Presenting Data Readiness to Stakeholders

Overview
In the PMI Cognitive Project Management for AI (CPMAI) methodology, AI projects run through six iterative phases. The first three are Business Understanding, Data Understanding and Data Preparation. Identify Data Needs sits inside Data Understanding. Presenting Data Readiness to Stakeholders is the step where the project team tells decision-makers what it has learned about the data. It covers whether the data needed for the AI solution exists, whether the team can get it, whether its quality is acceptable, and whether it can legally and ethically be used. It also shows where the gaps are and what it will cost to close them. Stakeholders use this to make an informed go/no-go or adjust decision.

Why It Is Important
1. Data is the foundation of AI. CPMAI stresses that AI projects are data-centric, not code-centric. Most AI project failures trace back to data problems: missing data, poor quality, bias, inaccessibility, or no permission to use it. Raising these issues early prevents expensive failures later.
2. It enables informed go/no-go decisions. CPMAI follows the principle of 'think big, start small, iterate often.' A clear data readiness presentation lets sponsors decide to proceed, narrow the scope, collect more data, or stop before major investment.
3. It aligns expectations. Business stakeholders often assume their organization has plenty of usable data. Readiness findings reset expectations about timelines, accuracy and feasibility.
4. It surfaces risk and compliance concerns. Privacy (GDPR, HIPAA), consent, intellectual property, security and ethical issues must be visible to the people accountable for them.
5. It secures resources and commitment. Fixing data gaps often needs budget, data engineering effort, data-owner cooperation, labeling services or third-party data. Stakeholders must approve these.
6. It supports trustworthy AI. Transparency about data provenance, representativeness and bias supports responsible AI principles.

What It Is
Presenting data readiness is a structured communication activity. It summarizes the findings of the data assessment against the requirements set in Business Understanding. It usually answers these questions:
- Availability: Does the required data exist? Where does it live? Who owns it?
- Accessibility: Can the team actually get it? Consider technical access, permissions, cost and time.
- Quantity: Is there enough data, including enough labeled examples and enough rare-class examples?
- Quality: Is the data accurate, complete, consistent, timely and valid? How much cleansing is needed?
- Relevance and representativeness: Does the data reflect the real-world population and conditions the model will face? Is there bias?
- Labeling: For supervised learning, are labels available? If not, what is the cost and approach to create them?
- Governance and compliance: Are there legal, privacy, consent, security or ethical constraints?
- Data in production: Will the same data be available at inference time, with the same format and latency?
- Gaps, risks and mitigations: What is missing? What is the plan, cost and timeline to fix it?
- Recommendation: Proceed, proceed with conditions, re-scope, gather more data, or stop.

How It Works (Step by Step)
1. Revisit the business objective and AI pattern. Tie data needs back to the problem defined in Business Understanding. Examples of AI patterns include recognition, predictive analytics, conversational, patterns and anomalies, hyper-personalization, goal-driven systems and autonomous systems. The pattern drives which data types are needed.
2. Inventory and assess data sources. Catalog internal and external sources, owners, formats, volumes and refresh rates.
3. Evaluate against readiness criteria. Score each dimension (availability, quality, quantity, labeling, compliance and so on). Many teams use a scorecard, heat map or RAG (red/amber/green) status.
4. Identify gaps and options. For each gap, list options such as:
- acquiring third-party data
- synthetic data
- data augmentation
- new data collection
- labeling services
- narrowing scope
- changing the AI approach
5. Quantify impact. Estimate cost, time, risk and effect on model performance and project value.
6. Tailor the message to the audience. Executives need business impact, risk and a decision. Data owners need specific access requests. Legal and compliance staff need privacy and consent details. Technical teams need detailed profiling results. Use plain language and visuals, not jargon.
7. Present findings honestly and transparently. Do not hide bad news. Clearly state assumptions and uncertainties.
8. Make a recommendation and request a decision. Offer go, no-go or conditional-go options, with the trade-offs of each.
9. Document and baseline. Record decisions, approvals, ownership and action items. Update the risk register, project plan and data requirements.
10. Iterate. CPMAI is iterative. If readiness is insufficient, loop back to Business Understanding to re-scope, or continue in Data Understanding to gather more data.

Common Artifacts
- Data readiness scorecard or assessment report
- Data inventory or catalog
- Data quality profiling summary
- Gap analysis and remediation plan
- Risk register updates (data, privacy, bias)
- Go/no-go decision record

Key Stakeholders
- Project sponsor and executives
- Business owners and subject matter experts
- Data owners and data stewards
- Data engineers and data scientists
- Legal, privacy, compliance and security teams
- IT and infrastructure teams
- End users or operations (data in production)

Exam Tips: Answering Questions on Presenting Data Readiness to Stakeholders
1. Prefer transparency. If a scenario shows data is insufficient, biased or noncompliant, the correct answer is almost always to communicate it openly and early. Hiding it, downplaying it or 'proceeding and fixing it later' is wrong.
2. Data before models. Answers that jump into model building, algorithm selection or vendor procurement before data readiness is confirmed are usually distractors.
3. Link to business objectives. The best answers connect data readiness to the business problem and success criteria defined in Business Understanding.
4. Recommend a decision with options. Look for answers where the PM presents gaps, mitigation options, cost or impact, and a recommendation (go, no-go, conditional go or re-scope). Avoid answers that only list problems.
5. Iteration is acceptable. CPMAI expects looping back. Re-scoping to a smaller, data-feasible problem ('start small') is often the best choice when data is lacking.
6. Tailor communication. If the question involves executives, choose the answer focused on business impact, risk and decisions rather than technical detail.
7. Involve the right owners. Privacy or consent issues call for legal and compliance. Access issues call for data owners and stewards. Do not bypass governance.
8. Watch for the 'data exists, so it's ready' trap. Existence alone is not readiness. Accessibility, quality, labeling, representativeness, permission and production availability all matter.
9. Do not overpromise. Answers that guarantee model accuracy before data assessment are incorrect.
10. Document decisions. Updating the risk register and recording stakeholder approval are hallmarks of correct answers.
11. Keywords for correct answers: 'communicate gaps,' 'go/no-go,' 'feasibility,' 're-scope,' 'iterate,' 'data owners,' 'governance,' 'trustworthy AI.'
12. Keywords for wrong answers: 'skip,' 'assume,' 'proceed anyway,' 'build the model first,' 'collect data later,' 'avoid alarming stakeholders.'

Sample Question
An AI team finds that only 30% of the required historical records are labeled, and some customer data lacks consent for secondary use. The sponsor expects model development to start next week. What should the project manager do FIRST?
A) Start modeling with the labeled 30%
B) Present the data readiness findings, including labeling and consent gaps, remediation options and impacts, and request a stakeholder decision
C) Purchase a labeling tool
D) Remove consent-restricted records quietly and proceed
Answer: B. It reflects transparency, governance, options and an informed decision, which are core CPMAI principles.

Summary
Presenting data readiness turns the technical data assessment into a business decision. It confirms whether the AI project is feasible with the data available. It exposes gaps and risks and proposes remediation. It also secures stakeholder alignment before significant investment. On the exam, choose the answers that are honest, data-first, business-aligned, governance-aware and iterative.