---
name: Stakeholder Analysis Frameworks
status: draft
territory: stakeholder-conflict
host_mode: stakeholder-mapping
also_loadable_in: []
msi_wired: false
sources:
  - title: Bryson, John M. (2004), "What to do when Stakeholders matter," Public Management Review 6(1):21-53
    url: https://doi.org/10.1080/14719030410001675722
  - title: Mitchell, Agle & Wood (1997), "Toward a Theory of Stakeholder Identification and Salience," Academy of Management Review 22(4):853-886
    url: https://doi.org/10.2307/259247
  - title: "Freeman, R. Edward (1984), Strategic Management: A Stakeholder Approach, Pitman"
    url: https://openlibrary.org/works/OL4980191W
---

# Stakeholder Analysis Frameworks

## Why it matters

Who ends up at the table — and who gets quietly forgotten — usually decides whether a plan survives contact with reality. These frameworks turn the vague instruction "consider everyone" into a disciplined grid that catches the party you'd otherwise miss.

For example: a transit authority planning a new rail line maps the parties that obviously matter — the investors who fund it, the regulators who approve it, the city that wants it built — courts each of them carefully, and clears every hurdle. Then a neighborhood association nobody had ranked above "monitor" discovers the line will run an elevated track past a school, files suit, draws the press, and the project stalls for two years. The association had almost no formal power. What it had was an unmistakably *legitimate* claim and, the moment the survey stakes went in, sudden *urgency* — and those two things, which the planners never scored, were enough to stop the train. "Consider everyone" had failed precisely because it was vague: it gave no way to notice the quiet party until it was loud.

- **What it reveals.** The full roster of parties with a stake in an outcome — not just the obviously powerful ones, but the marginalized, the latent, and the absent — each scored on *how much* it can affect the result and *how strong* its claim is, rather than lumped together as "the stakeholders."
- **How it changes the read.** You stop asking *"who are the important players?"* — which silently means *the powerful ones* — and start asking three separate questions: who has the **power** to shape this, whose claim carries **legitimacy**, and to whom is it **urgent** — because a party can be near-zero on one and decisive on another.
- **When to foreground it.** Whenever a decision touches several parties with different stakes and you need to know who must be engaged, who only kept informed, and — most of all — who is about to be left off the map entirely.
- **What you'd miss without it.** The quiet-but-legitimate party with no organized voice — the community downstream, the residents under the flight path — who has little power today but can stop the plan the day the claim turns urgent.
- **Where it misleads.** Rank parties on power alone and you rebuild the org chart and call it analysis; chase a clean diagram and you'll drop the low-salience long tail that was the whole point — the frameworks are a discipline against tidiness, not a license for it.

## How to invoke it in Ora

You're looking at a decision, rollout, or institutional change that touches several parties, and you want a disciplined map of who has a stake — who must be engaged, who only kept informed, and crucially who is about to be left off entirely.

Describe the situation and the parties you already know about, and ask:

> "Map the stakeholders in this rail project — who's involved, how much power and legitimacy each one has, and who's affected but isn't in the room."

Ora inventories the parties, plots them on the power-interest grid and the salience classes, names each one's concrete stake, and — the part that earns the analysis — actively scans for the parties you didn't list. The frameworks supply the grid; you supply the situation.

One thing to know: the words *stakeholder map*, *stakeholder analysis*, *who needs to be at the table*, *who has standing*, or naming the **Bryson power-interest grid** / **Mitchell-Agle-Wood salience** are what route you here. Describe the situation that way and Ora supplies the grids itself — you don't need to know the frameworks by name.

Give it the parties you already have in mind, but say so — because the most valuable output is usually the party you *didn't* name. "Here's who I think matters; who am I missing?" is the prompt that puts the analysis to work, since a map that only ranks the parties you already listed has done the easy half of the job.

