An inventory cycle count process should do more than tell you whether the shelf matches the system. It should help you find errors while they are still small enough to investigate, correct records safely, and prevent the same discrepancy from returning.
Yet many teams treat cycle counts as an occasional counting exercise. Someone prints a list, counts a few locations, and sends differences to a manager. By the time an adjustment is approved, stock may have moved again. A reliable process controls what happens before, during, and after the count.
Choose What to Count and How Often
Cycle counting means checking selected inventory throughout the year rather than relying only on a full physical count. It does not automatically replace a year-end count or other requirements your business may have. Its operational advantage is that you can investigate a smaller set of discrepancies closer to when they occurred.
Start by defining the count population: the items, locations, and inventory statuses included in your process. Be explicit about quarantined stock, returns, work in progress, consignment inventory, and goods in transit. If your team is unsure whether an item belongs in the count, you will get inconsistent results before anyone picks up a scanner.
Set a frequency based on risk, not convenience. An ABC approach groups items by importance, often using annual usage value. You might count high-value A items more frequently than lower-value C items, but value should not be your only criterion. Fast-moving items, products with frequent adjustments, and stock needed for critical orders may warrant more attention regardless of value.
A practical starting schedule could be:
Count group | Example selection rule | Starting frequency |
|---|---|---|
A | High annual usage value or critical supply | Monthly |
B | Moderate usage value and stable movement | Quarterly |
C | Lower usage value and low movement | Twice a year |
Exception group | Recent variances, returns, or high-risk locations | Weekly until stable |
These are examples, not universal targets. Adjust them to your transaction volume, available counting time, and the cost of an inaccurate record. Whatever schedule you choose, name an owner to maintain it and check that every eligible item or location receives its planned counts.
Prepare Each Count So the Result Is Trustworthy
The best counting method cannot compensate for an unclear snapshot. If receiving, picking, and transfers continue while a count is under way, your team needs a precise way to distinguish a real discrepancy from a transaction recorded at a different time.
Before the count starts, define the scope and the inventory position you will compare against. Depending on your system and operation, you may pause transactions for the affected locations, count during a quiet window, or record a timestamp and reconcile all movements that occur afterward. Do not assume that a system export and a physical count taken hours apart describe the same stock position.
Give counters clear instructions on units of measure and locations. A case of 12 recorded as one unit can create a large apparent variance. The same problem arises when stock sits in a temporary location, a returns area, or an unrecorded overflow bin.
A count assignment should include:
Scope: The item identifiers, locations, and inventory statuses to check.
Timing: The count window, snapshot time, and rules for transactions during the count.
Method: Whether the count is blind, meaning the counter cannot see the expected quantity, and how to record units of measure.
Evidence: The required quantity, location confirmation, notes, and any photos or supporting files.
Escalation: What to do if stock cannot be accessed, an item is unlabelled, or the count cannot be completed safely.
Blind counts can reduce the temptation to confirm the system quantity rather than count what is present. They also create more useful evidence when a variance needs investigation. If a second count is required, assign it to a different person where practical.
Investigate Variances Before Adjusting Inventory
A variance is a signal, not permission to change the system. First check whether the difference is caused by counting technique, timing, or a missing transaction. Only then decide whether an adjustment is needed.
Use a consistent investigation sequence:
Recount the item and location. Confirm the identifier, packaging, unit of measure, and any nearby or overflow locations.
Review recent movements. Check receipts, picks, transfers, returns, and adjustments around the count timestamp.
Identify the likely cause. Separate a recording delay from a physical shortage, mislocation, damage, or incorrect master data.
Assess impact. Consider quantity, value, customer commitments, and whether other items or locations may be affected.
Approve and record the correction. Apply the required approval rule, update the inventory system of record, and retain the reason and supporting evidence.
Set approval thresholds before discrepancies occur. For example, your policy might allow a supervisor to approve a small, explained adjustment while requiring a different approver for a high-value or unexplained loss. The right thresholds depend on your business, but the rule should be documented and consistently applied. For stronger control, the person who counts stock should not be the sole approver of their own adjustment.
If you find the same cause repeatedly, an adjustment is not a complete resolution. A receiving error may point to unclear purchase order and receipt steps. Repeated mislocations may call for better labelling or a change to put-away instructions. Use a corrective action process when the issue needs an owner, a due date, and a check that the fix worked.
Measure Accuracy and the Health of the Process
Inventory accuracy is useful, but one headline percentage can hide operational problems. Your team could report high overall accuracy while repeatedly missing the items needed for next week's orders.
Track a small set of measures with clear definitions:
Measure | How to calculate or assess it | What it tells you |
|---|---|---|
Count completion | Planned counts completed on time divided by planned counts | Whether the schedule is being followed |
Location accuracy | Locations with matching recorded and physical quantities divided by locations counted | Where records match the shelf |
Variance rate | Counts with a discrepancy divided by completed counts | How often investigation is needed |
Adjustment value | Total value of approved inventory adjustments | The financial scale of corrections |
Investigation time | Time from variance discovery to documented resolution | How quickly the team closes uncertainty |
Repeat variance rate | Items or locations with another variance within a defined period | Whether fixes are holding |
Define what counts as a match. Some teams need exact quantity agreement; others use documented tolerances for particular materials. If you apply a tolerance, still record the underlying difference so small, recurring losses remain visible.
Review measures by location, item group, and cause, not just as company-wide totals. If one receiving area produces most variances, increasing count frequency everywhere may consume time without fixing the source. The purpose of the dashboard is to direct action, not merely to display a better number.
Turn the Procedure Into Repeatable, Auditable Work
A written counting policy describes the standard. Your team also needs a reliable way to execute each count, assign follow-up, and see whether a discrepancy was resolved. This is where many spreadsheet-based programs become difficult to manage: the count sheet, approval message, inventory adjustment, and root-cause note live in separate places.
Create a reusable workflow with distinct stages for preparation, counting, variance review, approval, and closure. Give each stage an owner. Make the required evidence and decision points explicit, including what happens when a location is inaccessible or a variance exceeds the approval threshold.
In OKiDO, you can build that procedure as a versioned SOP template and launch a RUN for each count batch. Steps can be assigned to counters and reviewers, with structured fields for quantities, comments, file uploads, and approval gates. The resulting run records completion and decisions in one place, while your warehouse or inventory system remains the source of truth for stock balances and adjustments.
For more complex operations, a Decision Tree can guide the first variance investigation, and a System workflow can route different outcomes to recount, approval, or exception handling. This connects the procedure to its execution without pretending that a checklist alone controls inventory. For a broader view of who should approve changes and why, see our guide to internal controls for small businesses.
The strongest cycle count process does not aim to count everything more often. It makes each count trustworthy, resolves differences before they become routine, and uses the evidence to improve the work that created them.
If your counts are happening but follow-up is scattered, start with one item group and one location. Map the steps from count assignment through approved resolution, then run that process consistently. OKiDO can help you turn the procedure into assigned, auditable work and give your team a clearer record of what happened.
