Interest-Based Negotiation

Quick facts

Mechanism: Produces a negotiation preparation map that separates both sides’ positions from their underlying interests, anchors the accept-or-walk decision on a named, valued best alternative inside the mapped zone of possible agreement, and invents options for mutual gain across differently-valued issues before dividing value against objective criteria.

Produces: negotiation-preparation-map

Evidence: P · Domain: Decision and Option Evaluation

Use when agreement is required from a party not under your control and more than one issue is in play (salary, vendor, partnership, resource, or multi-party disputes), and a position fight has hardened before anyone checked whether the interests actually conflict. Preparation deskwork only, not live-table tactics, and not for a single-issue distributive haggle or a choice with no counterparty.

Most negotiation prep fixes on positions - what each side says it wants - and then plans to fight position against position, splitting the difference at the end. Interest-based negotiation refuses that frame. It models the counterparty (the move no solo decision method makes): it surfaces the interests behind the positions on BOTH sides of the table, anchors the accept-or-walk decision on a named, valued best alternative away from the table rather than on hope or the opening offer, and creates value across differently-valued issues before dividing it. The durable move is not the five-element checklist. It is treating the deal as a structured decision against an explicit alternative, with the counterparty’s interests and alternatives modeled, not assumed. The output is a negotiation preparation map: positions and interests for both parties, the best alternative and reservation point, the zone of possible agreement, options for mutual gain built from valuation differences, the legitimacy standards a division can appeal to, and the follow-through an agreement must carry. It is explicitly preparation deskwork - not a script for the live table.

This method consolidates three formulations as one: Fisher and Ury’s principled negotiation (separate the people from the problem, interests not positions, options for mutual gain, objective criteria, the best alternative as power anchor); Raiffa’s decision-analytic skeleton (reservation prices, the zone of possible agreement, the efficient frontier); and the Mutual Gains Approach (prepare, create value, distribute value, follow through - the follow-through adopted here as a map section). The mechanisms coincide; the brand vocabulary is set aside for the descriptive move.

When to Use

• Agreement is required from a party you do not control, and more than one issue is in play - salary packages (pay, start date, scope, remote terms), vendor and partnership contracts (price, term, service level, exclusivity), cross-team resource conflicts, licensing and acquisition terms, multi-party public disputes.
• A position fight has hardened and no one has checked whether the interests are actually opposed. This is the documented failure mode the interests step exists to attack.
• A consequential negotiation is upcoming and the discipline is worth it on its own: naming and valuing your best alternative converts “I need this deal” into a priced choice.

When NOT to Use

• Do not manufacture win-win on a genuinely single-issue, one-shot, distributive haggle. With one issue and no relationship there is nothing to logroll; only the best-alternative, reservation-point, and zone apparatus applies. Inventing imaginary integrative options on a pure price split is theater. Report the honest zone-only read instead.
• Do not assume good faith by default. Against a hard bargainer acting in bad faith, naive openness about interests is exploitable information disclosure (White, 1984, the standard critique of Getting to Yes). Mark which interests are safe to reveal and which to hold; interest disclosure is a calibrated act, not a default.
• Do not use it when there is no counterparty. A choice you alone control is option evaluation, not negotiation - route to think-decision-option-review or think-expected-value-decision-tree.
• Do not emit an accept-or-walk recommendation without a named, valued best alternative. The walk-away anchor is the load-bearing element; without it the structure floats. If no alternative exists, the first move is to develop one (or to recognize you are deciding under duress), not to bargain.
• Do not present the package as validated procedure. The component mechanisms are evidenced; the five-element procedure as a package is practitioner codification (see evidence/dossier.md). The map improves the decision’s structure; it does not guarantee outcomes.
• Do not value-create your way around a power asymmetry. When one side can walk costlessly and the other cannot, the honest output is sometimes “improve your alternatives or do not enter,” not a clever package.

