ROI Is a Useful Number That Hides Two Important Things

ROI earned its popularity honestly. One subtraction and one division give a figure anyone can interpret, which is why it appears in board packs, property listings and campaign reports alike.

The first omission is time. A shop refit costing 40,000 that adds 26,000 of profit shows a 65% ROI whether it took eight months or six years. Annualised, those are wildly different decisions: about 105% a year in the first case and roughly 8.7% in the second. The fix is one step — raise (1 + ROI) to the power of 1 divided by the number of years.

The second omission is risk. ROI treats a government bond and a speculative product launch identically if both return 12%. Any serious comparison has to ask what could have gone wrong, and how much capital was exposed while you waited.

There is also a definitional trap worth naming. Because both 'gain' and 'cost' are chosen by whoever builds the model, ROI is easy to flatter — leave out your own labour, count revenue instead of profit, ignore the fees, and a marginal project starts to look excellent.

Use ROI as a first screen, then test survivors with an annualised rate and, for anything multi-year, a discounted cash-flow view.