Salience Model

Purpose

The salience model is a typology of stakeholders. It is descriptive: it shows which claims hold the attention of managers.

The stakeholder matrix places each actor by power and interest. A loud claim and a well-founded claim often land in the same cell there. The salience model separates them: the first carries urgency, the second legitimacy. It also shows the actor without power who may gain some.

The team derives an order of priority from the class: which claims it handles first. The model also points to the actors to watch: those that a foreseeable event would move to another class.

The deliverable pairs a map with a register. The map shows three overlapping circles with each stakeholder placed in its zone. The register gives one row per stakeholder: the fact behind each attribute, the class, the trigger to watch, who judged and on what date. The PMBOK Guide lists stakeholder mapping among the tools of the Identify Stakeholders process. The team repeats this process throughout the project. Its outputs feed the stakeholder register and the stakeholder engagement plan.

Usage

When to use it

• Large or dense network of relationships: the PMBOK Guide intends the model for large and complex communities.
• Competing claims: separate the pressing claim from the well-founded claim before allocating effort.
• Legitimate actors without formal power: residents, tenants or a staff committee stay visible on the map.
• Public procedure ahead: a public inspection of plans, a referendum or a consultation can give power to an actor who had none.
• Milestone or external event: redraw the map.

When not to use it

• Small project with simple relationships: the PMBOK Guide considers the stakeholder matrix sufficient there.
• Actors' distance from the solution: the onion diagram places actors by proximity.
• Gap between current and desired engagement: the stakeholder engagement assessment matrix measures it actor by actor.

Description

The three attributes

Power is the ability of an actor A to get an actor B to do something that B would not otherwise have done. Mitchell and his co-authors take this definition from Dahl and Pfeffer. They take the sources of power from Etzioni. Coercive power rests on force or constraint: a site occupation, a blockade. Utilitarian power rests on material or financial resources: a budget, production capacity, access to land. Normative power rests on symbolic resources: a reputation, a platform in the press, the backing of an umbrella association. The PMBOK Guide states the attribute in project terms: the actor's level of authority or its ability to influence the outcomes.

For legitimacy, the authors adopt Suchman's definition: the generalised perception that the actions of an entity are desirable, proper or appropriate within a socially constructed system of norms, values and beliefs. The PMBOK Guide sums it up: the actor's involvement is appropriate. A law, a contract, a property right or an elected mandate are its clearest bases. A project that directly affects an actor's property or life also gives the actor's claim a legitimate basis. The team's sympathy for a cause confers no legitimacy.

Urgency is the degree to which an actor's claim calls for immediate attention. Mitchell and his co-authors distinguish two components. Time sensitivity measures how unacceptable a delay in responding is to the actor. Criticality measures how important the claim is to the actor. The PMBOK Guide accepts either one: a time constraint or high stakes. The analyst who applies this rule records in the register which component grounds the urgency.

The model treats each attribute as present or absent. From the number of attributes the authors derive levels of increasing salience: latent stakeholders hold one attribute, expectant stakeholders hold two, the definitive stakeholder holds all three.

The classes

The class names are those of Mitchell and Lee (2019).

The dormant stakeholder holds power without a legitimate or pressing claim. It becomes salient when it acquires a second attribute. The discretionary stakeholder has a legitimate place in the project and exerts no pressure. The team deals with it at its own discretion, hence the name. The demanding stakeholder presses the team without the power to force it and without a right to do so. The dominant stakeholder has power and a legitimate claim that is not yet pressing. The dangerous stakeholder combines power and urgency without legitimacy. It may resort to coercion. The dependent stakeholder has a legitimate and pressing claim. It depends on another actor to carry the claim through. The definitive stakeholder holds all three attributes and calls for an immediate response.

Mitchell and Lee count eight categories by including the non-stakeholders, whom the map places outside the circles.