Instructions

When asked to prepare for a negotiation or to build a negotiation strategy, follow these steps:

1. Confirm there is a counterparty and more than meets the eye. State the deal in one line and the parties whose agreement is required. If there is no counterparty, stop and route to a solo decision skill. If it is genuinely single-issue and distributive, say so and run the zone-only path (steps 4-5 only), refusing to manufacture integrative options.
2. Separate positions from interests, on both sides. List each party’s stated position, then the underlying interests it serves. Rank your own interests. Infer the counterparty’s interests and flag each inference’s confidence. For every interest of yours, mark whether it is safe to disclose or should be held - never assume good faith by default.
3. Name and value your best alternative; derive your reservation point. Identify the best alternative to a negotiated agreement (what you do if this deal collapses), value it honestly, and set the reservation point it implies - the worst deal still better than walking. If your alternative is undefined, name developing it as the first action and do not emit an accept-or-walk call yet.
4. Estimate the counterparty’s alternative and bound the zone. Infer the counterparty’s best alternative and reservation point (with confidence flags), then state the zone of possible agreement - the interval between the two reservation points. If the zone is unknown or negative (no overlap), say so plainly; that is a valid and important output.
5. Invent options for mutual gain across differently-valued issues. Find issues the two sides value differently and build candidate trades - what is cheap for you and dear to them against what is dear to you and cheap for them. Create value before dividing it. (Skip on a genuine single-issue distributive case; there is nothing to trade.)
6. Name the objective criteria for dividing value. List the legitimacy standards a division can appeal to - market rate, precedent, an independent appraisal, an industry benchmark - that both parties could accept, rather than relying on willpower or splitting the difference.
7. Specify the follow-through. Name the monitoring, milestones, and dispute-handling an agreement must carry so that “yes” survives contact with reality (the Mutual Gains contribution).
8. Emit the negotiation preparation map per references/TEMPLATE.md, carrying the evidence caveat. Keep it to preparation; decline to script live-table tactics or manipulative moves.

Output Format

Use the template in references/TEMPLATE.md. The deliverable is the filled negotiation preparation map - parties and positions, both-sides interests with confidence and disclose-or-hold flags, the best alternative and reservation point, the zone of possible agreement, options for mutual gain, objective criteria, and follow-through - not a prose essay. The pre-printed evidence-caveat line ships in the artifact. Never script the live table.

Quality Checklist

Before finalizing, verify:

• A counterparty is confirmed; if there is none, the run routed to a solo decision skill instead.
• Positions are separated from interests on BOTH sides; the counterparty’s interests are inferred with confidence flags, and each of your interests is marked disclose or hold.
• A named, valued best alternative and the reservation point derived from it are present; no accept-or-walk recommendation is made without them.
• The counterparty’s alternative is estimated and the zone of possible agreement is stated - including an explicit “unknown” or “negative / no overlap” read when that is the truth.
• Options for mutual gain are built from genuine valuation differences - or, on a single-issue distributive case, the map honestly reports a zone-only read with no manufactured win-win.
• Objective criteria for dividing value and the follow-through commitments are both named.
• Which interests are safe to disclose and which to hold is marked; good faith is not assumed by default.
• No overclaiming: the evidence is practitioner-grade with moderate-grade components, transferred from human studies; claim a preparation aid, not a guaranteed better outcome (see evidence/dossier.md).

Evidence

Tier P (governing). The component mechanisms carry genuine moderate (M) experimental support on human dyads - the fixed-pie misperception is real and pervasive (Thompson and Hastie 1990; Thompson and Hrebec 1996, a review of 32 experiments on lose-lose agreements), alternatives confer measurable power on own and joint outcomes (Pinkley, Neale and Bennett 1994), and integrative behavior raises joint outcomes under prosocial motive plus high resistance to yielding (De Dreu, Weingart and Kwon 2000, a meta-analysis of 28 studies; Pruitt and Lewis 1975). But the prescriptive package itself - the five Fisher-Ury elements, the four Mutual Gains steps - has no controlled validation as a package, and training evidence does not isolate it (Movius 2008 limited and mixed; ElShenawy 2010 moderate effects for training generally). Per this library’s conservative-split rule the governing grade is the lower half, P; grading M would launder component evidence onto the procedure. All evidence is transferred from human dyads; nothing validates this preparation move agent-side (LLM benchmarks such as Bianchi et al. 2024, NegotiationArena, measure bargaining outcomes, not this prep procedure). The skill ships as a preparation aid with hard when-NOT walls, never as a guarantee of a better outcome. Full grading, sources, and caveats: evidence/dossier.md.

Examples

See references/EXAMPLE.md for a completed negotiation preparation map on a real decision.

Deep dive: worked example

A full worked run (the shared Northwind scenario)

Negotiation Preparation Map - Worked Example

A completed run of the interest-based-negotiation skill on a real, consequential decision. This is the quality bar a generated preparation map should meet.

