Escalation That Works
Escalation has a reputation as a political act because most of it is done badly: a problem handed upward with no options and an implicit request that someone else choose a side. Done well it is a one-page artifact with two or three costed options, a recommendation, the decision needed, a date, and what you will do by default if no answer arrives. That last line is what converts a message into a decision, and it is the part almost everyone omits.
TL;DR: Escalate with an artifact, not a complaint. One page, in a fixed order: the decision needed in one sentence and the date it is needed by, two or three options each with what it delivers and what it costs, your recommendation in three sentences, and what you will do by default if no decision arrives. Show it to the other party before it goes up, because an escalation that surprises them is remembered as an attack. Escalate when a risk is owned by someone above you or when new evidence has appeared, and not because you lost an argument, since that is the version that ends your influence on the next decision.
The artifact, in order
The order is not cosmetic. A reader who stops after two paragraphs should already know what is being asked and by when.
| Section | Contents | Why it is there |
|---|---|---|
| The decision | One sentence, and the date it is needed by | A reader who reads nothing else knows what is wanted |
| Options | Two or three, each with what it delivers, its cost, its lead time and its risk | Choosing between costed options is a decision; choosing between a problem and nothing is not |
| Recommendation | Three sentences, including what would change it | You have the most context, so withholding a view wastes it |
| The default | What you will do if no decision arrives by the date | Converts the message into a plan and makes silence safe |
| Appendix | The data, the analysis, the details | Available without being in the way |
The default line is the one that changes outcomes. Most escalations receive no reply, because the recipient has many of them and no forcing function. A message ending "if we have not heard by the 15th I will proceed with reclaim, which is useful under either path" removes the need for a reply while still offering one, and it means the deadline passing is not a decision made by accident.
When escalation is correct, and when it is not
Two conditions justify it. The decision accepts a risk the person deciding does not own, such as potential data loss, a safety control, or a commitment to a customer, in which case the person who owns that risk has to be the one accepting it. Or evidence has appeared that did not exist during the original discussion, which is not really escalation at all but new information presented as such.
Outside those two, a lost argument is a judgment call someone was authorized to make. Escalating it teaches the organization that persistence beats process, and it marks the escalator as difficult in a way that is hard to undo. The behaviour that reads as senior is disagreeing clearly once, committing fully, keeping a written note of what you expected to break, and bringing that note later as data rather than as vindication.
Showing it first
An escalation the other party learns about from their manager is an attack, whatever its contents. Showing them the page before it goes up costs nothing, frequently improves the options because they know things you do not, and occasionally resolves the issue without the escalation happening at all.
It also changes what the reader above sees. Two engineers who disagree and have jointly written down the options present a solvable decision; two engineers escalating separately present a conflict, and conflicts are resolved by seniority rather than by analysis.
What interviewers are listening for
Whether the escalation was an artifact or a message. Candidates who describe going to a manager with a problem are describing the version that fails; candidates who describe a page with costed options, a recommendation and a default are describing the version that works. The second signal is the narrowness of the trigger, because someone who escalates only when a risk is misowned or when evidence changed has a rule rather than a reflex. The third is whether the other party saw it first.
Key takeaways
- One page in a fixed order: the decision and its date, two or three costed options, a recommendation, and a stated default.
- The default line converts a message into a plan and stops a passing deadline from becoming a decision by accident.
- Escalate when the risk is owned above you or when new evidence appears, and not because you lost.
- Show the page to the other party before it goes up, because a surprise escalation is remembered as an attack.
- After a decision, commit fully and keep a written record; bring it back later as information rather than as vindication.