WarehouseSOP
Menu

Cycle count procedure: how to run cycle counts, step by step

A cycle count procedure is how a warehouse keeps its inventory records true without shutting down for a full count: a small, scheduled slice of locations or items is counted blind every day, differences are recounted and researched, and only confirmed, approved corrections reach the system. This guide covers the counting methods, the steps, who does what, how to handle variances, and the mistakes that make a program report good numbers while the shelves disagree.

Cycle counting is inventory best practice, not an OSHA requirement. The safety of the work around a count is regulated, and the safety section names the sections that apply.

What cycle counting is

Cycle counting replaces, or reduces, the annual wall-to-wall physical inventory with small counts spread through the year. Every location or item is counted on a planned schedule while the building keeps shipping. Because each count happens close to the transactions that touched the location, a difference is easier to trace to its cause: a receipt keyed into the wrong unit of measure, a pick from the adjacent slot, a putaway that never got confirmed.

The output is two things. Corrected inventory, and a record of why it was wrong, which is the part that improves receiving, putaway and picking. A program that only produces the first is a write-off process with extra steps.

Cycle count methods

Most programs combine a schedule (what gets counted when) with blind counting (how it gets counted). Pick the schedule that matches what you need to be accurate.

ABC cycle counting

Items are ranked by annual usage value or pick frequency and cut into A, B and C classes. A items, the few that carry most of the value or movement, are counted most often; C items, the long tail, least often.

Use it when: You want count effort to follow the money and the movement, and you can keep the classes current as demand shifts.

Daily location counts

A fixed slice of locations is counted every day, often in a rolling sequence through the building, so every location is counted over a set period regardless of what is in it.

Use it when: Location accuracy matters as much as item accuracy, for example where pickers are directed to a slot by a WMS.

Blind counts

The counter is never shown the expected quantity, on the first count or on any recount. They record what is physically there and the system compares it afterward.

Use it when: Always. Blind counting is a discipline that sits on top of whichever schedule you use, because a counter who can see the expected number tends to find it.

Opportunity counts

A location is counted when an event makes it cheap and useful: a pick that takes it to zero, a low-quantity alert, a replenishment, or a pick exception.

Use it when: You want to catch errors at the moment they are easiest to see, alongside a scheduled program rather than instead of one.

The cycle count procedure, step by step

These steps follow the floor card of the Daily Cycle Counting template in the library.

  1. 01

    Generate the count list. The day's locations or items come from the schedule (ABC class, location sequence, or opportunity triggers). The list shows locations and items, never expected quantities.

  2. 02

    Freeze or coordinate activity at the location. Confirm nothing is being picked, put away or replenished at the location while it is counted, or record the cutoff so in-flight transactions can be reconciled.

  3. 03

    Count blind. Scan the location, scan the item, and count the actual units in the unit of measure the system holds. Never look up the system quantity first.

  4. 04

    Reach upper locations safely. If a location cannot be read from the floor, request the right access equipment and a trained operator. Never climb rack, pallets or product.

  5. 05

    Enter and confirm the count. Record the quantity once. The counter does not change a count after entry; a correction is a recount.

  6. 06

    Recount any variance blind. A count outside tolerance is recounted by a second person, also blind, before anyone researches it.

  7. 07

    Research the cause. Check open transactions, recent receipts and picks, adjacent locations, unit of measure errors and mislabeled product. Write down the cause found, or that none was found.

  8. 08

    Approve and post the adjustment. Someone other than the counter approves the adjustment, with its cause, before it posts. Large adjustments go to a higher approver.

  9. 09

    Keep the trail together. Count sheets, recount results, research notes and the approved adjustment are kept together, so the path from the physical count to the inventory change can be reproduced.

Roles in a cycle count program

The control that matters most is separation: the person who counts a location is never the person who approves its adjustment.

Counter
Counts blind, records what is physically there, flags damage, mixed product and bad labels, and never adjusts inventory.
Second counter
Recounts any variance blind, without seeing the first count or the system quantity.
Inventory control analyst
Builds the count schedule, researches variances, and documents the cause of each one.
Approver
A supervisor or manager who did not count the location approves adjustments, with a higher level for large ones.
Operations leadership
Staffs the plan, reviews coverage against plan and the trend in causes, and owns the procedure's review.

Variance thresholds and reconciliation

A variance is the difference between the counted and the system quantity. The tolerances are yours to set; what matters is that they are written down, applied the same way every time, and never loosened to make a report look better.

  1. 01

    Set a tolerance in units and in value. A count is accepted when it is within the unit tolerance and the value tolerance, whichever is tighter. A small unit miss on an expensive item should still trigger a recount.

