ROI

The formula is gain minus cost, divided by cost. Spend 10,000 and end with 13,000 and the ROI is 30%. Its appeal is that it compares things that are otherwise not comparable: a marketing campaign, a piece of equipment, a hire.

That comparability is also where it misleads, because three things it omits usually matter more than the number itself.

What the percentage does not say

Time. A 30% return is excellent over one year and poor over five. ROI has no time dimension at all, so comparing two figures without knowing their periods compares nothing. Annualised return exists precisely to fix this, and is the figure worth asking for.

Risk. A 30% return on something that was nearly certain and 30% on something that could have lost everything are not the same result, though the arithmetic cannot tell them apart.

What was counted as cost. This is where most business ROI claims quietly fail. Software that “returned 300%” often counts the licence and not the two months of staff time spent implementing it, the training, or the work that stopped while people learned it.

Where it gets abused

Vendor case studies almost always calculate ROI with the narrowest possible cost and the broadest possible benefit, frequently including savings that were projected rather than measured.

The useful defence is to ask three questions: over what period, compared with doing nothing, and what is in the cost figure. A claim that cannot answer all three is a marketing number.

For anything with ongoing costs — most software, most automation — payback period is the more honest measure. It asks how long until the thing has repaid what it cost, which is harder to inflate and closer to what anyone actually wants to know.