The last mile of the whole track
Five sessions of design: what the system is for, the manager as the mechanism, ladders, promotion, and underperformance as an organisational output. This session is the feedback loop. Everything you built shows up here as three readable signals - who leaves that you wanted to keep, whether anyone could step into a critical role tomorrow, and what your exits taught the people still in the building. All three are measurable, all three are usually measured badly or not at all, and all three are leading indicators of whether your performance system is actually working. The track ends with a maturity path, because a plan that starts everywhere at once starts nowhere.
Regretted and unregretted are opposite events 7 min live
A single attrition percentage is one of the least useful numbers on an executive dashboard, because it adds together the thing you are trying to prevent and the thing you sometimes need. Split it and the number starts telling you what to do.
LiveThe stay interview beats the exit interview3 min▶
Exit interviews are collected diligently and act on almost nothing, for two structural reasons: the person has already gone, and they have every incentive to be diplomatic about the manager they may need as a reference. You get "for a new challenge" and file it.
- The stay interview asks the same questions while the answer still matters. What would make you consider leaving? What is the best part of your week, and could there be more of it? What would you change if you could? What do you want to be doing in two years, and is that visible here?
- Run it with the people you would fight to keep, roughly twice a year, and not by the person who decides their pay if you can avoid it - a skip-level or an HR partner gets more honest answers.
- The discipline is doing something visible with one answer. A stay interview that changes nothing is worse than none, because you have now asked and demonstrated that asking is theatre.
What Cartwheel's exit data said, and what was true. Cartwheel's exit interviews cited compensation most often. Its stay interviews, run six months later with people still employed, surfaced something quite different: no visible path to the next level and a manager who never had time for a 1:1. Pay was the socially safe answer on the way out - and it was also the most expensive possible thing to act on. Cartwheel had been considering a pay review; the actual fix was in a3 and a2.
Could anyone actually do the job? 6 min live
Most succession plans are a list of names next to roles, produced annually and never tested. The useful version asks a harder question about each critical role: if this person left tomorrow, who covers it, how long would they need, and what would we lose in the meantime?
LiveWhat the team learns from an exit3 min▶
An exit is the highest-attention internal communication your organisation produces. People will not be told the details - correctly, because it is confidential - so they will read the shape of it instead, and draw conclusions with total confidence.
- Handled well - no surprise to the person, dignity in the leaving, no gossip from management, work redistributed thoughtfully - the team concludes that the standard is real and that they would be treated fairly.
- Handled badly - abrupt, no visible prior process, or someone competent removed for reasons nobody can see - the team concludes that nobody is safe and that performance is not what determines outcomes. Your best people update their assumptions first, because they have options.
- The tell: whether colleagues are surprised. A team that is surprised was watching someone struggle without anything visibly being done, which tells them feedback here happens invisibly and suddenly.
You cannot explain an exit, but you can be consistent enough that the pattern speaks for itself. That is the practical value of everything in a5.
Self-studyPsychological safety is the precondition, not the perk2 min read▶
Amy Edmondson's work on psychological safety is usually filed under culture, which undersells it here. For a performance system it is infrastructure: safety is what determines whether you find out about problems while they are still cheap.
- Without it, ramp problems stay hidden until they are performance problems - the new joiner who does not understand the work will not say so.
- Without it, coaching cannot happen, because coaching requires someone to admit a gap to the person who writes their review.
- Without it, your stay interviews return the same diplomatic answers as your exit interviews.
- Where it exists, the leverage curve from b1 works - problems surface at week three instead of month nine, which is the whole economic argument of this course.
Safety does not mean no standards. The two go together: high standards plus safety is where people improve, whereas high standards plus fear produces concealment and then formal processes.
Split your attrition and test your bench ★ 12 min · think it through
Two exercises, both of which usually produce an uncomfortable surprise within ten minutes.
Take last year's leavers and sort them into the four quadrants. Be honest about the "would have fought to keep" line - it is the keeper test used constructively.
Look only at the regretted-voluntary box and ask what those people had in common. Manager, tenure band, team, level - the pattern is usually specific.
Now list your critical roles and mark bench depth: ready now, ready in a year, nobody.
For every "nobody", decide whether you are accepting the risk or funding a second person. Both are valid; not deciding is not.
| Finding | What it looked like | The action |
|---|---|---|
| Regretted leavers clustered | Six of eight came from two teams, both with managers carrying 10+ reports | Split both teams - the a2 span-of-control fix |
| Tenure band | Most left between 18 and 30 months, just after mastering the role | The growth conversation was missing at exactly the point it mattered |
| Bench depth | Three critical roles had nobody, all held by long-tenured people | Two funded a deputy; one risk formally accepted and recorded |
| Key-person trap | The one person who understood the pricing model had been passed over twice | She could not be promoted because nobody could replace her - so she left |
Pick where you start ★ 8 min · your org
The whole track in one path. Choose the row you are actually on, not the one you would like to be on, and commit to a single first move.
Crawl. A written bar per role, protected 1:1 time, and managers taught one coaching conversation. Nothing else. This is where most organisations should start.
Walk. Observable ladders, levelling calibration with the five rules, growth conversations on a cadence, and a plan policy that requires teaching.
Run. Split attrition metrics, stay interviews with your keepers, bench depth per critical role, advertised stretch assignments, and a consistency check before any exit.
Name the one thing you will do first, and what you will stop doing to pay for it. Then send your managers to the practitioner track, because none of this happens without them.
Why crawl is the right answer more often than it feels like. Cartwheel's exec team wanted to launch ladders, calibration, a new platform and a talent-review cycle in the same quarter. What their managers actually needed was time and one skill. They did the crawl row instead - a written bar per role, protected 1:1s, and one training session on the coaching conversation - and the following year they had fewer formal cases and better-evidenced promotions without having bought anything. The heavier machinery landed in year two, on managers who could now use it. Sequencing is not caution here; the machinery only works if the coaching layer exists underneath it.
Try it yourself - this week ◐ 30-45 min total
- Split last year's attrition into the four quadrants and look for the pattern in the regretted-voluntary box.
- Book stay interviews with the five people you would most hate to lose, run by someone other than their manager.
- Mark bench depth for every critical role. For each "nobody", either fund a second person or formally record that you are accepting the risk.
- Ask, after your next exit, whether the team was surprised. Surprise is a diagnosis of your feedback culture.
- Choose your row - crawl, walk or run - name the first move, and name what you will stop doing to pay for it.
- Send your managers to the practitioner track. The design does nothing on its own.
Sources covered (cited)
Taught from the retention and organisational-safety literature. This session covers:
Three questions before you go 🎯 ◐ 90 seconds
1 · Why is a single attrition percentage close to useless?
Split it by regretted or not, and by who decided. Only one of the four boxes is a genuine alarm. And zero voluntary attrition is not a goal - an organisation nobody ever leaves is usually one nobody can grow in.
2 · Why does the stay interview outperform the exit interview?
Exit interviews collect socially safe answers - often "compensation" - from people who have already gone. Stay interviews, ideally run by a skip-level or HR partner, surface the real causes: manager quality and an invisible path. The discipline is visibly acting on one answer.
3 · A critical role is understood by exactly one person. What is that?
Key-person risk is not a compliment. The role is fragile, and the person cannot be promoted or take real leave because nobody can cover them - so eventually they leave for somewhere they can grow. Either fund a second person or formally record that you accept the risk.