o Research shows that the number of serious errors in a spreadsheet is about 2% of the number of input cells plus the number of calculated cells (labels and blank cells do not count)
o The number of cells in the CarsExpanded sample model is 1759
o In a conventional spreadsheet program, we would expect about 35 cells to have serious errors
o But because CarsExpanded is a Quantrix model, it has only 33 input cells and 9 formulae, or 42 places where an error might lurk
o That means we expect less than one error!
o The only reliable way to find errors is laborious inspection by a group (individual inspectors only find about half the errors)
o In particular, certain common errors are impossible in Quantrix: calculated cells are always protected and can’t be overridden with a numerical value (“hard-wired”)
Feel free to reuse any or all of this.
I just created a model that had 3,300 SKU’s each with sales and unit values over 7 sales channels with
each SKU assigned one of 50 possible grouping identifiers evaluated over
36 months. The total number of formulas, six!
In the model, I now have available the average selling price, grouping pie charts, sales
trends by SKU, group or sales channel. I can change the grouping
assignments or add SKU’s without changing a single formula. The formulas necessary in