must be visible on the record itself rather than reconstructed afterward, to avoid transcription errors.

For the items concerned, the information sheet must also include the batch number and the expiry date, two essential pieces of information as soon as a product is subject to a requirement of batch traceabilityWithout this information, a discrepancy observed on an overall reference may mask a more specific problem: an expired batch still in stock, or a recent batch counted in error instead of an older batch that should have been released first.

The table below illustrates an extract from a tally sheet covering seven items, with two actual discrepancies on the seven lines.

In this extract, five out of seven items show no discrepancies: the theoretical stock corresponds exactly to the observed reality. The two remaining discrepancies, one negative and the other positive, are those that warrant immediate attention, and above all, those that must be identified in the next section to determine if they recur with the same items.

Conducting a passage without interrupting the activity

The main practical advantage of cycle counting, compared to annual inventory, lies in its ability to be carried out without disrupting operations. However, this requires certain precautions during the counting process itself. The first is to temporarily freeze stock movements on the items being counted, and only on those items, for the duration of the count. If a stock entry or exit occurs for an item being counted, the measured discrepancy becomes impossible to interpret correctly.

The second precaution is to choose a time of low activity for the inventory check, without necessarily isolating it completely from the rest of the organization. A check performed early in the day, before order processing begins, limits the risk of competing movements on the items being counted. This constraint remains minor since the check only involves a fraction of the stock: the rest of the warehouse continues to operate normally.

The third precaution is to immediately document any unusual movement that occurs during the count, rather than ignoring it. An urgent receipt or an unexpected issue involving an item being counted must be noted to avoid skewing the discrepancy calculated at the end of the count. An apparent discrepancy explained by a documented movement is not a true discrepancy, and it would be a mistake to confuse it with a genuine warning sign.

These precautions remain easy to implement because the scope of a cycle is intentionally limited. It is precisely this limitation that makes cycle counting compatible with continuous operation, unlike annual inventory, which requires a complete shutdown to ensure the accuracy of the count.

Who is responsible for the inventory, and why is it never the same person managing it on a daily basis?

The person who manages the stock on a daily basis knows the locations, storage habits, and sometimes even the small inaccuracies that have developed over time. This knowledge is invaluable for organizing a stock run, but it becomes a drawback when it comes to counting, because it introduces confirmation bias. A person accustomed to believing that a particular location contains a certain quantity tends to see what they expect to see, rather than what is actually present.

Assigning the counting to someone other than the person who manages the stock daily reduces this bias. This person has no preconceived notions about what they should find and therefore counts what is physically present more accurately. This separation of roles doesn't require a dedicated inventory team: in a small organization, it may suffice to swap counting areas between two employees, each counting the items usually handled by the other.

This separation also serves as an internal control mechanism, regardless of any unintentional errors. If a discrepancy stems from theft or fraudulent manipulation, it is easier to detect if the person in charge is not the one with daily access to the physical inventory and recorded transactions. This is not a matter of systematic suspicion, but a common-sense precaution that protects both the company and the individuals who manage the inventory on a daily basis.

In smaller organizations, where a single person manages all the inventory, this complete separation isn't always possible. In such cases, it's still useful to have a second person review the inventory sheets, even without specific inventory expertise, simply to provide an outside perspective on any discrepancies before validating them.

Can rolling inventory and annual inventory coexist?

The two methods are not competing; they address different needs and can perfectly coexist within the same company. An annual inventory remains necessary to establish a reliable stock value at the end of the fiscal year, an accounting requirement that does not disappear simply because a cycle counting system is implemented concurrently during the year.

Cycle counting, conducted regularly throughout the year, reduces the workload at the time of the annual inventory. If discrepancies have already been detected and corrected on priority items during the regular checks, the year-end inventory is based on a largely reliable stock. The time required for a complete count is therefore reduced, because there are fewer surprises.

This coexistence also has a direct impact on the decisions of replenishment These discrepancies are identified throughout the year. A discrepancy detected and confirmed by a cycle count can trigger a corrective order before the situation escalates into a stockout, whereas a single annual inventory would only reveal the problem several months later, once the damage is done. This is a strong argument in favor of a cycle count, even in a company that continues to conduct a full annual inventory.

The coexistence of the two methods does not mean they must be organized by the same people, nor follow the same schedule. The annual inventory can remain a one-off exercise, prepared in advance and requiring more resources over a short period. The rolling inventory, on the other hand, is integrated into the normal rhythm of the business, without any exceptional mobilization of resources.

A concrete example: a discrepancy that appears three times on the same reference

Let's take the example of a small spare part, regularly sold by a company that repairs household appliances. This part has been identified as a priority for cycle counting due to its high turnover, and is therefore checked every two weeks. During the first check of the year, the quantity counted exactly matches the theoretical quantity. Nothing to report.