Uses the shared recurring scenario (Northwind, a B2B SaaS weighing a self-serve free-tier launch) so examples across skills read as one coherent product. Where think-scenario-planning builds the uncontrollable external futures the free-tier bet must survive, and think-futures-wheel traces the ripples of launching it, THIS skill prepares a specific negotiation that the free-tier strategy depends on: Northwind needs a cloud-infrastructure partner to underwrite the cost of a generous free tier, and that partner’s agreement is required. See docs/internal/AUTHORING.md.

Evidence caveat (ships with the artifact). This map is a P-tier (practitioner) preparation aid. Its component moves - separating interests from positions, anchoring on a best alternative, creating value across differently-valued issues - rest on moderate-grade evidence from human-dyad studies; the five-element package as a whole has no controlled validation, and none of it is validated for AI-produced preparation. Treat this as a structuring aid that improves the decision, not a guarantee of a better outcome. See evidence/dossier.md.

---

The deal in one line

• Deal / decision under pressure: Northwind needs a 2-year committed-spend agreement with a cloud-infrastructure vendor (“Cirrus”) to underwrite the compute and storage cost of a generous free tier, at terms that make the free-tier unit economics survivable.
• Parties (whose agreement is required): Northwind (buyer of infrastructure) and Cirrus (the infrastructure vendor). Cirrus’s agreement is required - Northwind cannot launch a generous free tier at the list rates.
• Issues in play: price per unit of compute/storage, committed-spend floor, contract term, overage / burst pricing, free-tier “credits” for non-paying users, co-marketing, and a ramp schedule. Multiple issues - this is integrative, not a single-issue haggle.
• Is this single-issue and distributive? No. Several issues are in play and the parties value them differently - the full map applies.

Positions and interests (both sides)

Party	Stated position	Underlying interests	Rank / confidence	Disclose or hold
Northwind	”We want 40 percent off list and no minimum commit.”	(1) Survivable free-tier cost per non-paying user; (2) protection against a cost spike if a free-tier launch goes viral before it monetizes; (3) not locking into a large fixed commit before the free-tier conversion rate is proven; (4) keeping the option to multi-cloud later.	1 and 2 top; 3 high; 4 lower	1-2 safe to disclose (they explain the ask); 3 HOLD (revealing low confidence in conversion weakens us); 4 HOLD (leverage)
Cirrus	”Standard enterprise pricing is 15 percent off list with a 2-year, high-floor commit.”	(1) Predictable committed revenue and a reference logo in a hot category; (2) growth in Northwind’s consumption over time (land-and-expand); (3) avoiding a deep per-unit discount that resets their internal price book; (4) a multi-year lock to amortize onboarding.	(1) and (4) high confidence; (2) high; (3) medium	n/a (inferred)

(Each Cirrus interest is an inference, flagged. Northwind’s interest 3 - low confidence in its own conversion rate - is the most dangerous to disclose: it both weakens the price ask and invites a higher floor. Held.)

Best alternative and reservation point

• Northwind’s best alternative away from the table: sign a 1-year, lower-commit deal with a second-tier cloud vendor (“Nimbus”) at roughly 25 percent off list, accepting weaker regional coverage and a migration cost later. This is real and quoteable, not “walk away.”
• Value of the alternative: acceptable but inferior - Nimbus’s coverage gaps would degrade the free-tier experience in two key regions, and a later migration to Cirrus would cost an estimated quarter of engineering time.
• Northwind’s reservation point: any Cirrus deal worse, all-in, than the Nimbus alternative (roughly: worse than ~25 percent effective discount once burst protection and the migration-avoidance value are priced in) is a walk.

The zone of possible agreement

• Cirrus’s estimated alternative: lose the deal and the reference logo to Nimbus or to a hyperscaler’s startup program; their alternative is weak because the category reference matters to them (medium-high confidence).
• Cirrus’s estimated reservation point: they will not go below a discount that resets their price book publicly, but they can give value through credits, burst caps, and ramped commits that do not touch headline per-unit price (medium confidence).
• Zone of possible agreement: likely positive. Northwind walks below ~25 percent effective value; Cirrus resists headline discounts past ~15 percent but has room on non-price terms. The overlap lives in the NON-price issues - credits, burst protection, a ramped commit - which is exactly where value creation happens. Headline-price-only, the zone is thin; widened by the issues below, it opens.

