When a percentage change lies

A percentage change is one of the most trusted figures in a management pack, and there are two cases where it should not be printed at all.

Not rounded, not footnoted, not shown in grey. Not printed.

Case one: the base is zero

A line that was zero last year and is $40,000 this year has increased by an undefined percentage. Dividing by zero is not a large number; it is not a number.

Spreadsheets tend to show #DIV/0!, which at least fails loudly. The dangerous version is a model that catches the error and prints a dash, a zero, or worse, a figure computed against a nudged base.

The movement itself is perfectly reportable: the line went from nil to $40,000. That sentence carries everything a reader needs, and the percentage adds nothing to it.

Case two: the values sit either side of zero

This one is subtler, and it survives into published packs.

A business unit lost $40,000 last year and made $10,000 this year. The percentage change calculation runs happily:

(10,000 − (−40,000)) ÷ |−40,000| = 50,000 ÷ 40,000 = 125%

125% is arithmetically correct and tells the reader nothing true. It suggests a magnitude that scaled by a little over one. What actually happened is that a loss-making unit became profitable, which is a change of kind, not of degree. No percentage expresses it, because percentage change assumes both values sit on the same side of zero and this pair does not.

The honest presentation is the movement: a $50,000 improvement, from a $40,000 loss to a $10,000 profit. Anyone reading that understands the year. Nobody reading 125% does.

What the tools do

Every Consilium tool suppresses the percentage in exactly these two cases, and the cell says which case applies rather than going blank — a blank looks like missing data, and a reader will go looking for the number that was supposed to be there.

The absolute movement is always shown, and it always still counts. It stays in the totals, in the driver ranking, and in the materiality assessment. Suppressing a percentage is a decision about presentation, never a decision to exclude the line from the analysis.

Why this is worth a rule

Both cases share a shape: the arithmetic succeeds, so nothing errors, so nothing is caught. The output is a plausible number in the right format, in the right column, coloured the right way. It passes every check except being meaningful.

Those are the errors worth building rules against. A formula that breaks gets fixed on the day it breaks. A formula that quietly returns 125% gets presented to a board.

See it

Load the example in the Budget vs Actual Analyzer or the EBITDA Bridge and look for a line where the percentage column is suppressed. The methodology panel on each page states the rule in full.