Two weeks later, the second count revealed a discrepancy of three units. The person in charge of counting noted the difference on the form, adjusted the quantity in the system, and attributed it to a possible data entry error during a recent sale. This is a plausible explanation, and nothing yet warrants a thorough investigation for such a large, isolated discrepancy.

On the third pass, two weeks after the previous one, the discrepancy reappeared for the same item, this time by four units. At this point, the explanation of an isolated data entry error became less convincing: it was unlikely that two independent data entry errors would affect the exact same item two passes in a row. The person responsible for inventory began to examine the recent movements of this item more closely, without yet formally alerting the team.

The fourth pass confirms the trend with another discrepancy, again negative. Three consecutive discrepancies for the same part, all in the same direction, can no longer reasonably be attributed to independent data entry errors. The company then decides to directly observe the storage location of this part between two passes and discovers that the label of the adjacent location, which contains a visually very similar part, is incorrectly positioned. The two part numbers are regularly confused during order preparation, which explains the regular leakage observed for one and, logically, the excess that should appear for the other.

This case perfectly illustrates the principle defended from the beginning of this article: an isolated discrepancy says nothing in particular, but a discrepancy which recurs on the same reference, passage after passage, eventually reveals a precise cause, here a labeling error which would never have been detected by a single annual inventory, lost in the mass of other discrepancies noted once a year.

Common mistakes and how to correct each one

Some causes of discrepancies recur more frequently than others, and each is identifiable by a specific signal in the count history. Knowing how to associate a type of signal with a probable cause significantly speeds up the correction process, rather than restarting the investigation with each new discrepancy.

A labeling error can be corrected by physically rechecking the location of the two affected items and then affixing a clearly legible label. Theft or markdowns require a more organizational response: increasing surveillance of the affected area or reviewing access permissions for that section of the warehouse. Systematic errors upon receipt are corrected by going directly to the supplier or reviewing the internal incoming goods control process, often by comparing the delivery note with the quantity actually unloaded.

Unreported breakage requires clear instructions for the teams: any breakage observed must be reported immediately in the system, rather than being discovered several weeks later during a count. Repeated data entry errors, often linked to a poorly understood interface or an incorrectly adopted habit, are corrected by reminding the person concerned of the procedure. Finally, confusion regarding units of measurement, frequent with items sold both individually and in batches, is corrected by clarifying the reference unit used in the system, to avoid any doubt during data entry. These discrepancies also impact the inventory valuation presented in the accounts, which gives an additional reason to process them at the source rather than simply correcting the quantity shown.

Djaboo and the rotating inventory: what the tool allows, and what it doesn't do

In Djaboo, a profile ofphysical inventory It focuses on a chosen warehouse, on a chosen date, and can cover the entire stock or only a portion of the items, depending on what is added. For each line, the counted quantity is entered, and the tool displays the theoretical quantity at that time and then automatically calculates the difference between the two. Each line can include a batch number and an expiration date. There's nothing stopping you from creating several successive records over time, each covering a different subset of items: this is the technical basis of a cycle counting system.

However, one limitation must be clearly stated: Djaboo does not categorize any items by criticality, does not schedule any inspections, and does not generate any automatic item lists starting this week. The grouping of items, the choice of inspection frequency, and the inspection order remain entirely at the discretion and discretion of the person managing the inventory, item by item. This limitation is unavoidable and will not be addressed by any future update announced here.

This limitation, however, aligns perfectly with the proper use of the tool, provided that the decision remains human. The practical approach involves deciding in advance which items warrant frequent inspection, based on the turnover, value, and risk criteria mentioned above. Once this list is established, simply create an inventory record for this specific subset on the scheduled inspection date, and then enter the quantities counted in the field as they are collected.

The next, equally important step is to review the discrepancy displayed by the tool by comparing it to previous records for the same item. This comparative review, performed by the person responsible for inventory, allows them to distinguish an isolated error from a recurring problem, exactly as in the example described above. Djaboo provides the figures and calculates them correctly for each record. However, it does not replace the human perspective that puts these figures into context from one inventory cycle to the next, and it is precisely this perspective that gives cycle counting its true meaning.

Regular use of these cards, organized according to a pre-planned schedule, transforms a simple counting tool into a real system for tracking stock over time, without ever claiming to automate the priority decision which remains, and will remain, on the company's side.

The division of counting into regular passes never replaces the decision of what to count and how often: this decision remains, from beginning to end, a management responsibility that no system can assume in place of the company.

And above all, a discrepancy which recurs on the same reference, passage after passage, always deserves more attention than an isolated discrepancy corrected without rethinking it: it is in this comparative rereading, and not in the simple repetition of counting, that the rolling inventory finds its true usefulness.

The field count is conducted with a inventory sheet ready to fill.