Compensation Without a Boss

Pay is where most self-managing organizations discover the limits of their commitment. Everything else can be distributed. Then someone asks what they should earn, and the group finds out whether it meant it.

There is no version of this that is comfortable. What follows is the four approaches that actually work, and what each one costs, so you can choose deliberately instead of drifting into the one that happens by default.

Before you choose: two things to settle

Transparency and process are separate questions. You can have secret salaries set by a group, or open salaries set by a founder. Decide them independently, because conflating them is how the conversation gets stuck.

Nearly everyone underestimates how much pay carries. Pay is not only money. It is a statement about worth, and it will be received that way no matter how carefully you frame it as market rate. Any process that treats it as a pure calculation will hurt people in ways the designers did not intend.

The four approaches

1. Formula

Pay is determined by a published rule: years of experience, a role band, a cost-of-living multiplier, tenure. Same inputs, same output, for everyone.

Costs you: nuance. Two people in the same band contribute very differently and are paid identically. Your best person can and eventually will leave for somewhere that will pay them for being exceptional.

Buys you: no negotiation, so the person most comfortable asking for money does not out-earn the person who is not. That is a much bigger effect than most groups realize, and it maps directly onto who is confident, who is senior, and often onto gender and background. A formula is the single most effective anti-bias mechanism available, and it is why many groups accept the loss of nuance.

Suits: groups that value fairness over retention of outliers, and groups without the maturity for peer conversations about contribution.

2. Self-set with advice

You propose your own salary. You seek advice from a defined set of people. You then set it, and both the number and the reasoning are visible to everyone.

Costs you: it rewards self-advocacy, which is the exact bias a formula removes. Without strong norms, confident people drift upward. It also puts real weight on individuals, and some people find it genuinely distressing.

Buys you: the most honest conversations about contribution that any of these produce, because you have to argue your own case to peers who know your work.

Suits: small groups with high trust and enough psychological safety that giving hard advice is normal. Below that bar it fails badly.

Make it survivable: publish the advice given, not just the outcome. Require advice from at least one person who will disagree with you. Set a floor and a ceiling so the range is bounded.

3. Peer panel

A small elected group sets pay for everyone, using published criteria, rotating annually.

Costs you: it recreates a small hierarchy, and being on the panel is a real burden that changes how people relate to you.

Buys you: consistency across people, and a decision made by people who actually know the work, unlike a distant HR function.

Suits: groups of roughly 15 to 60, where self-set stops scaling and a formula is too blunt.

4. Same pay

Everyone gets the same, or the same within a very narrow band, sometimes adjusted for dependents or need.

Costs you: you cannot hire scarce specialist skills at market rate, and you will lose people whose market value is far above your number. That is a strategic constraint, not a detail.

Buys you: the argument ends. All the energy that goes into pay conversations goes elsewhere. And it makes an unambiguous statement about what the organization believes.

Suits: mission-dense organizations where people have accepted the trade knowingly, and small groups with similar market values.

Transparency: the part people avoid

Open salaries are uncomfortable for about two months and then become normal. Secret salaries are comfortable indefinitely and quietly corrosive, because pay gaps exist whether or not they are visible, and invisible ones cannot be fixed.

If you open the books, do it in this order:

1. Publish the process before publishing any numbers. People need to understand how a number was arrived at before they see it, or they will read every gap as a judgment.
2. Publish ranges by role, then individual numbers.
3. Fix the indefensible gaps first. You will find some. Every organization does. Fix them before publication, not after, and tell people you did.

Do not open salaries in the same month you change the process. One at a time.

Where it goes wrong

Deciding pay in a meeting with everyone present. Whatever the intent, this becomes a public evaluation of a person's worth in front of their colleagues. Use a defined process with a small group or written advice.

Never revisiting. Pay set once and left drifts against market and against contribution, and the gap becomes an exit rather than a conversation.

Pretending the founder is not deciding. In many young organizations the process is real for everyone except the founders, who set their own compensation by a different route. Everyone knows. Either subject it to the same process or state plainly that it is different and why.

Confusing pay with appreciation. If the only way your organization says "your work matters" is money, every pay conversation carries far more than it can hold. Build other channels and pay gets easier.

Copying a famous company's model. The model is downstream of that company's size, sector and history. Choose from the costs, not from the case study.

What good looks like

A person can explain why they earn what they earn, and why a colleague earns more, without resentment, because they know the process and believe it was applied.

Someone raises pay as a tension in a normal meeting, and it is neither dramatic nor taboo.

Try this week

Do not change anything yet. Just write down, honestly, how pay is decided in your organization today, including who really decides.

If what you write differs from what you would tell a candidate, you have found the actual problem, and it is not which model to adopt.