An attribute is a dated judgement

The model describes what managers perceive (Mitchell and Lee). The same actor can therefore receive two different classes from two teams. A staff committee asks to be consulted on the new on-call schedules before the end of the month. Article 48 of the Employment Act gives workers or their representatives the right to be consulted on the organisation of working time and the arrangement of working hours. The team that does not know this right classes the committee as demanding. The team that knows it recognises the committee's legitimacy and classes it as dependent. In the register, the team that keeps it writes down the fact behind each attribute. Without that fact, a placement can be argued endlessly and cannot be checked.

Twenty years on, Wood, Mitchell, Agle and Bryan restate that the model is descriptive: it says whom managers judge salient. It does not say who deserves their attention. Some critics fault it for suggesting that it identifies the stakeholders who "really count". In a project, the team sets the order of priority itself. It records the reason for that choice in the register.

The proximity variant

The PMBOK Guide cites an adaptation that replaces legitimacy with proximity: the actor's direct involvement in project activities through physical presence on site or through contributions to the deliverables. This variant applies to the team itself and measures each person's involvement in the work. It helps the project manager divide attention between team members and external contributors to the deliverables. A map states which variant it uses and never mixes the two.

Running the analysis

1. Set the subject and the date. Salience varies with the decision at stake: Kapiriri and Razavi observe this in public health priority setting. The analyst writes the decision or scope being judged at the top of the map, then the date.
2. Start from the list. The model classifies the actors of the stakeholder list. The onion diagram helps check that the list covers the outer circles.
3. Judge each attribute separately. For each actor, the analyst asks one question per attribute and writes down the fact behind each yes. By what means can the actor impose its will or block a decision? Is the claim founded on accepted norms: a right, a contract, a mandate or a direct effect of the project on the actor's property or life? Is the claim pressing because of a deadline or because of high stakes for the actor? Two assessors who judge separately and then compare expose the fragile placements.
4. Split heterogeneous groups. "The residents" often lumps together an owner who has filed an objection with tenants who know nothing about it. A group whose members do not share the same attributes splits into several rows.
5. Classify and draw. The combination of yeses gives the class. Each actor takes its place in the matching zone of the three circles.
6. Derive the engagement. The class steers the stakeholder engagement plan. The stakeholder engagement assessment matrix then sets the target level for each actor.
7. Record the triggers. For each actor that a foreseeable event would move to another class, the register records that event and its likely date.

From class to engagement

Mitchell, Agle and Wood describe classes. They prescribe no course of action. The proposed stance for each class responds to the attribute that the class holds or lacks.

Kapiriri and Razavi point out a downside: favouring salient actors can deepen the power imbalance at the expense of latent actors. A discretionary or dependent stakeholder whose life the project changes deserves an engagement that its class does not guarantee.

Redrawing the map

Wood and his co-authors call the model dynamic: the team redraws its map at several points to keep it accurate. An actor changes class by gaining or losing an attribute. It also changes class by joining or leaving a coalition. A dependent residents' association becomes definitive when its members file objections during the public inspection period. A dominant municipal council becomes definitive when the credit vote goes on its agenda.

The team redraws the map at every milestone and at every trigger recorded in the register. It compares the versions and reads each change of class as a warning. A dormant stakeholder that acquires urgency becomes dangerous before it has acted.

What distorts the classification

Confusing legitimacy with agreement. An opponent of the project can have a legitimate claim. Classing the troublesome actor as "not legitimate" produces a false map.

Giving in to urgency. In 1999 Agle, Mitchell and Sonnenfeld published a test of the model among chief executives. All three attributes predicted salience. Urgency was the strongest predictor. A pressing claim draws the attention of managers whether it is well-founded or not. The team therefore checks the legitimacy of the loudest claims.

Forcing an actor into a single class. Kapiriri and Razavi struggled to place some actors in a single class. An ambiguous actor splits into two rows. Failing that, the register records both readings and the fact that would decide between them.

AI considerations

A language model can prepare the register. From minutes, correspondence, filed objections and press articles, it proposes for each actor the facts that support or undermine each attribute. It cites the source of each fact. Between two reviews, it spots signals of a class change: an announced deadline, a coalition forming, a question tabled in the municipal council. It also drafts a first version of the messages for each class.

