The handshake is the most dangerous moment in a community negotiation. Everyone in the room is relieved, the tension breaks, and the temptation is to treat the deal as done. It is not done. What you have is an agreement in principle, which is a shared intention that has not yet been made specific, ratified, funded, or monitored. Most agreements that later collapse were not badly negotiated. They were badly closed. The parties reached goodwill and stopped, and the goodwill did not survive contact with a change of personnel, a budget cut, or a drought two years on.
Closing is a distinct skill from bargaining, and it is the one most often skipped. I teach negotiation and mediation at the Universidade Catolica Portuguesa, and the closing phase is where I see experienced negotiators relax exactly when they should be tightening. A deal that protects both a community’s livelihoods and a company’s investment is not a warm feeling at the end of a long meeting. It is a document engineered to hold, with a defined process behind it. This piece is about that engineering.
An agreement in principle is not a deal
The gap between a verbal understanding and an enforceable agreement is where most value leaks out. In the room, both sides believe they agreed to the same thing. On paper, months later, they discover they did not. The community understood a commitment to local employment as a guaranteed number of jobs by a fixed date. The company understood it as a good-faith preference, subject to qualified applicants. Both were sincere. Neither wrote it down precisely, so the agreement contained a dispute it had not yet surfaced.
This is why the closing phase has to force specificity, not paper over it. Fisher and Ury, in Getting to Yes, argue that durable agreements rest on objective criteria rather than the will of the stronger party. Closing is where that principle earns its keep. Every commitment has to be pinned to a measurable standard: a number, a date, a named responsible party, and a way to verify it. A promise to “support local development” is not a term. A commitment to fund a named water system to a stated capacity by a stated month, verified by an agreed body, is a term. The difference is the difference between an agreement that can be enforced and one that can only be argued about.
The discipline here is uncomfortable because it reopens things the handshake seemed to settle. That discomfort is the point. Better to surface a hidden disagreement across the table now, while goodwill is high and a mediator is present. Discovering it later, as a grievance when trust has already thinned, costs far more. The practitioner and academic literature on benefit agreements is blunt about this. Vague language is the most common reason these agreements fail in practice. It leaves communities with no standard to hold a company to, and companies with no clear target to meet.
Ratify through the community’s own process
A signature from a few leaders is not consent, and treating it as consent is one of the fastest ways to build an agreement that unravels. In many settings, the people who negotiated the deal are not the people who have to live under it. They may not hold the authority to bind the wider community, even when they hold the title. An agreement ratified only at the top is legitimate on paper and hollow underneath. The moment it requires a sacrifice or a compromise, the members who never endorsed it treat it as something done to them, not something they agreed to.
Closing well means building ratification into the process rather than assuming it. That takes time and it is worth it. The negotiated text has to go back to the community and be explained in plain terms, in the community’s language. People need room to ask questions and raise objections before anything is final. Where a community decides through an assembly, a council of elders, or a broader consultation, the agreement has to pass through that channel and be seen to pass through it. Reading who actually holds that authority is a negotiation task in itself, one I set out in reading the table you’re actually at. Ratification is where a wrong reading of authority finally comes due.
This matters to the company as much as to the community, which is the part that is often missed. A company does not want a signature it will have to defend later against the people it supposedly bound. It wants an agreement the community will enforce from the inside, because its own members understood and accepted it. Genuine ratification is what converts a document into that kind of agreement. The ICMM position statement on Indigenous Peoples now treats consent as an ongoing mutual responsibility that a community can withdraw if terms are breached. An agreement that was never truly ratified is an agreement that was never really given, and it can be taken back the moment it is tested.
Write it so it cannot be read two ways
Once the terms are specific and the deal is ratified, the writing itself becomes a closing task. The goal is a text that reads the same way to the company’s lawyers and to the community’s committee. It has to hold three years from now, when neither of the original negotiators is still in post. That is a high bar, and most agreements do not clear it, because they are written in the language of one side.
Three habits close the gap. First, define every material term inside the document rather than relying on shared memory of what was meant. If “the affected area” matters, draw its boundary in the text or in an annexed map. If “meaningful consultation” is a trigger, state what counts as meeting it. Second, name responsibility and timing for every commitment, so that each obligation has an owner and a deadline rather than floating as a collective good intention. Third, write in both registers where the parties operate in different languages, and treat both versions as authoritative, not one as a courtesy translation of the other. A term that is precise in English and vague in the local language is a term that will be disputed in the local language.
There is a temptation to leave things loose on the theory that flexibility preserves goodwill. In practice, looseness transfers power to whichever party controls interpretation later, which is almost always the company. The community feels this, even when it cannot name it, and its trust in the whole agreement erodes. Precision is not the enemy of the relationship. Precision is what lets both sides rely on the relationship without having to trust each other’s memory. This is also where interest-based drafting beats positional drafting. A term built from the underlying interest tends to survive changed conditions well. Record why the community wanted that job number, and why the company could commit to that date. That term outlasts a split-the-difference figure nobody can now explain.
