You offered the distributor a margin that beats what they make anywhere else. They said no. You explained the arithmetic again, more carefully. They said no again, and now they are slower to return calls.
They understood it the first time. They are not pricing the money. They are pricing the split.
Why these three models
The decision is what to change after a profitable offer has been refused. The features that fire are a counterparty whose payoff depends on the division rather than only on their own share, a refusal that other people will hear about, and a threshold you cannot discover by asking.
Three lenses. Inequity and reciprocity produce an equilibrium answer about why the refusal is rational. Transmission produces a cycle answer about what the refusal does beyond this deal. The probe produces a random answer about how to find the acceptable split without a conversation that cannot happen. The first explains, the third acts, and the second is why you cannot simply keep testing on the same counterparty.
1. The refusal is rational, and the model is old
Start with the fact that makes the rest of the article necessary. People refuse profitable offers reliably, in controlled conditions, with real money, across many countries and in field settings as well as laboratories.
The behavioural literature gives this formal treatment rather than describing it as an anomaly. Inequity-averse preferences, in the Fehr and Schmidt formulation and the ERC alternative, model a decision maker whose utility falls with the difference between their payoff and others’, so a division that is profitable in absolute terms can be worse than nothing when the gap is large enough.1 Reciprocity models add a second mechanism: payoffs depend on the intention read into the offer, not only on the amounts, which is why the same split lands differently depending on how it arrived.1, 3
Two consequences matter for an operator, and they point in different directions from the usual advice.
The first is that explaining the arithmetic again cannot work. Your counterparty is not confused about their own margin. They have computed it and compared it to yours, and the comparison is the object. More clarity makes the comparison sharper.
The second is that the same total can be acceptable or unacceptable depending on how it is divided and how the division appears to have been arrived at. That is not a loophole. It is where the actual lever is, and it means restructuring beats discounting.
Inequity and reciprocity, the split lens
- Assumes: the counterparty’s payoff depends on the division and on the intention read into it, not only on their own amount.
- Fits because: a profitable offer was refused by someone who understood it.
- Breaks when: the refusal is strategic rather than fairness-driven, meaning they are holding out for more and this framework will flatter them.
- Evidence: grade B. Robustly replicated in direction, with parameters that vary substantially by population and setting.
- Counteracts: re-explaining the value of an offer that was understood.
- May reinforce: conceding to a negotiating tactic dressed as principle.
2. What the refusal does to everyone else
The second lens is why this is more expensive than one lost deal, and it is the part that argues for fixing the structure rather than the instance.
Terms travel. Distributors talk to each other, founders compare notes, and an offer that read as extractive gets described in one sentence to the next person who was considering you. The refusal is not a private event between two parties; it is the first case of something that spreads through a network at a rate set by how well connected that network is.
The practical consequence is asymmetric and worth being clear about. A generous split spreads slowly, because people have less reason to repeat it. An offer perceived as unfair spreads quickly, because it is a better story and it warns people. So the expected cost of a badly structured offer is not the deal you lost, it is the deals that never opened, and you will never see the denominator.
This is also why fixing it one counterparty at a time does not work. A special arrangement for the one who complained is itself information, and what it communicates is that the standard terms are negotiable for people who push.
Transmission, the reputation lens
- Assumes: terms and their perception propagate through a connected population at a rate set by contact.
- Fits because: your counterparties know each other, which in most distribution networks they do.
- Breaks when: the population is genuinely disconnected, where each negotiation is isolated and this lens adds nothing.
- Evidence: grade B. The mechanism is well established and the transmission rate in any specific market is not measurable in advance.
- Counteracts: treating a refusal as a single lost deal.
- May reinforce: paralysis about ever varying terms.
3. Finding a threshold you cannot ask about
The third lens is the practical one, because the obvious research method is unavailable.
You cannot ask a counterparty what split they would accept. The question itself is a negotiating move, the answer is strategic, and asking reveals that you were prepared to offer more. The fairness threshold is inside a system you cannot inspect.
So do not model the mechanism. Manipulate the input and read the output. Beer’s illustration is a child who has no idea how the side of his cot is built and adopts a black box strategy, manipulating the inputs until he obtains the output he wants, without ever understanding the construction.2
The design rule is the half that gets skipped: the most efficient searching procedure is the one offering the highest entropy at each selection.2 Offering slightly better terms to the next counterparty tells you almost nothing, because you can already guess the answer. Offering two materially different structures to two comparable groups tells you a great deal, and it is the only way to locate a threshold nobody will state.
The transmission lens constrains how you can run that. Vary the structure across segments that do not talk to each other, or across time rather than across neighbours. A split test your counterparties can compare is not a test, it is a grievance.
Black box probe, the threshold lens
- Assumes: you can vary the offer and observe acceptance, and nothing about their internal threshold.
- Fits because: the question you would want to ask cannot be asked honestly.
- Breaks when: the counterparties compare notes, which converts the probe into evidence of discrimination.
- Evidence: grade A. An experimental design policy resting on information theory rather than a claim about behaviour.
- Counteracts: guessing at an acceptable split and calling it market rate.
- May reinforce: treating partners as subjects, which they will eventually notice.
The levers, cheapest first
- Stop re-explaining. The arithmetic was understood. Another explanation sharpens the comparison you are losing.
- Show the split, not just their share. If your side is defensible, showing it converts a suspicion into a fact. If it is not defensible, you have learned something before they do.