Legitimacy remains a judgement of the team. It involves norms and values that the language model cannot settle for the team. The model reads only what is written: the informal power of a silent actor escapes it.

The register attaches labels such as "dangerous" to named people. These labels fall under the Federal Act on Data Protection (FADP). Where possible, the team classifies organisations and groups rather than individuals. It does not send the register with names to an AI service whose data processing it does not control. It restricts access to the register.

Examples

A town's municipal utilities are building a district heating network in one neighbourhood. While preparing the public inspection, the project manager and the business analyst draw up the following register at the review of 12 January. For urgency, they apply the rule of the PMBOK Guide: a deadline or high stakes is enough.

The gas distributor imposes no deadline. The stake of 300 customers is enough to ground its urgency. Without legitimacy, it stays in the dangerous class. Three actors carry a trigger: a single event would move each of them into a more salient class. The map shows the register before and after the objections.

During the public inspection period, 40 members of the association each file an objection. The municipality can grant the building permit only after it has ruled on each one. The association gains the power that it lacked. The review of 16 March classes it as definitive. Its claim has not changed.

Visualisations

The salience map is drawn as three overlapping circles: power, legitimacy, urgency. A frame around the circles holds the actors without attributes. For a given review, each stakeholder appears in one zone only. Its short name points to its row in the register.

The header names the decision at stake and the date of each review. When the map compares two reviews, the old position appears faded. An arrow links it to the new position. The map is often printed in black and white for a committee. Each of the seven zones therefore carries the name of its class in full, so that the map reads without colour.

The map shows the positions. The register holds the facts that justify them.

Cost

Tooling

A spreadsheet holds the register: one row per stakeholder, three columns for the facts behind the attributes, a calculated column for the class, one column for the trigger and one for the date. One tab per review keeps the history of the classes.

In the workshop, the team draws the three circles on a physical whiteboard or on an online board such as Miro or Mural. The assessors move sticky notes there and compare their placements.

Stakeholder management software suits an organisation that already keeps its register there, if the software accepts three attributes per actor and keeps their history.

Sources

• Project Management Institute (2025), A Guide to the Project Management Body of Knowledge (PMBOK Guide), 8th edition: §2.5.2.1 Identify Stakeholders for the place of the tool in the process; the "Stakeholder mapping/representation" entry in Section 5 Tools and Techniques for the three attributes, the proximity variant and the use with large communities.
• Mitchell, R. K., Agle, B. R. and Wood, D. J. (1997), "Toward a Theory of Stakeholder Identification and Salience: Defining the Principle of Who and What Really Counts", Academy of Management Review, 22(4), pp. 853-886. The original article of the model.
• Mitchell, R. K., Agle, B. R., Chrisman, J. J. and Spence, L. J. (2011), "Toward a Theory of Stakeholder Salience in Family Firms", Business Ethics Quarterly, 21(2), pp. 235-255: the definitions of salience, power, legitimacy and urgency as the 1997 article sets them out.
• Mitchell, R. K. and Lee, J. H. (2019), "Stakeholder Identification and Its Importance in the Value Creating System of Stakeholder Work", The Cambridge Handbook of Stakeholder Theory: the typology in eight categories, the names of the classes and the salience that grows with the number of attributes perceived.
• Wood, D. J., Mitchell, R. K., Agle, B. R. and Bryan, L. M. (2021), "Stakeholder Identification and Salience After 20 Years: Progress, Problems, and Prospects", Business & Society, 60(1), pp. 196-245: the dynamic and descriptive character of the model, the critiques and the 1999 test.
• Kapiriri, L. and Razavi, S. D. (2021), "Salient stakeholders: Using the salience stakeholder model to assess stakeholders' influence in healthcare priority setting", Health Policy OPEN, 2, 100048: a documented application and its limits.
• Swiss Confederation, Federal Act on Work in Industry, Crafts and Commerce (Employment Act, EmpA), SR 822.11, art. 48: the right of workers or their representatives to be informed and consulted on the organisation of working time and the arrangement of working hours.