Stakeholder analysis is the systematic process of identifying, categorizing, and prioritizing the individuals and groups affected by or influencing a project, program, or organization. The standard practice uses one of five canonical frameworks — the Mendelow Power-Interest Grid, the Salience Model, the Stakeholder Onion Diagram, the RACI Matrix, or a Stakeholder Map — to organize stakeholders by their influence, interest, and legitimacy. The goal is a defensible plan for engagement, communication, and risk management.

The five steps in stakeholder analysis are: 1) Identify every stakeholder affected by or influencing the work. 2) Categorize each stakeholder by power, interest, legitimacy, and urgency. 3) Prioritize using a framework — Mendelow grid for most projects, Salience for crisis or ethical contexts. 4) Design an engagement strategy per category — manage closely, keep satisfied, keep informed, or monitor. 5) Track changes continuously, because stakeholder relationships shift and the analysis is only useful when it stays current.

The Mendelow Power-Interest Grid is a 2x2 matrix that plots stakeholders along two axes — their power to influence the project and their interest in its outcome. The four quadrants generate four engagement strategies: manage closely (high power, high interest), keep satisfied (high power, low interest), keep informed (low power, high interest), and monitor (low power, low interest). Developed by Aubrey Mendelow in 1991, it remains the most widely taught stakeholder analysis framework.

The Salience Model, developed by Mitchell, Agle, and Wood in 1997, classifies stakeholders by three attributes: power (ability to influence), legitimacy (validity of their claim), and urgency (time-sensitivity of their concern). The combinations produce seven stakeholder types ranging from dormant (power only) to definitive (all three). The model is stronger than Mendelow for crisis management and ethical decision-making because it explicitly weighs whose claims deserve attention, not just whose voice carries volume.

A stakeholder map is a visual representation of all the individuals and groups connected to a project, showing their relationships to each other and to the work. Forms include onion diagrams (concentric rings of distance from the core), influence-network maps (showing who influences whom), and ecosystem maps (showing the broader system the work sits within). Stakeholder maps are most useful at project kickoff for surfacing missing stakeholders and at major decision points for visualizing how a change affects the network.

A RACI matrix is a stakeholder analysis tool that assigns four roles to each task or decision: Responsible (does the work), Accountable (signs off on the work), Consulted (provides input), and Informed (receives updates). Variants include RASCI (adds Supportive) and DACI (Driver, Approver, Contributor, Informed). Unlike the Mendelow or Salience frameworks which categorize stakeholders broadly, RACI is granular — it operates at the task level and clarifies decision rights when the project actually moves.

A worked stakeholder analysis example for a workforce training program would identify five groups — program participants, employer partners, funders, instructional staff, and community referral partners — then plot each on the Mendelow grid. Funders would land in 'manage closely' (high power, high interest). Employer partners in 'keep satisfied' (high power, lower active interest until hiring season). Participants in 'keep informed' (low formal power but high interest in their own outcomes). The output is an engagement plan with frequency, channel, and content tailored to each group's quadrant. Four full worked examples appear in the Examples section above.

Stakeholder analysis produces five concrete benefits. It surfaces missing stakeholders before they become surprises. It clarifies which voices should have the most weight in decisions. It allocates engagement effort proportional to influence. It builds the communication plan that prevents preventable conflict. And it creates a defensible record showing the project considered the people it would affect — which is increasingly required by funders, regulators, and ESG-reporting standards.

Conduct a stakeholder analysis in five steps. First, list every individual or group that could affect or be affected by the project — be exhaustive, then narrow later. Second, gather data about each stakeholder's current position, interests, and influence. Third, plot them on a framework appropriate to the project — Mendelow for most, Salience for ethical or crisis contexts, RACI for execution. Fourth, build the engagement plan from the categorization. Fifth, set a review cadence so the analysis stays current as positions shift.

Stakeholder mapping is one step inside stakeholder analysis. Mapping is the visualization activity — drawing the matrix, the onion diagram, or the ecosystem chart. Analysis is the broader process that includes identification, data collection, categorization, prioritization, engagement planning, and ongoing tracking. Mapping is the picture. Analysis is the discipline.

Stakeholder analysis tools fall into four eras. Paper templates from project management textbooks still work for one-off analyses. Digital tools like Lucidchart, Miro, and Microsoft Visio handle the diagram layer. Cloud platforms like Smartsheet and Monday provide stakeholder registers integrated with project tasks. AI-native stakeholder intelligence platforms — Sopact Sense in this category — replace the snapshot model with continuous data capture, persistent stakeholder IDs, and automated sentiment and theme analysis from feedback collected over time.

Stakeholder intelligence is the practice of treating stakeholder relationships as continuous data rather than a one-time mapping exercise. Where traditional stakeholder analysis produces a static matrix at project kickoff, stakeholder intelligence captures every interaction — survey response, interview transcript, sentiment shift, escalation event — onto a persistent stakeholder ID. The output is a living view of who matters, why, and how their position has changed over time. Sopact Sense is the first purpose-built stakeholder intelligence platform. See the engine pillar at stakeholder intelligence.

Stakeholder sentiment analysis is the systematic measurement of how positively or negatively a stakeholder group regards an organization, project, or decision. Traditional sentiment analysis ran on social media at the population level. Modern stakeholder sentiment analysis runs on first-party data — survey open-ends, interview transcripts, support tickets, complaint logs — and ties each sentiment score to a specific stakeholder ID so trends are trackable per group over time. For methods on collecting the underlying feedback, see stakeholder feedback.

Traditional stakeholder analysis was treated as a project-kickoff deliverable updated at major milestones — annually at most, often only once at the start. Modern practice treats it as continuous: stakeholder positions, sentiment, and influence are re-scored every time new evidence arrives. A funder who was previously 'keep satisfied' may move to 'manage closely' the week before a renewal decision. A continuous system surfaces that shift in time to act on it. A static document does not.

Stakeholder analysis is about identifying and categorizing stakeholders by influence and interest. Stakeholder impact analysis is the next step — measuring what changes for each stakeholder group as a result of the work. Stakeholder analysis answers 'who matters?' Stakeholder impact analysis answers 'what changed for them, and by how much?' Strong programs run both: analysis to plan engagement, impact analysis to evaluate the engagement's results. For the deeper impact-measurement guide, see stakeholder impact analysis.

Stakeholder analysis matters because most project failures are stakeholder failures. The technical work is rarely the problem. What kills projects is a missed stakeholder whose objection was preventable, a misaligned engagement strategy that wasted effort on people who could not move the work, or a shift in stakeholder position the team did not see coming. A well-run analysis is the cheapest insurance against the failure modes that cost the most.

Use both. Generative AI tools like ChatGPT, Claude, and Perplexity are excellent at synthesizing stakeholder feedback and drafting engagement plans, but they cannot solve the structural problems of stakeholder analysis. They cannot maintain a persistent stakeholder ID across years, validate data at submission, or join survey responses to interview transcripts on a shared key. The best practice in 2026 is coexistence — a stakeholder intelligence platform provides the clean, joined substrate; generative AI reasons over it. See the two best practices section above for the full argument.

A stakeholder analysis report documents the output of the analysis for distribution to project leadership, sponsors, and engagement teams. Strong reports include: a stakeholder register listing every identified party, a categorization framework applied (Mendelow, Salience, or RACI), an engagement strategy per category, a communication cadence and channel matrix, identified risks and mitigations, and a review schedule. Continuous-intelligence platforms produce these reports as queryable views on live data rather than static documents that age out within weeks.