The quick answer
Each cell in the Pivot - Median Unit Cost per Assembly by Quarter sheet is the median of the row-level Cost/Unit values on the table sheet that fall in that assembly and that quarter. Rows with a blank Cost/Unit are ignored, an even number of values gives the average of the two middle ones, and a year or Grand Total is worked out again from all rows, not from the quarter cells. To check one, list the assembly's rows on the first sheet, sort their Cost/Unit values and take the middle.
The screenshots show sample data. Company, person, item and location names are fictional. Use your own records and values when following the steps.
Before you start
You need SuiteAnalytics Workbook access and a role that can open the Manufacturing Workbook and see the transactions it reads. This example uses a company that builds assemblies with work orders and Work in Process. The Manufacturing Workbook is a standard template built on the Manufacturing Dataset. In its delivered configuration the dataset is restricted to Assembly Build and Work Order Close transactions and uses a rolling default date filter that starts about two years before the current date, so other transaction types and older transactions may not appear. Confirm the exact row grain and criteria in your dataset rather than inferring them from a sheet. The sheets, fields and criteria can differ with your features, permissions, release and template version, so check the dataset's criteria and each sheet's filters in your account.
A workbook is an analytical view, not an accounting report. It's good for seeing where a unit cost figure comes from, but it doesn't replace GL reports or a transaction's GL Impact.
01. Open the median unit cost pivot
With the Manufacturing Workbook open on its first sheet, click the sheet tab Pivot - Median Unit Cost per Assembly by Quarter. The layout panel shows Assembly as the row, Date (Year) and Date (Quarter) as the columns and Cost/Unit (Median) as the measure. The Oak Shaker Door row reads 31.10 under Q3, 30.55 under Q4, 31.10 under 2026 and 31.10 as the Grand Total. 30.55 isn't the Cost/Unit of any transaction, and the year isn't the average of the two quarters.
02. Find the assembly's rows and check Q4
Click the first sheet tab, Manufacturing Workbook, and scroll the grid to the right so the Assembly and Cost/Unit columns are on screen together. Oak Shaker Door has six rows. Three are dated in October 2026, which is Q4. Work Order Close #WCL215 has a quantity of 0, so its Cost/Unit is blank. Work Order Close #WCL213 is 30.00 and Assembly Build #ASB5111 is 31.10.
The blank value is skipped, not counted as zero. That leaves 30.00 and 31.10, and with an even number of values the median is their average, (30.00 + 31.10) / 2, which is 30.55.
03. Check Q3 and the year
The other three Oak Shaker Door rows are dated in Q3: Work Order Close #WCL209 is 18.64, and Assembly Builds #ASB5102 and #ASB5090 are 31.10. Sorted, the values are 18.64, 31.10 and 31.10, so the middle one, 31.10, is the Q3 median.
For the year, take all five values together: 18.64, 30.00, 31.10, 31.10 and 31.10. The middle value is 31.10. Each row counts once whatever its quantity, so a 4-unit close and a 9-unit build weigh the same. If you want a cost per unit weighted by quantity, divide a total Amount by a total Quantity instead.
04. Compare the year and Grand Total with the pivot
Click the pivot sheet tab again. For Oak Shaker Door the 2026 column and the Grand Total are 31.10, the median of all five rows. They aren't the average of the two quarter cells, 30.83, and they aren't their sum. A year total or Grand Total is calculated again from every row in its scope.
The Grand Total row at the top of the pivot works the same way. It's the median of all 14 rows that have a Cost/Unit: 23.00 under Q3, 26.50 under Q4 and 23.00 for 2026. That's a median of rows, not a cost per unit for the company.
Check the result
- The Q4 cell, 30.55, is the average of the two non-blank Q4 values, 30.00 and 31.10. The row with quantity 0 and a blank Cost/Unit is ignored.
- The Q3 cell, 31.10, is the middle of 18.64, 31.10 and 31.10.
- The 2026 and Grand Total cells, 31.10, are the median of all five Oak Shaker Door values, not the average, 30.83, of the two quarter cells.
- If a cell differs, check the sheet filters, the dataset's date criterion, the transaction Date against the Posting Period, the currency and whether any rows are duplicated by a join. Rounding the figures shown can also make a hand calculation differ slightly.
Common questions
How is the median unit cost in the Manufacturing Workbook calculated?
Each pivot cell is the median of the row-level Cost/Unit values of the dataset rows in that assembly and date scope, matched on the transaction Date. Cost/Unit itself is the transaction amount divided by the quantity, only when the quantity is above zero. With an even number of values the median is the average of the two middle values after sorting, so it can be a number that no row shows, like 30.55.
Are blank Cost/Unit values counted as zero?
No. A Work Order Close row with a quantity of 0 has a blank Cost/Unit, and a median ignores blank values. In the example, the Q4 cell would be 30.00 if the blank row counted as zero. It reads 30.55.
Why isn't the year the average of its quarters?
A year total, a Grand Total row or a Grand Total column isn't built from the quarter cells you see. It's calculated again from all underlying rows in its scope. Here the quarters are 31.10 and 30.55, their average is 30.83 and the year is 31.10.
Can I add up Cost/Unit to get a cost per unit for the company?
No. A unit cost is a rate, so adding the cells or the rows doesn't give you anything meaningful, and the Grand Total of a median is a median of rows. For a quantity-weighted cost, divide a total Amount by a total Quantity.
Does the median match inventory valuation, standard cost or the general ledger?
Not automatically. Cost/Unit is calculated from a transaction's amount and quantity and isn't the same as an item's standard or average cost. A workbook is an analytical view, so to reconcile to the general ledger use GL reports or a transaction's GL Impact with the same accounting book, subsidiary, currency and date or period basis.
Why might my median differ from the one I work out by hand?
The median is calculated from the unrounded values and the sheets round for display, so a rebuilt cell can differ slightly, depending on the precision of the source values. Duplicate rows from joins, sheet filters, the dataset's date criterion, currency or subsidiary scope and a Date that falls in a different quarter from the Posting Period can also change it. Check the filters on each sheet before comparing.