Where we are
Everything from sessions 1 to 5 still applies. What changes tonight is that the reader has an emotional stake, which means one new thing goes at the top and one habit has to be actively resisted.
★ What you walk out with tonight. The bad-news message you have been avoiding, written; a disagreement framed so it can be settled rather than won; two questions for taking pushback; and a repair script for the conversation that already went wrong.
LiveWhy late is worse than bad2 min▶
Daybreak's founder said it in session 2: "I do not really mind that it slipped, things slip. What I cannot have is finding out in the room."
That is not one founder's preference, it is the general shape. The slip costs a fortnight. The surprise costs your credibility, because the person now has to wonder what else they do not know - and that wondering does not switch off after you fix the timeline.
Which produces the rule that governs this whole session: tell them while it is still information. A delay reported at week one is a project update. The same delay discovered at week three is a character question, and it is a much more expensive conversation.
Three things a bad-news message must contain
7 min liveThe instinct is to pad - to soften, to explain first, to bury the sentence in reasonable-sounding context. Every one of those makes it worse, and worse in a specific way: padding reads as evasion, which converts an accident into something that looks like a cover-up.
LiveFour things people add that make it worse4 min▶
- The compliment sandwich. Good news, bad news, good news. Everybody has seen it, which means the opening praise now functions as an alarm rather than a cushion.
- The passive voice. Session 4 counted eight of them in Daybreak's original, all on the failure. "We underestimated" costs you nothing you were not already going to pay and buys you the reader's belief that you know what happened.
- The long explanation before the fact. Every sentence before the news reads, in hindsight, as preparation of the ground. Say it, then explain if they want it - and often they will not, which tells you how much of your draft was for you.
- The unforced apology. One "sorry" is fine where it is warranted. Repeated apology makes the reader manage your feelings, on top of the actual problem, at the moment they have least appetite for it.
The test: read it back and ask "does this sound like somebody who knows what happened, or like somebody hoping this lands softly?" Readers detect the second one reliably, even when they could not say what tipped them off.
"I wanted to flag that there have been some challenges on the subscription work. As you know the team has been working extremely hard, and unfortunately a number of factors outside our control have meant that it was determined the original timeline may not be achievable in its entirety..."
Nothing here is a lie. Every clause is defensible. And the founder finishes it knowing something has gone wrong, not knowing what, not knowing what it costs her, and now wondering how long you have known.
"The redesign will miss 18 September by three weeks - we underestimated the payments work. For the board on the 30th you can say it ships 9 October. I need you to pick the scope by Thursday."
41 words. What happened, what it means for her, what you want. She now has a line for the board, a decision to make, and the impression that the person running this knows exactly where it stands.
Disagreeing so it can be settled rather than won
6 min liveMost workplace disagreement fails not because somebody was wrong but because it got framed as a contest. A contest has a loser, and the loser has a long memory. Three moves stop that happening.
LiveTaking the pushback without defending4 min▶
The same two questions from the design critique in learn-product-design-with-phoebe, and they work identically in a conversation:
- "What made you land there?" Turns a verdict into a reason you can engage with. "This is too risky" becomes "we did this in March and it took the whole team out for a week", which is a fact, and facts can be addressed.
- "What would need to be true for you to be comfortable?" Turns opposition into a specification. Often the answer is small and you can just do it.
And the sentence that buys you thinking time without conceding anything: "That is fair, let me think about it and come back to you tomorrow." You have acknowledged, not agreed, and you have removed the requirement to be clever under pressure - which is where most people concede things they did not mean to.
"Well, the research showed..." is correct and it teaches the room that objecting to you costs an argument. Next time they will say "sounds good" and raise it with somebody else afterwards, which is the outcome you were trying to avoid.
Self-studyThe repair, after it went badly4 min▶
Sometimes you get it wrong. You were sharp in a meeting, or you let something slip that should have been said in week one, or you defended when you should have listened. The repair conversation is short and almost nobody has ever been taught it.
- Name it specifically. "I was short with you in the review on Tuesday." Not "I feel like things have been tense" - vagueness makes the other person do the work of identifying what you mean.
- No explanation attached. The reason you were short is true and irrelevant, and attaching it converts an apology into a defence, which the other person hears immediately.
- Say what you will do instead. "I should have asked what you were seeing. I will do that next time." Specific and future-facing beats general regret.
- Then stop. Do not ask them to reassure you. That is the version where they end up managing your feelings about your own behaviour.
