Count what is actually on the shelf without letting the expected balance lead the answer. Use two linked sheets to investigate differences before anyone changes stock records.
A count of 18 is not useful if nobody knows whether it means boxes or individual pieces, which bin was counted, or whether someone issued stock halfway through. A cycle count sheet needs to preserve those facts—not just calculate a difference.
This complete inline resource is for small internal stores holding ordinary consumable supplies. Copy the packet into a document and the review ledger into a spreadsheet. It covers one selected counting round, not inventory valuation, a statutory stocktake, or controls for hazardous, regulated or safety-critical materials. The roles, timing and review rules below are a proposed operating design to adapt locally.
Set the count boundary before printing a sheet
Give each round a unique session ID and select a small, physically identifiable area. Include every item found inside that boundary, including unexpected stock. Specify whether staging, damaged stock and goods awaiting receipt are inside or outside it; never silently mix different stock statuses or ownership.
Use one line per SKU, bin and stock status, adding lot or serial dimensions where your existing records require them. For the simple pilot below, quantities use whole individual units. Do not use it for weighed quantities or specialist stock without redesigning and validating the measurement rules.
Keep the expected balance away from the counter. Microsoft’s cycle-counting documentation provides a concrete software example of blind entry and review of count differences. This sheet adopts the separation of observation and review without depending on that software.
Use a controlled movement hold
For this small-store pilot, the stores lead pauses physical receipts, issues and transfers for the selected area, completes or identifies pending stock postings, then captures the baseline. Record the same cutoff time and time zone on both sheets. Keep the hold through counting, review and authorized correction. Other areas can continue operating.
A spreadsheet does not enforce this hold. Oracle’s physical inventory snapshot guidance illustrates why a saved balance alone is not a movement lock: its snapshot operation does not restrict inventory transactions. Your stock system may behave differently; check its actual counting and posting procedure.
If the hold is broken, mark the affected lines Invalidated, link every known movement and stop adjustment processing. The stores lead reconciles the movements, takes a new baseline and opens a linked replacement count attempt. Preserve the invalidated evidence. Do not compare a later count with an earlier balance and call the difference a loss.
Copy sheet A: the blind count packet
The stores lead prepares identifiers, locations and scope; the counter records physical observations. Do not include system quantities, differences or earlier count results in the counter’s copy.
Session ID: [unique] Attempt: [1, 2, ...] Store / selected boundary: [site, aisle, bins] Included / excluded areas and stock statuses: [explicit list] Hold authorized by: [name] Backup: [name] Hold and baseline time: [date-time, zone and UTC offset] Count starts / ends: [date-time and zone] Counter: [name] Review owner: [name] Review deadline: [date-time and zone] Movement-hold breach? [No / Yes: incident reference] Packet pages: [page x of total] Added lines: [references] Line ID | SKU | Description | Bin | Stock status | Unit Physical quantity | Counted at | Counter | Observation / evidence ref
- Line ID: unique within the session; retain it on every recount and review record. Number added lines rather than inserting untraceable notes.
- SKU and bin: verify the physical label. An unreadable or unknown identifier needs a temporary reference and review; do not guess a familiar SKU.
- Unit: use the stock record’s defined unit. For a sealed pack, use only a verified conversion and note the packaging basis; never assume a box contains the usual quantity.
- Physical quantity: a nonnegative whole number. Blank means not counted; an explicit 0 means the location was checked and none found. An inaccessible bin stays blank with a reason.
- Evidence: record observations such as open packs, mixed items or separate damaged stock. Follow local access and information-handling rules for any photos.
The counter signs the completed packet and returns all pages to the review owner. For a recount, issue a new blind copy with the same line identity and a new attempt reference. Keep the first observation rather than overwriting it.
Copy sheet B: the restricted reconciliation ledger
Only the stores lead, reviewer and authorized stock-record operator should change the review ledger. One row represents one count line. Use the following columns in order; copy the header into a sheet and split on commas.
Session,Line,SKU,Bin,Stock status,Unit,Baseline qty,First count,Recount,Accepted qty,Variance,Variance pct,Decision state,Reason code,Evidence ref,Approval ref,Posting ref,Reviewer,Closed at
Keep descriptions, timestamps and counter names in the linked packet rather than losing them. The baseline must be the authoritative stock quantity for the exact line dimensions at the recorded cutoff—not an available-to-promise figure or a total across all bins.
Calculate only after validating the inputs
With these columns, use the following formulas in row 2 and fill down. G is baseline, J is the reviewer’s accepted count, K is signed variance and L is percentage variance.
K2 = IF(OR(G2="",J2=""),"",J2-G2) L2 = IF(OR(G2="",J2=""),"",IF(G2=0,"N/A",(J2-G2)/G2))
Format L as a percentage for numeric results. Negative variance means fewer units counted than recorded; positive means more. A zero baseline has no meaningful percentage denominator: retain N/A and review the unit difference. A formula returning blank is not proof of a match.
The reviewer enters J only after checking identifiers, units, count evidence and the hold. In this conservative pilot, every first-count difference requires an independent blind recount. If counts disagree, investigate the cause and record why a specific quantity is accepted; do not average observations or repeatedly count until a preferred answer appears. Every nonzero accepted variance needs explicit adjustment approval. These are local design choices, not universal tolerance requirements.
