Where we are
Last session's leverage curve made an economic argument: the same issue costs a short conversation in week three and an improvement plan in month nine. This session is about actually being in week three with something useful to say. Two ideas do the work. First, a new person cannot meet a standard nobody has written down - and most ramp failures are not capability failures, they are undocumented-expectation failures. Second, the first ninety days is the only period when feedback is socially free. Nobody is offended by being told how things work here in week two. Say the same thing in month eight and it lands as criticism, because by then they reasonably assume they are doing it right.
The 30, 60 and 90 day milestones 7 min live
A ramp plan made of activities - "meet the team", "read the docs", "complete training" - measures attendance. A ramp plan made of observable milestones measures whether the person is becoming useful, which is the only thing either of you actually cares about.
LiveGive them the document, and the map of who is who3 min▶
Two things to hand over in week one that most managers keep in their heads:
- The ramp document itself. A new person who can see the 90-day milestones will self-correct against them, which converts your standard into their checklist. Withholding it means they have to guess what you want and then be judged on the guess.
- The informal map. Not the org chart - who actually knows how things work, who to ask when something breaks, whose approval matters in practice, which meetings are real. New people burn months discovering this, and the ones who never get it look slow when they are only uninformed.
Then set a standing weekly slot for the whole first quarter. High task-relevant maturity earns lighter oversight later, as b3 covers; a new joiner has none, by definition, however senior they are.
The senior hire who was left alone. Cartwheel hired an experienced buyer from a larger retailer and gave her a light ramp, on the reasonable grounds that she had done the job for a decade. She had - somewhere with different systems, a different margin structure and different supplier relationships. Four months in she was visibly struggling and had been too senior to admit she did not know how anything worked. Seniority is maturity at a task. Her task was new, so her maturity was low, and she needed more ramp than the graduate, not less.
Feedback is socially free for about twelve weeks 5 min live
There is a window at the start of a job when almost any correction is welcome, because the person expects to be learning. It closes, and once it does the same sentence costs far more to say and lands far worse.
LiveHow to use the window without being oppressive3 min▶
The instinct that stops managers using the free window is a good one - nobody wants to nitpick a nervous new joiner. Two things resolve it:
- Name the frame out loud in week one. "For the first three months I am going to tell you small things as they come up, because that is much easier than saving them. Please do the same to me." Now every correction is a fulfilled promise rather than a surprise.
- Keep it to how-things-work, not judgment. "We flag delays early here" is a norm. "You are not proactive enough" is a verdict, and it is also the vague personality language the sibling OKR course shows lands unevenly across different people.
Then, at day 90, do a real checkpoint rather than letting it drift: what is working, what is not yet, and what changes in the next quarter. If a joiner is genuinely off track at day 90, that is the moment to say so plainly - it is still cheap, and it is still recoverable.
Self-studyWhen the ramp is failing and it is not the person2 min read▶
Before concluding a slow ramp is a hiring mistake, rule out the four things that are yours:
- The role changed between the job description and their arrival. Extremely common after a reorg, and it means you hired for one job and handed over another.
- Nobody has time to teach them. A ramp assumes an available teacher. In a peak period, that teacher does not exist and the joiner is left to infer everything.
- The work is not available yet. Access, permissions, tools, or a project that has not started. People cannot ramp on work they cannot reach, and they will absorb the delay as their own failure.
- There is no written standard, so they are optimising for whatever you last mentioned.
All four are context causes in b6's taxonomy, and none of them are fixed by expecting more from the person.
Build a Cartwheel ramp for one real role ★ 12 min · everyone tries
Take the merchandising-analyst role and write the 30/60/90 as observable milestones. Then check every line against one question: could someone who cannot do this job still tick it?
Write day 90 first. Independence is the destination, so define it before the route.
Work backwards to day 60 - contributing but reviewed - and day 30 - oriented, with one small real thing delivered.
Strike out any milestone that is an activity. "Attended", "read", "completed training" all go.
Name the teacher for each phase, by person. A ramp with no named teacher is a wish.
Add the week-one framing sentence about small feedback, and diarise the day-90 checkpoint now.
| Milestone | Observable evidence | Teacher |
|---|---|---|
| Day 30 · oriented | Has produced one weekly category report that went out unchanged after review | Senior analyst |
| Day 30 · oriented | Can explain how margin, sessions and conversion connect for our business | Manager, in the 1:1 |
| Day 60 · contributing | Owns the weekly report end to end; review finds wording, not errors | Senior analyst |
| Day 60 · contributing | Has raised one problem nobody asked about | Manager |
| Day 90 · independent | Answers a buyer's ad-hoc question without escalating, and the answer holds up | None needed - that is the point |
| Day 90 · independent | Another analyst has asked them for help with something | - |
Write the week-three sentence ★ 8 min · do your own
Pick something a current or recent joiner is doing that you have not mentioned. Write the version you would say in week three, and notice how much smaller it is than the version you would have to say in month nine.
Name the behaviour and the norm it misses. Norm, not judgment.
Say why it matters here, in one clause. Context makes a norm learnable rather than arbitrary.
Offer the mechanism, not just the instruction - the template, the channel, the person to ask.
Then compare it to the month-nine version of the same problem, and notice that the later one needs evidence, examples and probably a plan.
Same issue, two prices. Week three: "One thing about how we work - if a report is going to slip, flag it the day before rather than the morning it is due, because the buyers build their orders around it. Easiest way is to drop a note in the team channel." Total cost: forty seconds, and it is heard as helpful. Month nine, the same issue, now habitual and having caused two missed buys: a prepared conversation, three dated examples, an agreed change, a follow-up, and a note on file in case it does not improve. The behaviour is identical. The eight months are the entire difference, and they were free to spend and expensive to have spent.
Try it yourself - this week ◐ 30-45 min total
- Write the 30/60/90 for one role on your team, as observable milestones, and name a teacher for each phase.
- Test every milestone against "could someone who cannot do this job still tick it?" Strike the ones that fail.
- Send the ramp document to your most recent joiner, even if they are past 90 days. It still helps them self-correct.
- Say one week-three-sized thing to someone this week, whether or not they are new.
- For anyone ramping slowly, check the four context causes before concluding anything about the person.
- Diarise a real day-90 checkpoint for every current joiner.
Sources covered (cited)
Taught from the management-craft canon and ramp practice. This session covers:
Three questions before you go 🎯 ◐ 90 seconds
1 · What is wrong with a ramp plan built from "meet the team, complete training, read the handbook"?
Test every milestone with "could someone who cannot do this job still tick it?" Observable milestones - shipped one real thing, owns a recurring piece of work, answers a question without escalating - measure whether the person is becoming useful.
2 · A senior hire with ten years in the same role joins from a larger competitor. What ramp do they need?
Seniority is maturity at a task, not a property of the person. A senior joiner in a new context has low maturity there, and is often too senior to admit they do not know how anything works - which is exactly how a four-month problem forms quietly.
3 · Why is the same corrective sentence cheap in week three and expensive in month nine?
Nothing about the sentence changes - only how long you waited. Name the frame in week one ("I will tell you small things as they come up") so every correction is a fulfilled promise, and keep it to how-things-work norms rather than verdicts.