Options for mutual gain (value creation)

Option / trade	What Northwind gives (cheap for it)	What Northwind gets (dear to it)	Why it is mutual gain
Ramped commit	A 2-year term and a commit floor that STEPS UP as conversion proves out (cheap once conversion is real; protects Cirrus’s lock interest)	Avoids a large fixed commit before conversion is proven (interest 3) and gets a better effective rate at the higher tier	Cirrus gets the multi-year lock and land-and-expand; Northwind avoids the early over-commit
Free-tier credits, not discount	Accepts headline price near Cirrus’s price book	A pool of low-cost “free-tier credits” that absorb non-paying-user cost without resetting Cirrus’s public price	Cirrus protects its price book (interest 3); Northwind protects free-tier unit economics (interest 1) - the same dollar, different ledger
Burst cap	A modest annual minimum	A hard cap on overage pricing if a launch goes viral pre-monetization	Cheap for Cirrus to grant (tail risk), high-value for Northwind (interest 2)
Co-marketing	A joint case study and a conference logo	A small additional discount or credit top-up	Cheap for Northwind, valuable to Cirrus’s reference interest

Objective criteria for dividing value

• Published cloud list pricing and standard enterprise discount bands (market rate).
• A comparable committed-spend deal at Northwind’s stage and volume (precedent / comparable).
• A third-party cloud-cost benchmark for the workload profile (independent benchmark).
• The Nimbus quote as a concrete reference point for effective rate (legitimate alternative-based standard).

Follow-through (so “yes” survives contact with reality)

• Monitoring / milestones: quarterly consumption review against the ramp; conversion-rate checkpoint at month 6 that gates the next commit step.
• Dispute-handling: a named escalation path and a true-up mechanism if actual usage diverges sharply from the ramp.
• Review / renegotiation triggers: a re-pricing window if Northwind’s volume crosses a defined threshold early, and a coverage-SLA review if a new region is needed for the free tier.

The accept-or-walk read

Press for a deal in the zone, but win it on the non-price terms, not the headline discount: a ramped commit plus a free-tier-credit pool plus a burst cap delivers Northwind’s real interests (survivable free-tier cost, spike protection, no premature over-commit) while leaving Cirrus’s price book and lock interests intact. Hold the conversion-confidence and multi-cloud interests. If Cirrus refuses both burst protection and credits, the Nimbus alternative is genuinely better all-in - walk to it rather than signing a high-floor, full-price commit that puts the free-tier economics underwater. A named, valued best alternative (Nimbus) is in hand, so this accept-or-walk call is grounded.

---

Note how this differs from its neighbors on the same Northwind decision. The think-scenario-planning example builds four alternative external futures Northwind does not control and asks which free-tier moves survive all of them - no counterparty is modeled. The think-decision-option-review family scores options Northwind alone chooses among. This skill is the only one that models another party: it surfaces Cirrus’s interests and alternatives, anchors Northwind’s accept-or-walk on a named alternative (Nimbus), maps the zone where a deal beats no deal for both, and constructs trades from the fact that the two sides value the issues differently. The deliverable is a preparation map for a two-party decision, not a solo evaluation and not a forecast.

Grounding: the full evidence dossier

What the research does and does not show, with graded sources

Evidence Dossier: Interest-Based Negotiation

The single source of truth for the interest-based-negotiation skill. The SKILL.md, the sidecar (skill.meta.yml), the templates, and the eval cases all derive from this file. If a claim is not here, it does not belong in the skill. Reformatted from the vetted research dossier (_local/proposed-builds/interest-based-negotiation/dossier.md) and admitted as a Build at the governing tier P.

Skill	thinking-framework-skills.interest-based-negotiation (installable name think-interest-based-negotiation)
Family	decision-and-option-evaluation
Evidence tier	P governing (the component mechanisms carry M human-dyad evidence; the prescriptive package has no controlled validation as a package - see “What the evidence shows”)
Confidence	Moderate that interest-elicitation and a named, valued walk-away alternative improve the structure of a negotiated decision; low that any specific outcome effect transfers to an agent-produced prep map
Status	cand (admitted from the v0.7.0 phase-2 tranche; governing tier P confirmed, not laundered to M)

---

1. The mechanism (what actually does the work)

Interest-based negotiation is a structuring method for any decision that requires another party’s agreement. The durable cognitive move is threefold:

