How do you know the ROI for eliminating a risk?
It’s hard to measure because the hazard didn’t actually occur and you don’t know how likely it was.
You can make inferences based on historical data. This has a limit, since very unlikely hazards will also not have occurred enough times for you to get statistically significant data.
Contracts are a minefield for unlikely or poorly understood hazards. Especially if the contract is bespoke or has been heavily negotiated from a template. You simply won’t have enough (if any) historical data.
This is a problem because such contracts tend to be your most valuable ones. Otherwise you wouldn’t have gone to the effort of such extensive customisation.
Expenditure in service of mitigating the risk of these hazards might be best thought of as insurance rather than just an investment (whether or not you’re buying actual insurance).