Use states that separate counting from correction
| State | Entry condition | Owner and exit evidence |
|---|---|---|
| Prepared | Boundary, hold, baseline and packet assigned | Stores lead verifies cutoff and authorizes counting |
| Counting | Blind packet released | Counter supplies every line or a documented unresolved blank |
| Recount required | First count differs or identity/unit is uncertain | Independent counter returns new blind observation |
| Under investigation | Counts disagree, evidence is incomplete or cause unresolved | Reviewer resolves with evidence; no posting allowed |
| Ready for approval | Accepted count differs; investigation complete | Authorized approver records signed quantity, line, attempt and decision |
| Ready to post | Exact correction approved | Stock operator posts once in the authoritative system and records reference |
| Closed | No unresolved lines; accepted quantity verified against resulting stock record | Reviewer records reconciliation and closure time; stores lead releases hold |
| Invalidated | Hold or baseline integrity failed | Stores lead records reason and linked replacement attempt; no correction from this attempt |
A verified zero-variance line needs no adjustment approval or posting: record “not required—verified match” in those fields. Do not use a fake transaction number. A rejected adjustment remains unresolved unless the reviewer documents the alternative resolution; rejection alone is not stock reconciliation.
Worked example: two missing units, not an automatic write-off
Fictional example: session CC-026 covers bin B-04 at North Stores on November 16, 2026. The stores lead completes the movement hold and captures the baseline at 09:00 America/Chicago (UTC−06:00). Each quantity below is in individual units.
| Line | Baseline | First / recount | Accepted / variance | Decision |
|---|---|---|---|---|
| 01, cable clips | 40 | 38 / 38 | 38 / −2 (−5%) | Reviewer verifies dimensions and no unposted movement; approves −2 under AP-026-01 |
| 02, label rolls | 12 | 12 / not required | 12 / 0 (0%) | Verified match; no stock adjustment |
| 03, protective caps | 0 | 3 / 3 | 3 / +3 (N/A) | Unexpected stock; identity/ownership investigated before approval AP-026-03 |
| 04, storage bags | 8 | blank / blank | blank / blank | Bin inaccessible; unresolved, not zero and not a shortage of eight |
The final line keeps the session open. The stores lead arranges access; a counter finds eight bags, and the reviewer accepts a match. For line 01, the physical difference is confirmed but its origin remains unproven: use reason “unexplained difference after investigation,” not “theft.” The authorized approver can require further investigation instead of approving a correction.
Before posting, the operator checks the stock system’s semantics. A delta adjustment for line 01 is −2; a replacement balance is 38. Entering 38 as a positive adjustment would be wrong. Check the proposed transaction and confirm the resulting balance. Record references ST-026-01 and ST-026-03 only after confirmed posting; inspect existing transaction history before retrying an uncertain submission.
Resolve exceptions and test the packet
Set a named reviewer and backup before counting. For a pilot, choose a review deadline within the planned hold window. At that deadline, escalate unresolved lines to the stores manager with the affected bins, blocked operations and next review time. If the hold cannot continue, invalidate unresolved affected lines and arrange a fresh controlled count rather than silently releasing the area and using stale results.
- Unit mismatch: block comparison, verify conversion with the stock owner and recount in the agreed unit.
- Unposted receipt or issue: reconcile the source transaction first. Use the material requisition record for issue evidence and the stock transfer record for movements between stores. Do not post the same event again as a count correction.
- Unexpected goods: keep them distinct until identity, ownership and stock status are established; do not merge them into a convenient line.
- Duplicate packet or uncertain posting: match session, attempt, line and transaction history. One physical observation must not create two adjustments.
Before live use, test these outcomes: blank count stays unresolved; explicit zero against five produces −5 after review; three against zero produces +3 with N/A percentage; matching counts produce no posting; disagreeing recounts block approval; any hold breach invalidates the affected comparison. Confirm that the reviewer cannot accidentally interpret a blank as zero.
Where Workhint fits in count review
The sheets remain usable without Workhint. The Workhint Workflow Engine describes approval routing to defined roles and requiring information before a step advances. Its workflow automation solution describes configurable forms, stages, owners and next actions. Those capabilities are relevant to collecting a count packet and routing a discrepancy decision.
Keep the authoritative inventory system responsible for stock balances and corrections. The workflow pages do not establish a prebuilt cycle-count module, warehouse scanning, inventory costing or automatic ledger posting. Demonstrate any proposed integration separately and retain an explicit human posting check.
Frequently asked questions
Is hiding the system balance enough for a blind count?
No. Give the counter a separate packet without earlier counts or variance formulas as well. Keep the review ledger restricted; merely hiding a column in an editable shared file is not access control.
Should every difference trigger a stock adjustment?
No. First verify the item, bin, unit, movement cutoff and recount. A timing or posting error may need correction of the original transaction rather than an additional count adjustment.
Can we resume operations before review finishes?
Not while relying on this strict-hold pilot’s baseline. Use a validated system-specific counting process if operations must continue. Otherwise preserve the incomplete attempt and arrange a new controlled count for affected lines.
Is a signed sheet proof of inventory accuracy?
It documents a particular observation and review. It does not prove every location was counted, eliminate measurement error or establish accounting compliance. Retain scope, exceptions and resulting stock-record evidence with it.
Start with one manageable boundary. A useful count round ends with explained decisions and verified records—not merely a completed column of numbers.

Leave a Reply