One thing Ora won't do: rank everyone by power and stop. It scores legitimacy and urgency as separate axes, keeps the low-power-but-legitimate parties on the map instead of trimming them for a cleaner picture, and names the affected-but-absent — because the quiet party in the long tail is the one the discipline exists to catch.

## How it works

A water authority sets out to build a dam. The case is strong: drought-prone farmland downstream, a growing city that needs the supply, investors ready to fund it, and an engineering plan that checks out. The team does its homework on the parties that obviously matter. The state regulators are walked through every environmental filing. The financiers get their projections. The city signs on. Approvals come through; the diggers roll in.

And then a few hundred people in a valley the reservoir will flood — farmers, a couple of villages, families who have worked that land for generations — plant themselves in front of the bulldozers and refuse to move. On any ranking the planners would have drawn, these people were negligible: no capital, no lobby, no seat on any board, nothing the spreadsheet recognized as leverage. What the planners had filed as a rounding error turned out to have two things their grid never measured. Their claim was *legitimate* in a way nobody could wave off — it was their homes. And the day the machines arrived, it became *urgent*. Power they did not have; the other two they had in full, and the other two were enough to halt the project, summon the cameras, and turn a routine build into a years-long standoff.

The mistake wasn't laziness — the team worked hard. The mistake was hidden inside the word *important*. Asked who the important parties were, the planners heard *the powerful ones*, and answered the question they heard. But "who matters here?" is not one question. It is three. How much **power** does a party have to affect the outcome — to fund it, block it, approve it, or break it? How much **legitimacy** does its claim carry — would a fair observer say it has a rightful stake, regardless of its clout? And how **urgent** is the matter to that party — is this a background concern or the thing keeping it up at night? Score those separately and a startling thing happens: a party can sit at near-zero on one axis and be decisive on another. The valley had no power and all the legitimacy and, eventually, all the urgency. Collapse the three into one and you never see it coming.

That three-question discipline is the heart of the salience model that Ronald Mitchell, Bradley Agle, and Donna Wood set out in 1997 — power, legitimacy, urgency, and the eight kinds of party you get from the presence or absence of each. A party with power but no legitimate claim and no urgency is *dormant* — latent, easy to overlook, dangerous if it wakes. One with a legitimate claim but no power and no urgency is *discretionary* — owed attention more than engagement. The valley, once the bulldozers came, was *dependent*: a legitimate, urgent claim with no power of its own, reliant on someone else to act for it. A party that holds all three is *definitive* — it has to be dealt with, now. The model's gift is that it makes the *quiet* combinations visible: the legitimate-but-powerless party that a power ranking files under "ignore" is exactly the one the three-axis view flags as a claim someone will eventually have to answer.

There is a simpler tool for the simpler question. When what you need is not *who deserves a hearing* but *where to spend a finite amount of attention*, a two-by-two grid — power on one axis, how much a party cares about this particular decision on the other — sorts everyone into four boxes: **manage closely** (high power, high interest), **keep satisfied** (high power, low interest — don't wake them needlessly), **keep informed** (low power, high interest — they care, so talk to them), and **monitor** (low power, low interest — a periodic glance). Aubrey Mendelow sketched this grid in 1991; John Bryson sharpened it for public and coalition-heavy settings, where parties slide between boxes as events unfold and today's "monitor" is tomorrow's "manage closely." It is fast and it is genuinely useful — but it reads only two of the three dimensions, which is why the careful analyst keeps the salience model alongside it. Both descend from the idea R. Edward Freeman planted in 1984: that an organization is answerable not only to its shareholders but to everyone who can affect, or is affected by, what it does.

The whole reason these frameworks exist is to stop the quiet-but-legitimate party from being dropped for a tidier diagram. Left to instinct, a stakeholder map drifts toward the parties with letterhead and lawyers, because those are the ones who show up and make noise — and the map ends up a mirror of existing power, which is to say it tells you nothing you didn't already know. The grids are a forcing function in the opposite direction: list everyone, score each on more than clout, and *keep* the low-power, legitimate, not-yet-urgent parties on the page — because the long tail of the map, the parties easiest to trim, is where the surprises that sink plans actually live.

