# Server commission: three models that work

A percentage of what the table spends, a flat rate per table, or a mix of the two. Each model produces a behaviour, and not always the behaviour you wanted.

- **Published:** 28 February 2026
- **Reading time:** 5 minutes
- **Category:** Management
- **Author:** The Oasys team

The way you pay your servers decides how your servers sell. That sounds obvious, but almost every owner treats commission purely as a cost, a line on the payroll. Commission is incentive design. Without realising it, you are programming your team's behaviour on the floor. And you get exactly what you incentivise.

## Model 1, a percentage of what the table spends

The server earns a percentage of what their table spent. It is the most common model and the most direct: sell more, earn more.

The behaviour it produces is active selling. The server suggests the starter, offers the second round, remembers dessert. That raises the average check, and that is the good side.

The side nobody talks about: the server also pushes. A table that only wanted one draught beer can get an insistence that grates. And low-spending tables become less attractive to the server, who prioritises whoever spends more. You gain sales, but you can lose on the experience of the people who spend little.

## Model 2, a flat rate per table or per shift

The server earns a fixed amount, per table served, per shift worked, or a combination of the two. What they earn does not depend on how much the table spends.

The behaviour here is stability. Every table gets the same attention, because none is worth more than another to their pocket. The server's income is predictable, which helps you hold on to the team.

The cost is the lost incentive. With no upside on sales, the server writes down what the customer asks for and that is that. Nobody suggests the second glass. The average check tends to sit still.

## Model 3, mixed

A fixed base plus a percentage of what the table spends. It is an attempt to take the best of both: the stability of the flat rate with the incentive of the percentage.

The behaviour is balanced. The server has predictable income from the base and still sees a gain in selling well. The low-spending table is not abandoned, because the base is already guaranteed, but active selling still pays off. For most venues that want to encourage selling without turning it into pressure, this is the healthiest design.

## How to choose

The question is not which model is best. It is what you want the server to optimise for.

If your goal is to pull the average check up in a high-traffic venue with a good margin per item, the percentage does the job. If your goal is a consistent experience and a good team atmosphere, a restaurant built on regulars where the customer comes back for the relationship, the flat rate or the mixed model protect that better. If you do not know, start with the mixed model: it errs less in both directions.

Worth remembering the Brazilian context: the 10% service charge is a tip, not a commission, and how it gets split is a separate conversation. What we are dealing with here is the incentive that comes out of your own pocket, and that shapes the behaviour the customer feels at the table.

## Why we are writing about this

Oasys records sales by server, table by table. Any of the three models becomes auditable: you see who sold what, you close commission with no spreadsheet and no argument, and you see in practice which behaviour your model is producing.

The commission model is a management decision, not an HR one. Choose the behaviour. The number follows.

## See also

- [All insights](https://oasysai.com.br/insights)
- [For bars](https://oasysai.com.br/para-voce/bares)
- [For restaurants](https://oasysai.com.br/para-voce/restaurantes)
- [For botecos, the neighbourhood bar](https://oasysai.com.br/para-voce/botecos)

---

This is the English markdown representation of https://oasysai.com.br/insights/comissao-por-garcom. The rendered page
is in Brazilian Portuguese and is the authoritative version.
