Fix the two lowest, ignore the rest

When you prepare for an interview loop, the instinct is to review everything. That is why most preparation is inefficient, and it is why six weeks of it often produces the same result as two.

Interviewers are scoring across roughly six dimensions: correctness, production judgment, system reasoning, execution under a clock, scope, and communication.

Almost everyone is strong on two of them and weak on two. The weak two decide the outcome. And almost everyone prepares by revisiting the strong ones, because reviewing what you already know feels productive and looking at what you cannot do feels bad.

Raising a four out of five to a five changes nothing about the verdict. Nobody in a debrief has ever said the correctness was a four rather than a five, so it is a no. They say the reasoning was not visible, or the scope was mid-level, or they could not follow the explanation. Those are the twos.

The diagnostic takes twenty minutes and you can do it alone.

Score yourself accurately on each of the six. Not aspirationally. Think about your last loop, or your last design review, and score what actually happened.

Correctness: do you get the technical facts right under pressure.

Production judgment: can you talk about failure, testing, rollout and maintenance rather than only the happy path.

System reasoning: can you turn an ambiguous prompt into constraints and boundaries.

Execution: can you clarify, plan, implement and recover while a clock runs.

Scope: does the size of what you describe owning match the level you are targeting.

Communication: can a listener follow your assumptions and trade-offs without asking.

Write the two lowest down. Those are your preparation for the next month. Everything else is optional.

The uncomfortable part is that the weak two are usually not technical. Scope and communication are the most common pair, and they are the two most people refuse to work on because they do not feel like engineering. They are what the verdict turns on.