## Framework & implementation

*This section uses Ora's own terms for the parts of an analysis, so that if you open the actual mode and lens files they line up. Each is glossed in plain language on first use.*

### Pipeline execution

Stakeholder Analysis Frameworks is the **catalog lens** of the Stakeholder Mapping analysis — `lens_type: catalog`, the one that supplies the mode's structure rather than merely informing it. It sits in the mode's **`ANALYTICAL PERSPECTIVES`** block under "always loaded," and like Ulrich's boundary categories inside a boundary critique, it is not a supporting model the analysis consults in passing: its two classification frameworks **are** the mode's middle output sections. Stakeholder Mapping's **Power-interest positioning** section is this lens's Bryson/Mendelow two-by-two, filled in; its **Mitchell-Agle-Wood salience classification** section is this lens's three-attribute model, filled in. The other always-loaded models — incentives, signaling, principal-agent-problem, schelling-point, fisher-ury-principled-negotiation, cooperation — supply discipline *within* that structure; this lens supplies the structure itself. The analysis runs at **Gear 4**, Ora's most thorough setting — a **Depth analyst** and a **Breadth analyst** map the parties in parallel, critique each other's reading, and revise under that critique before a consolidator merges what survives. The lens threads through those stages like this.

**Detection.** The lens engages on the cases in its **Detection Signals** — the situation has multiple parties whose interests are not fully aligned; a decision affects heterogeneous groups whose stakes differ in *kind*, not just degree; institutional change is afoot with constituencies both represented and not; the conversation has started treating "the stakeholders" as one homogeneous block; engagement resources are scarce and prioritization is contested; or coalition possibilities are being floated without a structured map of who could ally with whom.

**The Depth and Breadth analysts.** Two models map the situation in parallel and never see each other's work. Both run the lens's **Application Steps**, which use the **Bryson Stakeholder Identification Protocol** as the spine — *identify* parties broadly (including absent and latent ones), *classify* them, *analyze* each one's interests and probable response, *identify* alliances and conflicts, and *develop* a differentiated engagement strategy — and reach for the right framework at the classify step: **Mitchell-Agle-Wood** when the question is *who deserves attention*, the **Bryson/Mendelow power-interest grid** when the question is *where to spend engagement resources*, and a **Stakeholder Influence Diagram** when influence pathways are non-obvious. The **Depth analyst** drives the roster deep — pursuing latent and absent parties, working each party's stake down to concrete interests, and tracing indirect influence chains a flat list misses (the low-salience party who has the ear of a high-salience one). The **Breadth analyst** sweeps the full attribute space so no class is skipped, checking power *and* legitimacy *and* urgency as independent axes rather than collapsing them, and scanning for coalitions across the whole field.

**Cross-adversarial evaluation.** Each analyst's map is handed to the *other* to attack, and this is where the lens's signature failures — keyed to its **Critical Questions** and **Common Failure Modes** — are caught. The evaluator presses: are parties identified by *underlying interest* or merely by *stated position*, which misses the latent ones who haven't declared (the mode's **role-as-stake** and **frame-bounded-inventory** failures, staying inside the roster the user already had in mind)? Are *absent* parties on the map — those who'd be affected but aren't in the room (**silent-power-mirroring**, a map that ranks by existing power and omits the marginalized-but-legitimate)? Is salience scored on all three attributes, or quietly collapsed to power alone (the lens's **power-only proxy** → the mode's **single-axis-salience**, the precise failure the three-dimensional classification exists to prevent — the tell is a ranking that matches the org chart)? Are *conflicts* among parties surfaced, not just each party's interests in isolation (**coalition blindness** — a map with nodes but no edges)? And is the classification re-checked as the situation moves, or frozen at project start (**static salience**)?