Build the implementation and monitoring engine
Most agreements are written as if signing were the finish line. The signing is the starting line. An agreement without an implementation structure is a list of promises with no engine to deliver them. It degrades exactly as the research predicts. Personnel who negotiated it move on, and financial pressure invites cost-cutting. No one watches closely enough to catch the drift until it has become a grievance.
The standard remedy, and the one I build into every closing, is a joint body with real authority. A joint implementation committee with equal company and community representation, meeting on a fixed schedule, is that body. It is the venue where commitments are reviewed against reality, complaints are heard early, and disputes are resolved before they escalate. The literature on benefit agreements is consistent on this. A commitment without a monitoring mechanism and consequences for non-compliance is merely an aspiration. Yet monitoring and evaluation are rarely designed in. Closing well means designing them in from the start. Pair the committee with agreed indicators and an agreed cadence of reporting. Where the community’s trust in company data is thin, add joint fact-finding. Then both sides read the same numbers, rather than arguing over whose figures to believe.
This is not a bolt-on. It is the same standard the major frameworks already expect of the company. IFC Performance Standard 1 requires a management program, monitoring and review, adaptive management that responds to changed circumstances, and a grievance mechanism for affected communities. A well-closed community agreement mirrors that architecture on the relationship side. The company that already runs a management system for its environmental commitments has no principled reason to run its social commitments on trust alone. The habits that make grievance and monitoring systems actually resolve conflict, rather than log it, carry directly into how a committee like this should work. I take up those habits in the grievance mechanism that is not enough on its own.
Write the revision triggers before you need them
The last closing task is the one that feels least urgent and prevents the most damage. Every long-term agreement in an extractive setting will be tested by conditions no one foresaw at signing. Commodity prices swing, mine plans change, droughts arrive, leadership turns over, and a new regulation lands. An agreement with no way to adjust becomes brittle. It either freezes both sides into terms that no longer fit, or it breaks, and the break is experienced as betrayal rather than as the foreseeable event it was.
The move is to write revision triggers into the agreement while everyone is still calm. Name the circumstances that will reopen specific terms. Examples are a defined change in production, a shift in the resource being extracted, a material change in law, or the passage of a set number of years. Attach a defined process to each trigger. Reopening a term then becomes an orderly, pre-agreed step, rather than a crisis that reads as one party trying to escape the deal. A community that knows the water commitment can be revisited if the source changes does not experience a later shortage as a company failure. It experiences it as a shared problem the agreement already anticipated, with a path already written for how to handle it.
Designing this in is what separates a deal that holds from one that merely survives until the first surprise. It also protects the investment as much as the livelihoods. A company locked into a term that has become unworkable is exposed as surely as a community whose benefit has quietly evaporated. Franks and colleagues documented how community conflict converts into project cost and delay, and a brittle agreement that snaps under changed conditions is a direct route into that cost. A durable, dual-serving deal builds in its own capacity to bend, so that neither side is trapped when the world moves.
Pressure-test the close before you sign
The tool that accompanies this article is the Deal-Closing Readiness Checklist, a checklist of 17 checkpoints across five sections. The first section, “Specificity of Terms”, walks you through pinning every commitment to a number, a date, a named owner, and a way to verify it. The later sections cover ratification through the community’s own process and writing the text so it cannot be read two ways. They also cover the implementation and monitoring committee, and the revision triggers you set before conditions change. Score each checkpoint In place, Partial, or Absent. Any section carrying two or more Partial or Absent answers is a section where your agreement is not yet ready to sign. A promise there is still vague, unratified, unmonitored, or unable to bend. Run it in the days between the handshake and the signature, which is exactly the window most negotiators waste on relief. Download the Deal-Closing Readiness Checklist.
Why a mediated close holds better than a handshake
The pull toward a quick close is strongest at the end, and that is when a party to the deal is least able to slow it down responsibly. Your own relief works against you, and so does the other side’s. An independent mediator has no such stake in the relief. A skilled neutral keeps the parties in the room long enough to force specificity and to run genuine ratification rather than a top-table signature. The neutral also builds the monitoring and revision machinery while goodwill is still high enough to fund it. The mediator turns the closing phase from a rushed formality into the structured step it needs to be. That is why bringing one in for the close, not only for the conflict, so often pays for itself. This is the shift from positional bargaining to a joint problem-solving posture that Luis Oré Ibarra’s work on extractive negotiation describes, carried all the way through to signature.
Closing well still leaves the question of how the agreement is governed across the years that follow. That is the work of the Social Accord Architecture. The Social Accord Architecture, or SAA, treats a signed accord as a relationship to be run rather than a document to be filed. It carries the ratification, the monitoring committee, the revision triggers, and the dual accountability from the closing table into the daily management of the relationship. The deal keeps protecting livelihoods and investment long after the negotiators have moved on. A well-closed agreement is the foundation. The SAA is how you keep it standing. If you want to pressure-test the close on a specific agreement, reach me at thomas@thomasgaultier.com.