Managing up is keeping the person who owns your scope able to decide. It is not flattery, and it is not a shadow org. Politics is when information is used as currency. The alternative is a cadence: what moved, what is blocked, what you need from them, and what will blow up if we wait. Grove's "task-relevant maturity" still applies: more detail when they are new to the domain, less when they trust the system.

The no-surprise rule

If a date will slip, they hear it from you with a new date and a cut list, not from a dashboard in a QBR. If you disagree with a direction, you say so with an option, not with a leak to their peers first. Escalation is a tool: use it when the decision is above your pay band, and tell your manager you are doing it.

text
This week: canary 20%, error rate 0.4% → 0.2%
Risk: vendor SLA miss on Friday; need approval to fail open
Ask: 30 min with security on the exception

Written beats hallway for anything that will be repeated. A short doc scales to skip-levels.

Options, not just problems

"We need more people" is not a decision. "Drop feature B, slip two weeks, or pull Jane for 50%" is a decision. Staff engineers who only bring problems train managers to treat them as weather.

Don't optimize for looking busy. Optimize for decision quality. If they choose against your recommendation, execute or leave — lingering sabotage is politics.

Boundaries

You are not their therapist or their spokesperson in every slack war. Protect maker time; batch updates. If they want daily stand-up theater, propose a written pulse that takes five minutes.

Read Grove on meetings and Fournier on managing up chapters. Then replace one status meeting with a doc for two weeks. If your manager is more informed and you wrote less theater, you managed up. If you spent the week crafting a narrative that hid the risk, you did the other thing. The difference is whether they can still steer.