1. Separate stated positions from underlying interests, on both sides of the table. A position is what a party says it wants; an interest is the need the position is meant to serve. The move is to surface the interests behind the positions for yourself and to infer them for the counterparty (with confidence flags), rather than fighting position against position.
2. Anchor the accept-or-walk decision on your best alternative to a negotiated agreement, not on hope or the opening offer. The best alternative away from the table sets a reservation point - the worst deal still better than no deal. Estimating the counterparty’s alternative and reservation point bounds the zone of possible agreement: the interval between the two reservation points where any deal beats no deal for both sides.
3. Invent options for mutual gain before dividing value, then divide against objective criteria. Because the two parties usually value the issues differently, there is room to trade what is cheap for you and dear to them against what is dear to you and cheap for them (logrolling). Value is created first; only then is it divided, and the division is tested against legitimacy standards both sides can accept (market rate, precedent, an appraisal) rather than against willpower.

Three formulations are consolidated here as one method. Fisher and Ury’s principled negotiation (Getting to Yes, Harvard Negotiation Project, 1981) is the canonical practitioner codification: separate the people from the problem, focus on interests not positions, invent options for mutual gain, insist on objective criteria, with the best alternative as the power anchor. Howard Raiffa’s decision-analytic treatment (The Art and Science of Negotiation, 1982) is the formal skeleton: reservation prices, the zone of possible agreement as the interval between them, and the efficient frontier of agreements no joint improvement can beat. The Mutual Gains Approach (Lawrence Susskind and the Consensus Building Institute; Movius and Susskind, Built to Win, 2009) repackages the same core as four steps - prepare, create value, distribute value, follow through - whose distinct addition is the explicit follow-through step: building monitoring and dispute-handling into the agreement itself. The mechanisms coincide; this is one entry, one skill, with the follow-through adopted as a section of the artifact rather than a separate method.

The output is a negotiation preparation map, not prose: the parties and their stated positions; each side’s underlying interests (yours ranked, theirs inferred with confidence flags and a disclose-or-hold mark); your named, valued alternative away from the table with a reservation point derived from it; the counterparty’s estimated alternative and the resulting zone bounds; candidate options for mutual gain built from differently-valued issues; the objective criteria a division can appeal to; and the follow-through commitments an agreement must carry. The scope is this deskwork - preparation, modeling, and option design. Live-table conduct (phrasing, concession timing, responding to tactics) is explicitly out of scope.

2. Lineage

The modern form comes out of the Harvard Negotiation Project: Roger Fisher and William Ury, Getting to Yes: Negotiating Agreement Without Giving In (Houghton Mifflin, 1981; second edition with Bruce Patton, 1991), which coined the best-alternative concept and fixed the five-element vocabulary. Howard Raiffa, The Art and Science of Negotiation (Belknap/Harvard, 1982), is the decision-analytic companion: reservation prices, the zone of possible agreement, efficient frontiers - the formal structure under the prose. Lawrence Susskind carried the approach into multi-party public disputes (Dealing with an Angry Public, with Patrick Field, 1996) and, with Hallam Movius, into organizational practice as the four-step Mutual Gains Approach (Built to Win, Harvard Business Press, 2009; the Consensus Building Institute maintains the canonical description). The standard critique is James J. White, “The Pros and Cons of Getting to Yes” (Journal of Legal Education 34, 1984), on the underplayed distributive core of real bargaining. The experimental tradition runs through Dean Pruitt (with Steven Lewis, 1975), Margaret Neale and Max Bazerman (Cognition and Rationality in Negotiation, 1991), Leigh Thompson (with Reid Hastie 1990; with Dennis Hrebec 1996), and Carsten De Dreu (with Laurie Weingart and Seungwoo Kwon, 2000).

The terms “principled negotiation,” “best alternative to a negotiated agreement,” “zone of possible agreement,” and “interest-based negotiation” are generic vocabulary with named academic attribution; “Mutual Gains Approach” is associated with the Consensus Building Institute but is documented descriptively here. Nothing is trademark-gated. The skill ships under the descriptive mechanism name, crediting the lineage rather than branding it.