**Revision and claim-check.** The reviser repairs what the critique exposes — restoring the affected-but-absent parties, re-scoring any salience that had flattened to power, completing the alliances-and-conflicts pass a draft left half-run, and pushing role-labels down to concrete stakes. Where a reading rests on a checkable factual claim about a party or its standing, that claim is flagged and sent to a web-search tool before the revised draft proceeds. Crucially, the reviser resists the **tidy-diagram bias** — the pull to drop low-salience-but-legitimate parties for a cleaner picture — because the long tail of the map is the mode's value, not its clutter.

**Consolidation and output.** The consolidator organizes everything into the mode's seven set sections: the **Stakeholder inventory** (every party, including the latent and absent); **Power-interest positioning**, this lens's Bryson two-by-two with each party sorted into *manage closely / keep satisfied / keep informed / monitor*; the **Mitchell-Agle-Wood salience classification**, this lens's eight classes from power × legitimacy × urgency — *definitive · dominant · dangerous · dependent · dormant · discretionary · demanding · non-stakeholder*; **Stake per party**, each rendered as *Wants / Could lose / BATNA* (a party's best fallback if no deal is reached); **Relationships among parties**, the alliances and oppositions; **Absent or marginalized parties**, the constituencies with a stake but no seat; and **Confidence per finding**. Sections two and three are this lens, made concrete.

**What the analysis will not do.** It will not collapse salience to power — legitimacy and urgency are scored as independent axes, because a party can hold one and not the others and still be decisive. It will not silently drop a low-salience party for a cleaner diagram; where a party's salience is genuinely marginal, the map says so rather than omitting it. And it will not let the roster stop at the parties already named — an analysis that hasn't actively asked *who is affected by this who is not in the room* is not finished.

### Origin and evidence

The frameworks the lens catalogs come from a well-developed strand of strategic-management and public-administration scholarship. The field's founding text is R. Edward Freeman's *Strategic Management: A Stakeholder Approach* (1984), which established the stakeholder concept as a discipline — the claim that an organization is accountable not only to shareholders but to every party that can affect or is affected by its actions. The two classification frameworks the lens leans on hardest are the heart of the catalog. Ronald Mitchell, Bradley Agle, and Donna Wood's "Toward a Theory of Stakeholder Identification and Salience" (1997) supplied the three-attribute model — power, legitimacy, urgency — and the typology of stakeholder classes those attributes generate, answering the prior question of *who counts as a stakeholder at all* and *how much*. The power-interest grid traces to Aubrey Mendelow's 1991 conference paper "Environmental Scanning: The Impact of the Stakeholder Concept," which sorted parties by power and interest into the manage-closely / keep-satisfied / keep-informed / monitor quadrants; John Bryson's "What to do when Stakeholders matter" (2004) reviewed the full toolkit and sharpened the grid and an operational identification protocol for public and coalition-driven settings, where parties shift between quadrants as events unfold. The lens sits alongside a complementary tradition it explicitly cross-references — Werner Ulrich's Critical Systems Heuristics — for the harder, prior question these grids take as settled: who *should* count as a stakeholder when standing itself is contested.

### Applications and common uses

Stakeholder analysis is a working tool wherever a decision touches several parties with unlike stakes and someone has to decide who to engage, how, and in what order — used both to *map* a field before acting and to *audit* a map for the party it forgot.

