I'm engaged in work presently for which the objective is to generate an accurate cost for a large finance / ERP solution. This forecast will then inform the business case of whether or not to proceed with the implementation.
While the exact figures remain to be determined, if we can reduce the overall costs of the existing provision of several separate platforms by say, 20%, then there's probably a reasonable case for the proposed investment.
A forecast is only ever an estimate of a future state and while the whole life forecast costs may be precise (say £8,673,417.45p), the exact implementation costs as recorded of £10M would show it not to have been very accurate.
Conversely, a forecast of £9M +/- 15% would have been accurate, but not terribly precise. This second approach may seem preferable, but can marginalise any investment appraisal.
This is by no means the only area in which such distinctions between accuracy and precision can be made and if any of your product descriptions specify numerical data, then consideration should be given to questions of accuracy and precision