3. What the evidence shows, and what it does NOT show

The honest governing grade is P (practitioner). The full read is a split: the component mechanisms the method is built on carry moderate (M) experimental support, but that evidence sits on adjacent component claims measured in human dyads, while the prescriptive package itself - the five Fisher-Ury elements, the four Mutual Gains steps - has no controlled validation as a package. Under this library’s conservative-split rule (and the phase-1 precedent of downgrading candidates whose M evidence sat on adjacent claims), the governing tier is P, not M. Grading M would launder component evidence onto the procedure.

What the record supports, component by component.

• The misperception the interests step attacks is real and pervasive. Thompson and Hastie (1990, Organizational Behavior and Human Decision Processes 47: 98-123) documented that negotiators systematically assume the counterparty’s interests directly oppose their own (the fixed-pie error) and that this misperception predicts worse outcomes. Thompson and Hrebec (1996, Psychological Bulletin 120: 396-409) reviewed 32 experiments on the lose-lose effect: parties in interdependent decisions repeatedly settle on outcomes worse for both sides than an available alternative, by failing to discover compatible interests. (M, component.)
• Alternatives confer measurable power. Pinkley, Neale and Bennett (1994, Organizational Behavior and Human Decision Processes 57: 97-116) manipulated alternatives in dyadic negotiation: possessing an alternative increased the holder’s own outcome and the joint outcome; the more attractive the alternative, the larger the effect; the relatively better alternative claimed the larger share. This is the direct experimental footing for anchoring the accept-or-walk decision on the best alternative. (M, component.)
• Integrative behavior produces higher joint outcomes under specifiable conditions. Pruitt and Lewis (1975, Journal of Personality and Social Psychology 31: 621-633) showed in lab dyads that information exchange about preferences and a flexible-on-means, firm-on-goals stance produce more integrative agreements. De Dreu, Weingart and Kwon (2000, Journal of Personality and Social Psychology 78: 889-905), a meta-analysis of 28 studies, found negotiators with prosocial motives engaged in more problem solving and achieved higher joint outcomes, but only when resistance to yielding was high - the empirical echo of “soft on the people, hard on the problem.” (M, component and meta-analytic.)
• Training in these methods helps moderately, without isolating the package. ElShenawy (2010, Journal of European Industrial Training; six meta-analyses over 57 lab experiments from 36 studies) found negotiation training improves performance, with a medium effect on joint performance (r = 0.37) and larger effects for longer training. Movius (2008, Negotiation Journal 24: 509-531) reviewed the field and concluded systematic evidence on training effectiveness is limited and mixed across outcome levels. Neither isolates principled negotiation as the active package. (M for training generally; nothing package-specific.)

What the record does NOT support. No controlled study tests the Fisher-Ury or Mutual Gains package as a package against an alternative preparation procedure. The famous prescriptions are practitioner codifications whose component mechanisms happen to be well-evidenced, which is not the same thing. The normative superiority of the integrative style is itself contested for distributive contexts (White 1984). No outcome claim of the form “principled negotiators do X percent better” exists in this literature with a nameable primary source, and none is asserted here.

4. Transferred-evidence flag (required honesty for this library)

All of the above is human-dyad evidence. None of it studies whether an interests-and-alternative preparation procedure produced by or with an AI agent improves an agent’s or a user’s negotiation results. On the agent side, LLM negotiation benchmarks exist - Bianchi and colleagues (2024, ICML), NegotiationArena, measured LLM-vs-LLM bargaining and found outcome shifts of around 20 percent from behavioral tactics plus human-like irrationalities - but they measure bargaining performance and exploitability, not whether this preparation procedure improves results. The evidence is transferred from human contexts and not validated for AI-augmented use, which independently caps the grade at P. The agent value is mechanical and modest: an agent makes the prep cheap to run, forces the discipline (both-sides interest elicitation, a named and valued alternative, an explicit zone read, logrolling candidates built from real valuation differences, named legitimacy standards), and produces a durable, inspectable artifact - benefits that do not depend on any contested outcome claim. The skill ships honestly as a P-tier preparation aid, never as a guarantee of a better outcome.

5. When it works / when it fails (drives the eval negative cases and “When NOT to Use”)

Works best when:

• Agreement is required from a party you do not control, and more than one issue is in play (salary packages, vendor and partnership contracts, cross-team resource conflicts, M&A and licensing terms, multi-party public disputes).
• A position fight has hardened and the parties have not checked whether their interests are actually opposed - the documented failure mode the interests step exists to attack.
• A consequential negotiation is upcoming and simple preparation discipline is worth it: knowing your best alternative converts “I need this deal” into a priced choice.

