If the expected ROI on a particular project is so razor thin that accurate estimates are needed, don’t even start. Be sure to do your due diligence and aim for at least 10X (ideally > 100X). If this is provably true, then accurate estimates will not be needed. If this cannot be concluded, simply live with the limitations of off-the-shelf offerings.