Navigating stakeholder priorities
Unit
In ProgressML projects are a juggling act — product managers want features, engineers want stability, and finance wants lower costs. Fail to balance these voices, and your project could come crashing down.
Clear communication is only useful if it helps the project move forward, and in ML, that means managing competing demands. sit at the crossroads of business goals, technical execution and budget constraints, where stakeholders often have conflicting objectives. Your role is to navigate these differences and keep the project aligned with its larger purpose.
Prioritisation
Conflicting demands are inevitable in ML projects, but not all requests carry the same weight. Your job is to evaluate, prioritise and sometimes make tough trade-offs that protect the project’s larger business value.
When evaluating competing requests, consider these guiding questions:
- Alignment: Which requests directly support the project’s core business objective?
- Risk: What happens if this request is delayed or ignored? Does it increase business, ethical or technical risk?
- Impact vs effort: What’s the return on investment?
Remember
Saying ‘yes’ to everything often spreads resources too thin, leading to delays, burnout and diluted results. A well-prioritised backlog ensures the project delivers real value, not just activity.
Practical example: A product manager requests a new feature to improve customer engagement, while compliance insists on stronger data privacy checks. If the model is not compliant, it cannot be deployed at all, so compliance must come first.
Practical steps for negotiation and coordination
Clear roles, evidence-backed requests and active listening are three powerful tools that help you turn tension into collaboration.
1. Define roles and responsibilities
A RACI chart (Responsible, Accountable, Consulted, Informed) helps bring clarity.
| Task | Responsible (R) | Accountable (A) | Consulted (C) | Informed (I) |
|---|---|---|---|---|
| Build and test model | Data scientist | Project lead | Risk team | Executives |
| Approve compliance | Compliance team | Compliance lead | Data scientist | Executives |
| Deploy model | ML engineer | Project lead | Security team | Executives |
| Report impact | Project lead | Head of risk | Finance team | Executives |
2. Negotiate with data
Data is your strongest argument. Instead of vague statements, show measurable evidence.
- Less effective: ‘We need more time.’
- More effective: ‘To achieve our 95% accuracy target, experiments show we need an additional three weeks for hyperparameter tuning. Without this, the model will remain at 87% accuracy, which increases the risk of missed fraud cases.’
3. Active listening
Behind every demand is a deeper motivation. Active listening uncovers it.
- Ask clarifying questions: ‘What’s driving this request?’
- Restate their concern in your own words to confirm understanding.
- Look for opportunities to align.
Case study: Rolling out a fraud detection model
You’re leading the development of a fraud detection model for a mid-sized bank.
Prioritisation
- The CFO pushes for immediate deployment.
- The engineering team requests two more weeks for accuracy.
- The compliance team insists on privacy and bias checks.The Strategy: Compliance takes priority (regulatory requirement). Next, you weigh engineering’s request against the CFO’s urgency by showing the projected business impact of a short delay (catching more fraud).
Negotiation and coordination
- RACI chart: Defined roles to reduce overlap and confusion.
- Negotiate with data: Show that extending tuning improves recall from 88% to 94%, capturing 150 additional cases monthly.
- Active listening: Uncover the CFO's underlying pressure. Suggest a phased deployment as a compromise.
Action item: Quiz — navigating stakeholder priorities
Test your understanding of how to prioritise conflicting demands and negotiate effectively.