Fails or misleads when (poor-fit / anti-patterns):

• The negotiation is genuinely single-issue, one-shot, and distributive. With one issue and no relationship, there is nothing to logroll; only the alternative, reservation point, and zone apparatus applies. Inventing imaginary win-win on a pure price haggle is theater; the method must report the honest zone-only read instead of manufacturing integrative options that do not exist.
• The counterparty is a hard bargainer acting in bad faith. Naive openness about interests is exploitable information disclosure. White (1984) presses exactly this: the book underplays the distributive core of much real negotiation. Interest disclosure is a calibrated act, not a default; the prep map must mark which interests are safe to reveal and which to hold, and must never assume good faith by default.
• There is no counterparty. A choice you alone control is option evaluation, not negotiation - route to decision-option-review or expected-value-decision-tree.
• Your best alternative is undefined or unexamined. The walk-away anchor is the load-bearing element; without it the whole structure floats. If no alternative exists, the first move is to develop one (or to recognize you are deciding under duress), not to bargain. The skill refuses to emit an accept-or-walk recommendation without a named, valued alternative.
• The package is treated as validated procedure. The component mechanisms are evidenced; the five-element procedure as a package is practitioner codification. The prep map improves the decision’s structure; it does not guarantee outcomes and must never be presented as if controlled research certified the recipe.
• Power asymmetry makes the alternative the whole game. When one side can walk costlessly and the other cannot, value-creation work may be moot; the honest output is sometimes “improve your alternatives or do not enter.”

6. Output artifact

The skill must emit a negotiation preparation map, not prose: the parties and their stated positions; each side’s underlying interests (yours ranked; theirs inferred with confidence flags and a disclose-or-hold mark per interest); your named, valued best alternative with the reservation point derived from it; the counterparty’s estimated alternative and the resulting zone-of-possible-agreement bounds (or an explicit “zone unknown / negative” read); candidate options for mutual gain built from differently-valued issues; the objective criteria or legitimate standards a division can appeal to; and the follow-through commitments an agreement must carry. A pre-printed evidence-caveat line ships in the artifact by construction. The map is preparation deskwork; live-table conduct is out of scope.

7. Why it is a skill here (distinctness)

Verdict: Build, confirming the v0.7.0 phase-2 reconciliation (cand / build / P). The Build burden is to name the distinct durable move and show no shipped skill or chain produces it. The decisive structural fact: the catalog contains zero methods that model a counterparty. Every shipped decision skill evaluates options the decider alone controls. This method’s machinery - both-sides interest elicitation, best-alternative valuation and reservation points, zone estimation, value creation across differently-valued issues, principled value division, follow-through design - exists nowhere in the catalog, singly or in combination.

• Not contradiction-resolution (the nearest in spirit). Both reject splitting the difference in favor of inventing a both-win resolution, which is the visible kinship. But contradiction-resolution operates on a requirements contradiction inside a system the decider controls, via an Ideal Final Result and the separation principles; there is no counterparty whose interests must be modeled, no alternative away from the table, no zone, and no distribution step because there is no one to distribute to. The shared element - generate an option that satisfies both demands - is one sub-step of this method’s seven-part canvas, well under the roughly 20 percent overlap ceiling.
• Not decision-option-review. The objective-criteria element superficially echoes criteria-based evaluation, but here the criteria are legitimacy standards two parties can both accept for dividing value, not a solo weighted scoring matrix. Decision-option-review has no interests model, no walk-away anchor, and no counterparty.
• Not expected-value-decision-tree. Valuing an uncertain alternative can borrow expected-value mechanics for one cell of the canvas; the tree itself has no counterparty interests, no value creation, and no agreement structure. Complementary, not substitutable.
• Not boundary-critique or parallel-perspectives-review / role-storming. Boundary-critique audits who was counted as affected and makes no deal; perspective and persona methods neither value alternatives away from the table nor construct tradeable packages from valuation differences. Inhabiting the counterparty’s perspective could feed the interests-inference step, but the inference is an input, not the mechanism.
• No chain of shipped skills produces the move. Any sequence still lacks the best-alternative/reservation/zone apparatus and the logrolling step - the two elements the experimental record specifically supports (Pinkley, Neale and Bennett 1994; Pruitt and Lewis 1975).

