The best AI technology in the world won’t save a project with poor stakeholder management. Research from MIT Sloan Management Review confirms this: technically excellent implementations often fail because the right people weren’t aligned. I’ve also seen mediocre technology succeed because stakeholder management was masterful.

Politics isn’t a dirty word. It’s how organisations actually work.

The Stakeholder Map for AI Projects

Every AI project has these stakeholder groups. Miss any of them and you’re taking on unnecessary risk.

The Executive Sponsor

You need someone senior enough to:

• Approve budget
• Remove organisational obstacles
• Protect the project when it faces criticism
• Make decisions quickly when needed

The executive sponsor doesn’t need to understand AI deeply. They need to believe in the business value and be willing to spend political capital to make it happen.

Common mistake: Treating executive sponsorship as a formality. A sponsor who signs off but doesn’t actively champion the project is worse than no sponsor at all – it creates false confidence.

The Data Owners

These are the people responsible for the data your AI will use. They control access, understand quality issues, and have opinions about how their data should be used.

Data owners can make or break your project:

• Access delays can stall everything
• Quality issues they didn’t mention can derail models
• Privacy concerns they raise late can force redesign

How to engage: Early and often. Explain what you need, why you need it, and what happens to the data. Address their concerns before asking for access.

The IT/Infrastructure Team

Your AI needs to run somewhere. It needs to connect to systems. It needs to meet security requirements. IT controls all of this.

If IT sees your project as extra work dumped on them without consultation, they’ll slow-walk everything. If they’re partners from the start, they’ll find ways to make things work.

How to engage: Involve them in architecture discussions from day one. Ask for their input on approach, don’t just present them with requirements.

The Business Users

The people who will actually use the AI every day. Their adoption determines whether the project delivers value.

Business users have legitimate concerns:

• Will this make my job easier or harder?
• Will I be blamed for AI mistakes?
• Will this threaten my position?
• Did anyone ask me what I actually need?

How to engage: Listen before telling. Understand their current workflow, their pain points, their fears. Design with them, not for them.

The Sceptics

Every organisation has people who’ve seen technology projects fail before. They’re cynical about promises, and they have influential voices.

You might be tempted to ignore sceptics or work around them. Don’t. Unconverted sceptics become resistance leaders. Converted sceptics become powerful champions.

How to engage: Acknowledge their concerns as legitimate. Don’t over-promise. Show them evidence, not slides. Let them poke holes in your plan – they’ll find the weaknesses anyway.

Legal and Compliance

AI creates novel risks that legal teams need to assess. The Australian Information Commissioner and AIIA have published guidance on:

• Liability for AI decisions
• Privacy implications of data use
• Regulatory requirements
• Intellectual property questions

Legal involvement that comes late causes expensive rework. Legal involvement that comes early shapes design appropriately.

How to engage: Brief them early. Frame AI in terms of risks and mitigations, which is how they think. Ask for their guidance rather than asking them to approve your fait accompli.

The Stakeholder Engagement Plan

Mapping stakeholders isn’t enough. You need a plan for each.

For each stakeholder group, document:

1. What do they care about? Business outcomes, risk reduction, career advancement, workload – motivations vary.

2. What are their concerns? Legitimate worries about the project or its implications.

3. What do they need to know? Information that addresses their concerns and supports their objectives.

4. How will we engage them? Frequency, format, who leads the relationship.

5. What’s their current position? Supporter, neutral, sceptic – and why.

6. What would move them? What would change a sceptic to neutral or neutral to supporter?

Review this monthly. Positions change as the project evolves.

Communication Cadence

Different stakeholders need different communication:

Executive sponsor: Monthly steering committee plus ad-hoc escalations. Focus on progress against milestones, risks requiring decisions, and outcomes achieved.

Data owners: Weekly during data preparation, then monthly. Focus on requirements, timelines, and any issues discovered.

IT team: Weekly during implementation, then ongoing for operations. Focus on technical specifics, integration requirements, and support planning.

Business users: Frequent during design and rollout, then ongoing for feedback. Focus on how the AI affects their work and how to get the best results.

Sceptics: Periodic check-ins throughout the project. Focus on evidence of value and honest acknowledgement of limitations.

Legal/Compliance: Early briefing, then involvement at key milestones. Focus on risk assessment and policy implications.

The Difficult Conversations

Some stakeholder situations require particular care:

When IT Says No

IT teams often respond with “no” to protect the organisation from risks they’ve been burned by before. Don’t fight this directly.

Instead: Understand their concerns fully. What specifically are they worried about? Often there’s a version of your plan that addresses their concerns while still meeting your needs.

When Business Users Resist

Resistance usually stems from fear – of change, of job loss, of being blamed. Pushing harder makes it worse.

Instead: Slow down. Listen more. Involve resistant users in design decisions. Let them shape the solution they’ll use.

When Executives Lose Interest

Sponsors who disengage leave projects vulnerable. Usually this happens when the project stops being visible or when other priorities crowd it out.

Instead: Keep the project relevant through regular updates tied to strategic objectives. Escalate decisions appropriately so the sponsor stays engaged.

When Scope Expands

Stakeholders will inevitably want more than the original scope. Some of this is legitimate learning; some is scope creep.

Instead: Have clear criteria for what’s in and out of scope. When new requests come, evaluate them against the original business case. Say “yes, in phase 2” more often than “no.”

The Politics of Success

Here’s an uncomfortable truth: your project’s success isn’t just about outcomes. It’s about perception of outcomes.

A project that delivers modest improvements but has strong stakeholder advocacy will be perceived as successful. A project that delivers significant improvements but has poor stakeholder management will be perceived as problematic.

This isn’t fair, but it’s real. Manage perceptions alongside reality.

Practical steps:

• Communicate wins early and often
• Attribute success to the team and stakeholders, not yourself
• Frame problems as challenges to solve, not failures
• Build relationships before you need them

Final Thought

AI projects exist within organisations, and organisations run on relationships. Technical excellence is necessary but not sufficient.

The project managers who succeed with AI are the ones who understand that getting the technology right is only half the job. Getting the people right is the other half – and often the harder half.

Invest in stakeholder management as seriously as you invest in the technology. Your project’s success depends on it.