Project risk is the possibility that uncertainty will produce adverse effects, including cost escalation, unmet targets, or failure to deliver. Pre-project investigation reduces exposure by testing business requirements, data availability, security, privacy, ethics, and the likelihood that an ML model can achieve the required properties. section 2.0 section 2.3.1
Manage risk before committing the project:
- Review the risk register after estimating resources; include the effort and elasticity needed to mitigate or resolve identified risks. section 3.2.3
- Identify mitigations or workarounds, distinguish proven mitigations from novel ones, and record both their costs and confidence of resolution. section 3.2.3
- Apply higher premiums to low-confidence mitigations, investigate high-impact risks with weak resolution paths, and obtain accountable organizational sign-off. section 3.2.3
- Document decisions, share them with approvers, and review the risk register with stakeholders during project execution. section 3.2.3
Pricing mitigation uncertainty
If a low-confidence mitigation is estimated at $500, budget $1,000; a high-confidence mitigation at the same estimate receives a 30% premium and is budgeted at $650. Unmanageable data quality or availability problems may prevent the project from proceeding. section 3.2.3
Funding structure changes who bears delivery risk. Fixed-price, fixed-time work places much of that risk on the delivery team, where poor estimates can produce crunch, overtime, and commercially damaging cost escalation. Preparation and investigation are therefore safeguards against accepting an unmanageable commitment. section 2.3.1 section 2.0
