Where we are
You have a ladder written in observable scope, and a levelling process that produces a shared standard. Promotion is the moment that machinery gets used for real, with money and status attached - and it is where two specific leaks show up. The first is structural and well measured: the step from individual contributor to manager loses people who should have made it, and loses them unevenly. The second is linguistic. Most promotion processes run on two axes, performance and potential, and only one of those is ever defined. An undefined axis does not measure nothing - it measures whoever is judging. That is the same finding the sibling OKR course establishes about ratings generally, arriving here in its most consequential form.
The first rung is the one that breaks 7 min live
Attention usually goes to the top of the pyramid - the proportion of women and under-represented people in senior leadership. The measurement work says the decisive loss happens much earlier, at the first promotion into management, and that everything above it inherits that shortfall.
LiveMentorship is not sponsorship, and only one of them promotes people3 min▶
The distinction sounds like semantics until you look at what happens in the promotion room:
- A mentor talks to you. They give advice, share experience, help you think. Valuable, safe to offer, and easy for an organisation to scale into a programme.
- A sponsor talks about you. They spend their own credibility arguing your case when decisions are made, and they put your name forward for the assignment that builds the evidence. This is the one that moves promotions.
- Organisations run mentoring programmes and call the problem solved, because mentoring is cheap and comfortable. Sponsorship cannot be programmed as easily - it requires senior people to spend capital - which is exactly why it distributes unevenly and tends to follow affinity.
The same report notes four in ten entry-level women had received no promotion, stretch assignment, or leadership training in the previous two years. Stretch assignments are how promotion evidence gets built, so unequal access to them produces unequal evidence, which then looks like a difference in potential in the room.
The Cartwheel assignment nobody competed for. Cartwheel's cross-team returns project was the assignment that made two careers, and both times it was handed to whoever the director happened to think of. Nobody applied, because it was never announced. That is not malice, it is the default: high-visibility work distributed through the network of whoever is deciding. Announcing it changed the applicant pool immediately.
What "potential" has to mean to be usable 6 min live
The 9-box grid is a useful conversation tool with one severe flaw: one of its two axes is usually never defined. Performance can be evidenced. Potential, left undefined, absorbs everything the room cannot say out loud - confidence, similarity, availability for evening calls, how someone sounds in a meeting.
LiveFour fixes that change who gets promoted3 min▶
- Define potential, or drop the axis. Three evidenced behaviours, each requiring a named example. If your talent review cannot produce examples, you are ranking impressions.
- Advertise stretch assignments. Every high-visibility project posted, with a named selection basis. This is the cheapest intervention on this list and it directly attacks the evidence gap.
- Make sponsorship an expectation with a name attached. Ask senior leaders who they sponsored this year and what they did about it. Unmeasured sponsorship follows affinity by default.
- Audit the first rung specifically. Promotion rate into first-level management, by group, tracked over cycles - not the seniority pyramid, which is a lagging indicator of decisions made years ago.
And one thing not to do: set a promotion quota by group. That recreates the quota problem from a3 in a more politically dangerous form. Fix the evidence and the access; do not allocate the outcome.
Self-studyPromotion timing, and the trap of "acting up"2 min read▶
Two structural issues that quietly decide promotion outcomes:
- Evidence-first versus title-first. Most organisations require someone to demonstrate the next level before being promoted to it. That is defensible, but it means promotion depends entirely on access to work at that level - which loops straight back to who gets the stretch assignment. If you promote on demonstrated scope, you must actively distribute the scope.
- The indefinite acting-up. Someone covers the bigger role "temporarily" for a year without the title or the pay. It happens most to the people most willing to help, and it converts goodwill into a grievance. Set a hard limit - three to six months - after which you either promote them or end the arrangement. An acting arrangement with no end date is a decision you have made by not making it.
Rewrite a Cartwheel talent-review entry ★ 12 min · think it through
Here are two entries from a Cartwheel talent review, both marked "high potential". Strip out everything that is not evidence and see what remains.
Read each entry and underline every claim with no example attached.
Test each remaining claim against the three-behaviour definition: learns fast in new situations, gets results through others, seeks harder ambiguity.
Ask the access question: did this person have the chance to demonstrate the behaviour, or did they have the chance to be visible?
Rewrite both entries so a person outside the room could check them.
| Entry as written | What survives an evidence test | What it actually recorded |
|---|---|---|
| "Real star. Great presence with the exec team. Definitely ready for more." | Nothing. No behaviour, no example | Visibility to senior people, and the director's comfort |
| "Quiet but very solid. Not sure she wants it." | Nothing - and "not sure she wants it" was never asked | An assumption about ambition, made in her absence |
| Rewritten: "Took the returns analysis with no defined method and delivered it in three weeks (learns fast in new situations). Rebuilt the forecasting template that two other analysts now use (results through others). Asked for the cross-team project (seeks ambiguity)." Same person, second row. | ||
Audit your own first rung ★ 8 min · your org
Four numbers. Most organisations track the last one and none of the first three, which is why they see the problem years too late.
Promotion rate into first-level management, split by group, for the last two cycles.
Stretch-assignment allocation: who got the visible projects, and how were they chosen? Named basis or network?
Sponsorship: ask each senior leader who they actively sponsored, and what they did. Count the names and look at the pattern.
Then look at your seniority pyramid, and treat it as history rather than as a target.
Why the pyramid is the wrong dashboard. Cartwheel's exec team tracked the proportion of women in senior roles quarterly, and it barely moved for three years despite genuine effort. It could not have moved: the number is a function of first-rung promotions made five to ten years earlier and of who stayed. When they finally measured the first rung, the gap was visible immediately and was fixable with two cheap changes - advertising stretch assignments and requiring examples in the talent review. The lagging indicator had been hiding the leading one.
Try it yourself - this week ◐ 30-45 min total
- Find your written definition of "potential". If there is none, write the three-behaviour version and require an example per claim.
- Pull the promotion rate into first-level management by group for two cycles. Look at the first rung, not the pyramid.
- List the last five high-visibility assignments and how each was allocated. Advertise the next one.
- Ask each senior leader who they sponsored this year and what they actually did. Note the pattern in the names.
- Find anyone acting up without the title. Set an end date, then promote them or end it.
Sources covered (cited)
Taught from the promotion-equity research. This session covers:
Three questions before you go 🎯 ◐ 90 seconds
1 · Where does the "broken rung" occur?
The 2025 report found 93 women promoted to manager for every 100 men, and 74 for women of colour. Every level above inherits that smaller pool, which is why the seniority pyramid is a lagging indicator and the first rung is where to intervene.
2 · What distinguishes a sponsor from a mentor?
Only sponsorship moves promotions, because it also allocates the stretch assignments that build promotion evidence. Organisations run mentoring programmes because mentoring is cheap and comfortable, then treat the problem as solved.
3 · Your talent review has a "potential" axis with no written definition. What is it measuring?
An undefined axis does not measure nothing - it absorbs everything the room cannot say out loud. Define potential as three evidenced behaviours (learns fast in new situations, results through others, seeks harder ambiguity), each requiring a named example, or drop the axis.