Performance reviews when the platform moves

An operations person can run a clean quarter and still miss every number. The platform changed a commission slab, or the ranking behaviour shifted, or a dark store got deprioritised, or the category filled up with sellers willing to lose money. None of those were decisions your employee made. All of them land in the report you are about to review them on.

This is the structural problem with performance management in marketplace and quick commerce roles. The outcome is a joint product of two authors, and only one of them works for you. A manager who reviews on outcomes alone will let go of a competent person after a bad quarter and promote a lucky one after a good one, and will do both while believing they are being rigorous and data driven.

Write down what you expect the platform to do

The single most useful habit here costs about twenty minutes at the start of a quarter. Write down, in one place, what you expect the platforms to do over the next three months. Fee revisions you have been warned about, events on the calendar, a category you know is getting crowded, a fulfilment change already announced, a warehouse you suspect is slipping.

You will get plenty of it wrong. That is not the point. The point is that at review time you have a document that predates the result. Anything that shows up in the outcome and was on the list is a known condition the person should have planned around. Anything genuinely absent from the list is a surprise. Without that document, every explanation offered at review time sounds equally plausible, and you end up deciding based on how convincing the person is in a meeting, which selects for articulacy rather than competence.

Review the decisions, not only the result

Pick three or four decisions the person actually made during the period and examine those directly. Which SKUs they pushed and which they let go quiet. How they moved budget when performance shifted. What they escalated to the platform and how early. What they chose not to do.

For each one, ask what they knew at the time, not what turned out to be true. A defensible decision that produced a poor outcome is a good decision with bad luck attached, and a reckless decision that produced a great outcome is still reckless. The second is the more dangerous case, because a good result buries the reasoning and the person learns the wrong lesson.

Look also at speed of detection. Someone who spotted a payout anomaly in week two and raised it did their job even if the money was lost anyway. Someone who found out when you did, from the settlement, did not.

Compare across your own channels

Absolute performance tells you very little in a period where something moved. Relative performance across your own channels in the same window tells you a great deal, and it is free.

If every channel softened together, you are looking at demand or at something in your own supply, not at the person running one of them. If one channel fell while the others held, the question narrows usefully to that channel, and it becomes a real conversation about what happened there rather than a general accusation. Where you have a comparable brand or a sister category on the same platform, that is a cleaner control than anything else you have access to.

Be honest about the limits. This works for market-wide movements and it does not work when a change hits only one channel for reasons outside the person, which is exactly why the written expectations list matters alongside it.

The conversation when someone genuinely is not performing

All of the above is about not punishing people for noise. It is not a reason to avoid the other conversation, and managers who internalise the controllability argument sometimes swing too far and stop holding anyone to anything.

The tell for genuine underperformance is a pattern that survives every adjustment above. The decisions are poor on their own terms. Problems are found late or by someone else. The same issue recurs after being discussed. Performance is weak in periods where nothing external moved.

When you have that, say it plainly and early. Name the specific behaviour and the specific gap, not a vague sense of disappointment. Say what has to be different and by when, in terms the person could measure themselves. Ask what they need, and mean it, because sometimes the answer is access or tooling or a decision you have been sitting on. Then write down what you both said, the same day, and share it with them so there is no version gap later.

Documentation, before it becomes a termination

If the pattern continues, the file you need is the one you should have been building all along: dated records of what was discussed, what was agreed, what support was offered and what changed or did not. Contemporaneous notes carry weight that a reconstruction written after the decision does not.

Keep the record factual and specific to the work. Avoid characterisations of the person. Make sure the standard you are applying was communicated in advance and applied consistently to others in comparable roles, because inconsistency is where these situations come apart.

Employment law in India is state-specific and this is not legal advice. Notice, final settlement, applicable shops and establishments provisions and the process required before a termination vary by state and by the terms of the individual contract. Before a performance conversation becomes an exit, take advice from an employment lawyer qualified in the state where the person is employed.