Skip to content
Guide
Inventory Management
Counting Stock: Count and Read the Band

Counting stock: run the count, then read the band before the variance

Steps 3 and 4 of counting stock and deciding what a difference means: run the count, then read the report’s tolerance band before you read any variance. It follows Step 2, where the four controls were set up before the count. The running example is the weekly phone count at branch GS-KV-01, where a flagship smartphone costs about RM 2,450. The stock take measures the difference and changes nothing: no stock moves until an adjustment in Step 7.

Keep the session Open until Step 5’s reasons are saved; Step 3 says what closing it early costs. Two things from the front page guard these steps: the Stock Take applet’s Settings > Field Settings must have been saved once, or closing a session gets no Status Variance verdict (Before you start); and BigLedger has no blind-count mode, so the supervisor, not a counter, generates the report (Step 2).

  1. Step 3: count.
  2. Step 4: read the band before you read the variance.

Step 3: Count

The outcome: every record in the session is one the counters stand behind.

Stock Take > Stock Take Session > + (Create Session, one location) > Devices > + (Create Device Session) > Records

Open one session per location per count. For serialised phones, turn on Validate Serial Numbers in Field Settings so a mistyped serial is refused while the counter is still standing at the shelf.

Three things go wrong at this stage, and each one shows up in the report as a variance that never happened:

  • A case counted as a unit, or a unit as a case. Records are stored in single units, and the report shows packs, divided by the largest pack size. A variance of exactly one case quantity is a unit-of-measure mistake. It is not a loss.
  • A record with no device. A record saved without a device counts in the report’s total and does not show on the device’s Records screen. The supervisor who generates the report compares its total with the sum of the devices’ records. If the report total is higher, look for these records first.
  • Counting after Closed, and closing too early. Closing a session disables the buttons in the web applet only. A handheld can still write to it. It also disables GENERATE, REGENERATE, UPLOAD REASON and the session’s SAVE, and the web applet offers no way back to Open. So the report cannot be generated again once the session is Closed, and any Reason not yet saved cannot be saved. Close the session last, after Step 5’s reasons are saved: the supervisor collects the devices when the count ends, generates the report while the session is still Open, and closes it once the report is final. The supervisor knows nothing was added afterwards from the session’s Records tab, which stays readable after closing and lists records newest first by Modified Date. Note the time you close. If the top row is later than that, a device wrote to the closed session and that record is not on the report. Recount that item before Step 7 adjusts it.

Step 4: Read the band before you read the variance

The outcome: you know which differences are timing and which are real.

Stock Take > Stock Take Session > (the session) > Report > GENERATE

The report does not use a figure taken when the count started. It works out the expected quantity from the ledger each time you press GENERATE, over the window from the first record to the last edit. For each item it shows:

  • Min: what the shelf would hold if every sale during the count had already left it.
  • Max: what the shelf would hold if every receipt during the count had already arrived.
  • Variance: zero if the count is anywhere between Min and Max. Otherwise it shows the distance to the nearer edge.
  • Variance 2: the plain difference between counted and expected, ignoring the band.

At GS-KV-01 the count starts at 10:05 and ends at 10:50. Three phones of one model are sold during that time. The ledger opened at 40, so Min is 37 and Max is 40. The counters found 38. Variance is 0. One of the sold phones was still on the shelf when it was counted, and the other two had already gone. That is timing, not a loss. If they had found 35, Variance would be 2 short of the band, and that is a real difference.

A zero Variance means “within the band”. It does not mean “correct”. And the band only works if the count is short. A session whose last record is edited the next morning measures the whole night’s trading, and almost anything will fall inside it. Keeping the window short is the supervisor’s job. They know it was short when the session’s Start Date and End Date are the same day. The web applet moves those dates with every record it saves; a handheld may not, so for a handheld count the supervisor collects the devices at the planned end time. Leave the session Open until Step 5 is done, because a Closed session cannot be generated again (Step 3). If the End Date is later, treat every zero Variance on a high-value item as unproven and recount those lines.

The report’s cost is the moving average at that location before the count started, and never the FIFO cost. An item with no earlier movement at that location shows a cost of zero. Take the value of a write-off from the adjustment in Step 7, not from this report. The finaliser checks it there: each Adjust-Out line’s unit price is the item’s current moving average, which is the default.

What success looks like for this part

The report’s total agrees with the sum of the devices’ records, the session’s Start Date and End Date are the same day, so the lines inside the band can be read as timing (where the window ran longer, every high-value line inside it has been recounted), any line out by exactly one pack quantity has been checked as a unit-of-measure mistake, and the session is still Open, so the reasons in Step 5 can be saved. Then carry on to Investigate and write off.

Common mistakes in this part

The mistakes are listed together on the front page. Two belong here: expecting the stock take to post the variance, and letting a session run overnight, which puts nearly every item inside the band.

Related documentation

Last updated on
Counting Stock and Deciding What a Difference MeansCounting Stock: Investigate and Write Off