Four sentences, done within about two days. Left longer it calcifies into "how things are between us", and that is much harder to unpick than the original moment ever was.
Write the one you have been avoiding
★ 22 min · everyone buildsEverybody has one: a piece of bad news not yet delivered, a disagreement not yet raised, or a repair not yet made. Tonight you write it. You do not have to send it tonight, but you will have it.
Name it. Which is it - bad news, a disagreement, or a repair? They have different shapes and people routinely reach for the wrong one, usually delivering a disagreement dressed up as an update.
Three parts. What happened in one line, what it means for them, what you want from them. Write all three before improving any of them.
Cut the padding. Compliment sandwich, passives on the failure, explanation before the fact, repeated apology. Delete all four. Your draft will get roughly a third shorter and considerably more honest.
The ask. Even here. Especially here - a piece of bad news with nothing to do with it leaves the reader holding an anxiety and no action, which is the worst thing you can hand somebody.
Say it aloud. Once. If a sentence is hard to say, it is usually because it is not true, or because it is padded. Both are worth finding before you send.
Write the one-line version of what happened. Active voice, with a person in it. "We underestimated the payments work", not "the timeline was impacted". If that sentence is hard to type, notice - that difficulty is the exact reason the padded version exists.
Write what it means for them, specifically. Not for the project. For them: their board meeting, their commitment, their week. Session 2's listening pays here, because you know what they are afraid of.
Write the ask. A decision, an escalation, a date. If you genuinely want nothing, say "no action needed, telling you now so it is not a surprise later" - which is itself a good message, and it directly answers the fear.
Now delete your four padding types. Go through and physically remove them. Count the words before and after. Most people cut 30 to 40 percent and every one of those words was doing emotional work for the writer, not informational work for the reader.
If it is a disagreement, add the test. "If X is not true by Friday, we do Y." This is the move that ends disagreements, and it is the one people skip because it means giving up being proved right in the meeting.
Rehearse the pushback. Write down what they will say, and your two questions: "What made you land there?" and "What would need to be true for you to be comfortable?" Plus the thinking-time sentence.
Say the whole thing out loud once. Then grade it below - this is a ship-or-not gate on a message that actually matters.
If you take one thing from the whole course. Tell them while it is still information. Every technique in these six sessions is worth less than being the person whose bad news arrives early, and that is a decision rather than a skill.
Where to go from here
ongoing- Send the hard one. This week. It is written and it will not improve by waiting - it will only get later, and late is the expensive part.
- Put the ask first in every message for two weeks. It stops being a technique and becomes how you write, and it is the highest-return habit in the course.
- Ask the two questions before drafting anything significant. "What would you do with that?" and "Anything you are worried about I should get ahead of?"
- Run one real message through the scorer each week. Watch which row is consistently your weak one. Everyone has one, and it is rarely the one they expected.
- Then go further. learn-negotiation-with-phoebe and learn-public-speaking-with-phoebe are planned in this bucket; learn-course-design-with-phoebe is the craft of designing learning; and learn-ai-writing-with-phoebe is what happens when you point a drafting tool at all of this.
Three questions
Clear all three and session 6 gets stamped in your passport - six of six.
1 · Why is late bad news more damaging than bad news itself?
Less time is a real cost and a bounded one. The unbounded cost is that a person who was surprised once now has to price in the possibility of being surprised again, on everything you tell them. That is why the rule is to tell them while it is still information.
2 · Which move actually ends a disagreement?
Evidence turns it into a contest with a loser who remembers. Escalation does the same with an audience. A test converts the disagreement into a date, which means it gets settled by what happens rather than by who is more senior - and it is skipped precisely because it gives up being proved right in the room.
3 · You were short with a colleague in a meeting. What does the repair conversation contain?
Attaching the explanation converts an apology into a defence and the other person hears it instantly. Vagueness makes them do the work of identifying what you mean. And asking for reassurance leaves them managing your feelings about your own behaviour, which is the opposite of a repair.
What this session teaches, and where it came from
Every fact on this page, and its verification tier, is recorded in the course's source map. Daybreak is a teaching scenario throughout; the measurement code in session 4 is real.
Session 6 cheat sheet · pin this
Six sessions and one message, from ignored to acted on. The thing you can do now that you could not do in session 1 is not write more elegantly - it is to know, before you send anything, what you want the person to do, whether they can find it, and what it will cost them to read. That is a mechanical skill with measurable parts, which is the good news, because mechanical skills can be practised by anybody. Next: learn-course-design-with-phoebe for designing learning that changes behaviour.