What this track is really about
Six sessions on designing the system, not running it. You will decide what your ladders describe, how levelling gets calibrated, how promotion decisions get made and evidenced, what your policy on underperformance actually is, and what your exits communicate to everyone who stays. The running example is Cartwheel, a mid-size online retailer - the same fictional company as the sibling course on OKRs, so you are looking at one organisation rather than a fresh set of scenarios each time. One boundary up front: this is not the goal-setting course. Objectives, key results, metric design, rater bias and calibration mechanics belong to learn OKRs with Phoebe. This course is about what you do with a person over the arc of their time with you - and what happens when that arc has to end.
Grow capability, decide fairly, keep people 7 min live
A performance system exists to do three things. They pull in different directions, they need different cadences, and the classic mistake is to run all three through one annual event and one number.
LiveWhy the annual cycle collapsed3 min▶
When large organisations examined their own review cycles, the finding was consistent and uncomfortable: enormous cost, little capability change. Buckingham and Goodall's account of Deloitte's rebuild put numbers on the effort going into the annual machinery and asked what it bought. Adobe replaced its annual review with lighter, more frequent check-ins and reported less admin and lower voluntary attrition.
- What failed was not the rating scale - it was the frequency. A once-a-year conversation is a memory test with pay attached.
- What worked was more conversations, not better forms. None of the reformers succeeded by inventing a cleverer scale.
- The catch: "we abolished ratings" is not a strategy on its own. If pay and promotion decisions still happen, the judgment moves somewhere less visible and less auditable. Job 2 still needs evidence.
What Cartwheel's board actually asked. After a costly review cycle, Cartwheel's CEO asked the obvious question: "what did we get for it?" The honest answer was a distribution of ratings and no evidence that anyone was better at their job than the year before. That is the test to hold your own system to - not whether the cycle completed, but whether capability moved.
Four stages, and your lever at each 6 min live
Managers run the lifecycle. You fund it, staff it, and set its rules. Here is the same four-stage arc your managers see in the practitioner track, with the design decision that sits under each stage - the part that is yours.
LiveSpan of control is a performance-management decision3 min▶
The most common root cause of a broken performance system is not the framework. It is arithmetic. Weekly coaching costs a manager real hours per person, and those hours are the mechanism by which capability changes. Stack fifteen reports on someone who also has delivery targets and you have decided, structurally, that coaching will not happen - no matter what the policy document says.
- Count the hours before you buy the framework. If the maths does not allow a weekly conversation per person, the framework will be theatre.
- Watch for the player-coach trap. A team lead with a full individual workload plus eight reports will always sacrifice the coaching, because only one of those has a visible deadline.
- Promoting your best individual contributor into management is a decision to lose their output and to gain a novice manager. Fund the transition or expect both to go badly.
Self-studyThe three jobs and the confidentiality problem2 min read▶
Keeping the three jobs apart runs into a practical objection: the same manager holds all three, and the person knows it. You cannot fully firewall a coaching relationship from the person who will later write a pay recommendation. What you can do is design so that the leak is small and understood:
- Separate the meetings and say so. Steering and development weekly, decisions in rare, named, well-prepared conversations - and tell people which conversation they are in.
- Make the decision criteria public in advance. If the ladder and the promotion bar are written down and known, coaching conversations become preparation rather than assessment.
- Give people a second channel. A skip-level, a mentor outside the reporting line, or an HR partner means the coaching relationship is not the only route for an honest problem.
The sibling OKR course makes the same argument from the measurement side, and its rule is worth importing wholesale: an OKR score steers and teaches, it never sets pay.
Diagnose which stage you are starving ★ 12 min · think it through
Every organisation underfunds one stage, and the symptoms are specific enough to diagnose from the outside. Find yours.
Read the symptom column below and mark which line sounds most like your organisation in the last twelve months.
Check the arithmetic behind it: how many reports per manager, how much manager time is protected, whether a written bar exists per role.
Name the one stage you would fund first, and what you would stop doing to pay for it.
Resist the urge to fix stage 4 first. It is the loudest and almost never the cause.
| What you observe | Starved stage | What to fund |
|---|---|---|
| New joiners take 9 months to be useful, and some never get there | Ramp | A written bar per role and a real 90-day plan |
| Managers only give feedback in review season; every concern arrives late | Coach | Manager time and manager training, plus a lower span of control |
| Strong people leave saying "there was no path here" | Grow | Ladders, visible levelling, and growth conversations on a cadence |
| Exits are messy, inconsistent, or land in disputes | Decide | Policy, fair process, and consistency checks - after fixing the above |
| Ratings are calibrated beautifully and nothing improves | All of 1-3 | Stop investing in the measurement layer and fund the coaching layer |
Separate the three jobs on paper ★ 8 min · your org
Take your current performance calendar and sort every element by which of the three jobs it serves. Anything serving two is where trust leaks out.
List every scheduled element: 1:1s, check-ins, mid-year, year-end, calibration, promotion rounds, pay review, talent review.
Tag each with one job: grow capability, decide fairly, or keep people. Force a single tag.
Flag anything you could not tag with one. Those are the merged elements, and they are where honesty degrades.
For the worst merge, sketch the split: what becomes two meetings, and what you tell people about which is which.
The mid-year that was secretly a pay meeting. Cartwheel's mid-year check-in was described to staff as developmental. It also fed a "readiness" flag used in the pay round. Within two cycles people had worked that out, and mid-year conversations became careful performances in which nobody admitted a weakness - which destroyed the only developmental conversation in the calendar. Nothing was hidden maliciously; the flag was added for good administrative reasons. But a developmental meeting with a decision quietly attached is a decision meeting, and people will treat it as one.
Try it yourself - this week ◐ 30-45 min total
- Do the arithmetic: average reports per manager, and how many protected hours per week a manager actually has for coaching. Compare the two honestly.
- Sort your performance calendar by the three jobs and find every merged element.
- Ask three managers what they think the performance system is for. The variance in their answers is your real starting point.
- Ask what evidence exists that anyone is better at their job than a year ago. If none exists, that is the finding.
- Check whether a written bar exists for each role at each level. If not, note that a3 is the session that fixes it.
Sources covered (cited)
Taught from the management-craft canon. This session covers:
Three questions before you go 🎯 ◐ 90 seconds
1 · What actually failed in the traditional annual review cycle?
The organisations that improved things added conversations, not better forms. Note the catch too: abolishing ratings is not a strategy by itself, because pay and promotion decisions still happen and still need evidence.
2 · A manager has fifteen reports and a full delivery workload. What has the organisation decided?
Weekly coaching costs real hours per person, and those hours are the mechanism by which capability changes. If the arithmetic does not allow the conversation, the framework is theatre. Span of control is a performance-management decision.
3 · Exits at your company are messy and inconsistent. Where should you look first?
The decide stage is a symptom far more often than a cause. Fix policy and fair process too, but if stage 4 keeps arriving, the honest question is what stages 1 to 3 were not doing.