- **Public policy and planning.** Its native ground: a policy affects heterogeneous publics whose stakes differ in kind, and the frameworks force the analysis past the organized lobbies to the unrepresented — future generations, non-organized communities, those without procedural standing. The infrastructure project blindsided by a constituency it never scored is the canonical cautionary case.
- **Project and program management.** Before a rollout, the power-interest grid allocates a finite engagement budget — who gets active management, who a periodic update — so effort lands where it changes outcomes rather than spreading evenly and thinly.
- **Strategy and corporate decisions.** Turning the salience model on a merger, a market entry, or a restructuring surfaces which parties are *definitive* (must be dealt with now), which only *dormant* (latent power worth watching), and which *dependent* (a legitimate, urgent claim with no power of its own) — and keeps the analysis from mistaking the loudest party for the most consequential.
- **Negotiation and coalition-building.** A Stakeholder Influence Diagram maps who actually shapes whom — including the indirect chains where a low-salience party has the ear of a high-salience one — so an analyst can see which alliances are possible and where interests collide before sitting down at the table.
- **Conflict and standing disputes.** When the fight is partly about *who even has a rightful claim*, scoring legitimacy as a separate axis from power keeps the powerful from defining the legitimate — and flags the party whose claim is sound even though its clout is not.

In every case the move is the same: list everyone, including the parties no one named; score each on power *and* legitimacy *and* urgency rather than clout alone; map where they ally and collide; keep the quiet-but-legitimate parties on the page; and re-run it when the situation moves.

### Failure modes and when not to use it

The lens's characteristic ways of going wrong are catalogued in its **Common Failure Modes**:

- **Power-only proxy.** Collapsing salience to power alone, ignoring legitimacy and urgency. The tell is a stakeholder ranking that matches the org-chart hierarchy. Re-apply the three-attribute model with legitimacy and urgency scored as independent axes — this is the single failure the salience model most exists to prevent.
- **Stated-stakeholder bias.** Only the parties who self-identify get analyzed; those without an organized voice never make the map. The tell is that the stakeholder list correlates with which parties have communications staff. Actively scan for the affected-but-absent with the question *"who would be affected by this decision who is not in the room?"*
- **Static salience.** Classifying parties once and never revisiting as the situation evolves. The tell is an engagement strategy set at project start and never re-checked, even as a low-salience party acquires power, legitimacy, or urgency overnight. Re-run the classification at named decision points and triggering events.
- **Single-framework reduction.** Using one grid when the situation called for another — power-interest when the question was salience, or vice versa. The tell is a map that answers a question the situation didn't ask while the real question goes unaddressed. Match the framework to the question; combine them when the situation is complex.
- **Coalition blindness.** Listing parties individually without mapping the alliance structure. The tell is a map with nodes but no edges. Run the alliances-and-conflicts step explicitly, or build a Stakeholder Influence Diagram.

**When not to reach for it.** When a single party simply decides and no one else has standing or leverage, the apparatus of grids and salience classes is overkill — there is no field to map. When the question is not *who counts* but *whether a boundary already drawn is the right one* — when standing itself is the contested issue and the worry is that the frame quietly wrote someone out before any party was scored — that is a boundary critique, not a stakeholder map; these frameworks take the question of who counts as answerable, and where it is the whole fight, reach for the critical method built for it. And when the parties are already few, fully present, and clearly identified, the lens's central contribution — surfacing the absent and the misranked — has little to add; move on to negotiating among the parties who are all already at the table.

## Related

- **Stakeholder Mapping** — the analysis this lens structures; inventories the parties, plots their power and salience, names each one's concrete stake, and surfaces the absent.
- **Incentives** — the discipline that keeps each party's *stake* honest inside the same map: not the role on the business card, but the reward the party is actually chasing.
- **Principal-Agent Problem** — the formal frame for one recurring relationship a map surfaces: a party acting on behalf of another whose interests it doesn't fully share.
- **Fisher-Ury Principled Negotiation** — the natural next step once the parties are mapped: separating positions from underlying interests to find agreements that hold.

## Sources

- [Bryson, John M. (2004), "What to do when Stakeholders matter," Public Management Review 6(1):21-53](https://doi.org/10.1080/14719030410001675722)
- [Mitchell, Agle & Wood (1997), "Toward a Theory of Stakeholder Identification and Salience," Academy of Management Review 22(4):853-886](https://doi.org/10.2307/259247)
- [Freeman, R. Edward (1984), Strategic Management: A Stakeholder Approach, Pitman](https://openlibrary.org/works/OL4980191W)