- Change the shape before the size. A smaller share that rises with volume frequently beats a larger flat one, because it reads as shared upside rather than as a fixed division.
- Make the rule visible and general. Inequity aversion is about comparison, so a published schedule that applies to everyone removes most of the comparison.
- Vary structures across disconnected segments. That is how you find the threshold without generating the grievance.
- Fix the standard, not the complainer. A private concession is information about your standard terms and it travels.
What to do this month
Do now, sized at one afternoon, effect immediate. Write down your side of the split on your three largest partner deals, as your counterparty would see it if they could. Reversible, free, and dominant across every scenario about why the last refusal happened.
Hedge, where the premium is the whole loss. Convert one flat share into a share that rises with volume, on one segment. If the shape was never the issue you have complicated one contract for a cycle, and that is the entire downside.
Defer and trigger, size fixed now. Do not re-paper the partner base. Pre-commit the trigger: the next time a profitable offer is refused by someone who clearly understood it, the response is a structural change to the standard terms rather than a private exception. Decide that now, because in the moment the private exception is always faster.
Note the arrivals. The audit lands today. A structural change lands at the next contracting cycle. What it does to your reputation among the ones who never opened a conversation lands invisibly, over quarters, and cannot be attributed.
What usually happens next
Run the break test first. Has a rule changed, has an actor entered or left, has a measurement become a target? A competitor publishing their partner terms changes the comparison your counterparties are making, and the split that was acceptable last year is now being measured against a visible alternative.
If nothing broke, the pattern is consistent. The refusal is treated as a one-off, a private exception is granted to close it, the exception becomes known, and within two cycles the standard terms are the exception. What was a fairness problem is now a consistency problem, which is harder.
There is a second pattern worth naming because it is the expensive one. A counterparty who accepts a split they consider unfair does not become a happy partner; they become one who under-invests quietly. The deal closes, the numbers look fine for two quarters, and the effort they were never contractually obliged to give simply does not arrive. That failure never appears as a refusal, which is why it is easy to conclude the terms were fine.
Subtract the counterfactual before crediting a fix. A partner who accepted after you improved the split may have accepted because their alternative fell through. The test is whether acceptance rates move across a segment, not whether one person said yes.
What this ensemble cannot see
All three lenses assume the refusal is about fairness. Sometimes it is straightforward leverage.
A counterparty who has read the same literature knows that framing a refusal as a fairness objection is more effective than framing it as a demand, and this framework has no way to distinguish the two. Everything above is equally available to someone who simply wants more and has found a respectable way to ask. The transparency lever offers some protection, because a genuine fairness objection usually survives seeing your side of the split and a tactical one often does not, but that is a heuristic rather than a test.
There is also a magnitude problem. The direction of these effects is well replicated and the parameters are not stable across populations, settings and stake sizes. Nothing here tells you how large a gap becomes unacceptable in your market, which is the number you actually want.
And one property none of these models contains: your own sense of what is fair is doing work throughout, and it was formed by your costs, your risk and your history. The counterparty’s was formed by theirs. Both of you will experience the disagreement as the other side being unreasonable.
The one action that survives the ignorance: before you next respond to a refusal, write your side of the split on one line and ask whether you would sign the other side of it. If you would not, the objection was real and the structure needs changing. If you would, you are probably facing leverage rather than fairness, and the response is different.
Who has to move
The person who needs this is whoever owns partner terms, and their fastest path through a refusal is always a private concession. The cheapest first test is writing your own side of the split down and reading it as the counterparty. It takes an afternoon, and it usually settles whether you are looking at a fairness problem or a negotiation.
Sources and notes
- Sanjit S. Dhami, The Foundations of Behavioral Economic Analysis, Oxford University Press. Part 2 covers other-regarding preferences: chapter 5 surveys the evidence on human sociality including ultimatum and dictator games, gift exchange and trust games, and the question of how representative laboratory evidence is; chapter 6 gives the formal models, including the Fehr and Schmidt model of inequity aversion, the ERC model, and the evidence on social preference and reciprocity. Chapter 8 covers the interaction of incentives and other-regarding preferences, including cases where extrinsic incentives crowd out intrinsic motivation. Note that Dhami’s own treatment reports substantial parameter heterogeneity across studies, and no parameter is claimed here.
- Stafford Beer, Cybernetics and Management, English Universities Press, 1959. The black box strategy, the child and the cot, and the argument that the most efficient searching procedure is the one offering the highest entropy at each selection are in the chapter on the Black Box. Note: the copy consulted is an image-only scan read via optical character recognition, so it is cited qualitatively and no figure is quoted from it.
- Jeffrey Carpenter and Andrea Robbett, Game Theory and Behavior, MIT Press. The ultimatum game and non-cooperative bargaining are chapter 14, and psychological game theory, in which payoffs depend on beliefs and intentions rather than only on outcomes, is chapter 28, covering guilt aversion, intention-based reciprocity and the hold-up problem with vengeful players. Used in section 1 for the claim that the same division lands differently depending on the intention read into it.
A note on the uncomfortable use of section 3. Varying terms across segments to locate a fairness threshold is experimentation on partners who did not consent to being in an experiment. It is defensible when the variation is a genuine attempt to find terms that work for both sides and indefensible when it is a search for the lowest number they will tolerate. The models cannot tell those apart. You can.
Joshua Agonya Pi’Rwot, Founder.