Variable pay for ecommerce teams, in cash
Most ecommerce variable pay is tied to revenue or GMV, which the person usually does not control and which quietly pays them to discount. Here is what to tie it to instead, by role.
- A metric the person cannot influence is not an incentive, it is a lottery, and it drives out the people who understand the difference.
- Paying an operations person on GMV rewards discounting, which is the one lever they can pull that damages the business.
- A bonus is discretionary and backward looking while a commission is formulaic and contractual, and confusing the two creates a dispute you will lose.
- Any plan built on a marketplace metric needs a written clause for what happens when the platform changes the rules mid-period.
Almost every variable pay plan in a small ecommerce team is tied to revenue or GMV. It is the number the founder watches, so it becomes the number everyone is paid on. The problem is that most people on the pod cannot move it directly, and the one lever they can pull to move it fast is discounting. You have paid them to erode your own margin.
The equity side of pay is a separate conversation with its own mechanics. This is about cash, which is the part most people actually make decisions on.
The controllability test
Before you design anything, apply one test to every candidate metric: if this person did their job unusually well for three months, would this number move because of them?
If the answer is no, you are not designing an incentive. You are running a lottery with their household budget, and the people who notice first are the good ones. They understand exactly how much of the outcome they own, and being paid on noise reads as an employer who does not understand the job. The mediocre performer is happy to take a random payout. The strong one leaves.
Match the metric to what the role actually decides
The pod model this site works from puts three roles around a senior owner: operations, catalogue and advertising. Each of them controls something different, so each needs a different number.
- Operations. Fill rate and on-time dispatch. These are almost entirely within the role. If stock was in the building and the order went out late, that is theirs. If the order could not go out because purchasing never raised the order, that is not, which is why the plan needs an exclusion rather than a shrug.
- Catalogue. Content health and listing quality across the channels you actually sell on. Completeness, image compliance, attribute accuracy, how fast a new SKU goes fully live. Do not pay this role on category revenue, because the lag between a fixed listing and a revenue effect is long enough that they will never see the connection.
- Advertising. Contribution margin after ad spend, not ROAS on its own. ROAS can be hit by starving everything except branded search, which flatters the report and grows nothing. Making the ads person carry margin forces them to care what the order was worth after fees and returns.
The senior owner is the one role where a blended commercial number is fair, because they genuinely decide across all three.
A bonus and a commission are different instruments
People use the words interchangeably and then argue about it in March. A bonus is discretionary and backward looking. You decide after the period, using judgement, and you are not obliged to pay it. A commission is formulaic and contractual. If the formula produces a number, you owe it, including in a month where the outcome flattered them.
Both are legitimate. What is not legitimate is describing something as a commission while reserving the right to overrule the formula when the payout looks large. Decide which one you are offering, write it in the offer letter in those terms, and then behave accordingly. The first time you quietly reduce a formulaic payout because it felt too high, every plan you write afterwards is treated as fiction.
Caps, floors and the platform spike
An uncapped plan on a marketplace metric is a trap, because marketplace metrics have spikes nobody on your team caused. A platform runs an event you did not plan for, a competitor goes out of stock, your category gets a burst of app traffic. Under an uncapped commission you write a large cheque for weather.
Cap it. If an extraordinary period genuinely happens and the person genuinely drove it, pay a discretionary bonus on top because you chose to, which costs you the same money and buys considerably more goodwill.
Floors matter for the opposite reason. If a warehouse move, a platform suspension or a cash crunch makes the target unreachable through no fault of theirs, a plan with no floor pays nothing in the month they worked hardest. Set a floor for defined disruptions, and define them in advance rather than adjudicating each one while the person is upset.
Write the platform change clause before you need it
This is the clause almost nobody has and almost everybody needs. Marketplaces change commission structures, fee slabs, ranking behaviour and fulfilment rules mid-period. Any of those can make a target either unreachable or trivially easy through no action of the employee.
State in the plan what happens when it does: who decides that a material change has occurred, whether the target is restated or the period is treated as neutral, and by when that decision gets made. The point is not to predict the change. It is to agree the procedure while nobody has money riding on the answer.
Keep the plan small enough to hold in your head
One or two metrics per role. If a person cannot recite their plan from memory, it will not change a single decision they make on a Tuesday, which was the entire purpose. Complexity in a variable pay plan is usually a sign that the designer could not decide what mattered and hedged by including everything.
Review the plan once a year against what actually happened. If the same person hit maximum payout every period, the target was decoration. If nobody reached it twice, you did not build an incentive, you built a pay cut with extra steps.