The method clears all four commitments: a durable mechanism named descriptively (no brand required), an honest P grade with M-grade component evidence, a concrete reusable artifact (the negotiation preparation map), and a sharp set of when-NOT walls. The live negotiation is out of scope by design, which keeps the skill inside what an agent can actually do.

8. Sources

1. Roger Fisher and William Ury, Getting to Yes: Negotiating Agreement Without Giving In (Houghton Mifflin, 1981; 2nd ed. with Bruce Patton, 1991). The canonical practitioner codification: interests over positions, options for mutual gain, objective criteria, the best alternative as power anchor. Prescriptive, not controlled research. (P, foundational.)
2. Howard Raiffa, The Art and Science of Negotiation (Belknap/Harvard, 1982). The decision-analytic formalization: reservation prices, the zone of possible agreement, the efficient frontier. Foundational analysis, not an outcome study. (P, foundational.)
3. Leigh Thompson and Reid Hastie, “Social Perception in Negotiation,” Organizational Behavior and Human Decision Processes 47(1) (1990): 98-123. The fixed-pie misperception documented; misperceiving the counterparty’s interests predicts suboptimal outcomes. (M, component.)
4. Leigh Thompson and Dennis Hrebec, “Lose-Lose Agreements in Interdependent Decision Making,” Psychological Bulletin 120(3) (1996): 396-409. Review across 32 experiments: parties repeatedly settle on outcomes worse for both than an available alternative by failing to discover compatible interests. (M, component review.)
5. Robin Pinkley, Margaret Neale and Rebecca Bennett, “The Impact of Alternatives to Settlement in Dyadic Negotiation,” Organizational Behavior and Human Decision Processes 57(1) (1994): 97-116. Alternatives increase own and joint outcomes; attractiveness and relative strength scale the effect. The best-alternative evidence. (M, component.)
6. Dean Pruitt and Steven Lewis, “Development of Integrative Solutions in Bilateral Negotiation,” Journal of Personality and Social Psychology 31(4) (1975): 621-633. Information exchange and flexible rigidity produce integrative agreements in lab dyads. (M, component.)
7. Carsten De Dreu, Laurie Weingart and Seungwoo Kwon, “Influence of Social Motives on Integrative Negotiation: A Meta-Analytic Review and Test of Two Theories,” Journal of Personality and Social Psychology 78(5) (2000): 889-905. Meta-analysis of 28 studies; prosocial motive plus high resistance to yielding yields more problem solving and higher joint outcomes. (M, meta-analytic, component.)
8. Hal Movius, “The Effectiveness of Negotiation Training,” Negotiation Journal 24(4) (2008): 509-531. Systematic evidence on training effectiveness is limited and mixed; nothing isolates the principled-negotiation package. (P, review.)
9. Eta ElShenawy, “Does Negotiation Training Improve Negotiators’ Performance?,” Journal of European Industrial Training 34(3) (2010). Six meta-analyses over 57 lab experiments from 36 studies; training has a medium effect on joint performance (r = 0.37), larger for longer training. (M, training generally, not package-specific.)
10. James J. White, “The Pros and Cons of Getting to Yes,” Journal of Legal Education 34 (1984): 115-124. The standard critique: the book underplays distributive bargaining; openness is exploitable. Feeds the when-NOT wall. (Critique.)
11. Consensus Building Institute, “CBI’s Mutual Gains Approach to Negotiation”; Hallam Movius and Lawrence Susskind, Built to Win (Harvard Business Press, 2009). The four-step Mutual Gains Approach: prepare, create value, distribute value, follow through. Practitioner/consulting lineage. (P.)
12. Federico Bianchi, Patrick John Chia, Mert Yuksekgonul, Jacopo Tagliabue, Dan Jurafsky and James Zou, “How Well Can LLMs Negotiate? NegotiationArena Platform and Analysis,” ICML 2024. LLM-vs-LLM bargaining benchmark; measures outcomes and exploitability, not this preparation procedure. Context for the agent-transfer caveat. (Agent context.)

Excluded on the evidence rule: no specific decision-outcome or “X percent better” effect size for the principled-negotiation package is asserted as fact, because no robust, replicated, primary-sourced figure exists on the package as a package. Component effects (the fixed-pie misperception, the power of alternatives, integrative behavior under prosocial motive) are reported with their named primary sources and their human-dyad scope. The NegotiationArena ~20 percent figure is reported as a bargaining-tactics result, explicitly not as evidence for this prep procedure.