  2. 02

    Tighten it by class. A items usually carry a tighter tolerance than C items. Write the tolerance for each class into the procedure so it is not decided count by count.

  3. 03

    Recount before research, research before adjustment. An out-of-tolerance count gets a blind recount first. Only a confirmed variance is researched, and only a researched variance is adjusted.

  4. 04

    Never widen a threshold to reduce recounts. A rising recount rate is information about the process. Loosening the tolerance hides it.

Reconciliation is the research and approval half. The Inventory Adjustment and Reconciliation template sets out how an adjustment is researched, approved and posted.

How often to cycle count

Frequency follows class. A widely used convention for an ABC program is to count A items monthly or more often, B items quarterly, and C items once or twice a year. That is common practice, not a regulation and not a benchmark: the right numbers are the ones that put count effort where the value and the movement are, and that the counters you actually have can complete.

Turn the class frequencies into a daily workload, check it against the people scheduled to count, and report coverage against plan next to every accuracy number. The ABC Cycle Count Program template walks through classification, frequencies and reclassification. When the records are too far off for cycle counts to recover, a Physical Inventory Wall-to-Wall Count resets the baseline.

Safety while counting

No regulation tells you to cycle count, but counters work in live aisles and reach upper rack, and that part is regulated. The library's cycle count templates cite these sections of 29 CFR 1910 for the work around a count:

  • 29 CFR 1910.22(a) Walking-working surfaces: general requirements
  • 29 CFR 1910.176(a) Handling materials, general: use of mechanical equipment
  • 29 CFR 1910.178(l)(1) Powered industrial trucks: operator training and evaluation
  • 29 CFR 1910.23(b) Ladders: general requirements for all ladders
  • 29 CFR 1910.23(c)(3) Ladders: additional requirements for portable ladders

In practice: keep aisles clear and passable while counting, never climb rack, pallets or product, inspect a ladder before each shift's first use, and count elevated locations only from equipment designed to lift a person, run by an operator trained and evaluated on that truck.

Common cycle counting mistakes

Showing the counter the system quantity.

Counts drift toward the number on the screen and the program measures nothing.

Adjusting without a cause.

Posting the variance and moving on erases the evidence of what went wrong upstream, in receiving, putaway or picking.

Letting the counter approve their own adjustment.

It removes the one control that separates a count from a write-off.

Reporting accuracy without coverage.

An accuracy number from a plan that was half completed describes the locations that were easy to reach, not the inventory.

Counting a whole class in one block.

Spreading each class across the period keeps the workload level and catches errors sooner.

Climbing to read a location.

The count is not worth a fall. Upper locations are scheduled with the access equipment booked.

Cycle count SOP templates and checklists

Each of these is a full SOP in the library, with a free sample PDF on its page. All of them are in the cycle counting templates.

Free printable checklists

Write it for your building

Answer a facility questionnaire once and the generator fits these templates to your locations, your equipment and your job titles. The output is a PDF, and the review stays with you.

14-day free trial, no credit card. Launch pricing.

Cycle count procedure questions

What is a cycle count procedure?
A cycle count procedure is the written process for counting a small part of a warehouse's inventory on a schedule, comparing it with the system, and correcting confirmed differences, so the whole inventory is counted over a period without shutting the building for a wall-to-wall physical inventory.
Does OSHA require cycle counting?
No. Cycle counting is inventory control practice and no OSHA standard requires it. The work around a count is regulated, though: aisles kept clear, floors kept free of slip and trip hazards, ladders inspected and used as designed, and any truck that lifts a counter operated by a trained and evaluated operator. The templates on this page cite those sections.
What is the difference between cycle counting and a physical inventory?
A physical inventory counts everything at once, usually with operations stopped. Cycle counting counts a slice every day or week while the building keeps running, and it finds errors close to when they happened, which makes the cause easier to trace.
How often should each item be cycle counted?
It depends on the item's class and on the counters you can staff. A widely used convention for ABC programs is A items monthly or more often, B items quarterly, and C items once or twice a year. That is common practice, not a rule; the right frequencies are the ones your staffing can actually complete.
What is a blind count?
A count where the counter is not shown the expected quantity. It keeps the count independent of the system, so a difference between the two means something.
Who should approve an inventory adjustment?
Someone other than the person who counted the location, usually a supervisor or inventory manager, with a higher approval level for large adjustments. The approval should carry the cause the research found.
What should a cycle count SOP include?
Scope, roles, how the count list is built, the blind count and recount rules, variance tolerances by class, how variances are researched and approved, safe access to upper locations, and the records kept. The daily cycle counting and ABC cycle count program templates in the library cover each of these.