An enterprise that formalises its developmental relationships buys wider access and worse outcomes; one that leaves them informal buys better outcomes and reproduces the existing distribution of who gets developed. Almost no enterprise states which trade it has made.
Two statements about developing people are each defensible, and no enterprise can act on both at full strength. The first: the relationships that change a career are the ones nobody arranged, where a senior figure decides unprompted that a junior person is worth uncompensated attention, then explains the reasoning behind decisions, argues for them in rooms they are not in, and tells them the truth. The second: a relationship nobody arranges reaches only the people already close enough to be noticed.
Both survive in the material used to train managers in this discipline, and they survive separately. Four treatments assert without qualification that informal developmental relationships outperform assigned ones and that formalising them degrades what works. A fifth asserts with equal confidence that a formal programme is the only way the relationship reaches those who cannot obtain one for themselves, and that informal arrangements disproportionately benefit people who already resemble the senior party. No page reconciles them, or notes that an enterprise between them is choosing.
That silence is the finding. A contradiction surviving inside a discipline's own teaching material is not a gap in the literature; it means practitioners hold two operational instructions and no rule for choosing between them. Without a rule, enterprises run both, keep them in separate parts of the building, and never put them on the same page. The composition of the next executive cohort is then set by an allocation process nobody designed, priced or approved — a distribution of the enterprise's scarcest asset, the discretionary attention of those who already hold senior judgement.
The Strategic Context
Consider a regional dairy cooperative, hypothetical in every detail. Its leadership came up through the yards, the collection routes and the processing floor, and development happens there as it always has: a plant supervisor sees something in a shift leader and, over four years, puts them in front of problems slightly beyond them and defends them when they get one wrong. The cooperative has never spent a cent on this and could not name it as a process.
It also recruits graduates with food science and supply chain degrees, most of whom did not grow up on farms. They arrive competent and stay invisible — no senior operator decided against them; nobody decided about them at all. Five years on, the succession slate mirrors the current leadership, and the board that asked for a broader one is told internal candidates are scarce.
Formalising fixes that directly: a programme places the graduates in front of the general manager on a schedule. It also converts an act of judgement into a diary commitment, and the supervisor who used to argue for his shift leader in the promotion meeting becomes one of eleven people with an assigned mentee and a quarterly form.
The same structure appears where pairing is already administrative. An urban emergency ambulance service rosters crews — formalisation by another name, since it assigns who spends the shift beside whom. It can watch both instruments at once, because alongside the roster runs a second network: the senior clinician a newer paramedic rings at three in the morning about a decision they are unsure of. That relationship is on no roster, is unevenly distributed, and is the one the service's clinical judgement depends on.
In a project-based enterprise this operates at three levels. The project holds the developmental assignment — work slightly beyond someone's proven capability. The programme is where the same people persist alongside each other long enough for a relationship to acquire depth. The portfolio decides how many can be sustained and what they point at. An enterprise that moves people between projects at high frequency has taken a portfolio decision that eliminates the programme-level condition, then wonders why the formal scheme carries the whole load.
What Leaders Commonly Misread: Treating This as Programme Design
The standard resolution offered inside the discipline is elegant and does not work. Do not assign pairs, it says; create the conditions under which informal relationships form — cross-site forums, mixed-level teams, secondments — and let self-selection do the matching.
This is good practice and not a resolution of the trade. It widens the surface on which recognition operates without changing what does the recognising. Self-selection still runs on similarity — shared background, the same route in, the same shift, an accent that signals the same origin. Nobody intends it and no policy causes it; it is what an unmanaged selection process produces when the selectors come from a narrow population. An enterprise adopting that answer has widened the input and left the allocator untouched, buying a smaller version of the same trade rather than an escape from it.
The second misreading is measurement. Because the formal instrument is visible, it is the one counted: pairings made, sessions held, completion rates, satisfaction scores. Each measures the shell, not the content. An enterprise can report full coverage while every consequential act of sponsorship happens outside the scheme.
Reframing the Issue
The useful question is not whether developmental relationships work. It is what each instrument actually purchases, and for whom.
The formal instrument purchases coverage, legitimacy and a defensible answer to who was given access — the ability to say truthfully that the person who did not grow up in the industry was put in front of someone senior. What it cannot purchase is the senior party's discretionary investment, because that is what assignment removes. The value was never the meeting; it was that someone chose to spend a scarce, uncompensated resource here rather than elsewhere. Assignment substitutes obligation for choice, then measures attendance.
The informal instrument purchases depth, advocacy and the transfer of reasoning rather than conclusions. What it cannot purchase is distribution: left alone it allocates senior attention along lines of resemblance, so an enterprise relying on it draws its future leadership from the same distribution as its present leadership, less attrition, indefinitely.
These are two instruments with different yields, costs and beneficiaries, and an enterprise may rationally buy different proportions of each for different populations. What it may not do is buy both, conceal that it has, and report only one.
This article does not address the prior constraint — the cap portfolio composition places on how many leaders can be grown, whatever the budget says — which is the subject of [Related article: Your Pipeline Is Capped by Your Portfolio, Not Your Budget]. Here the assignments are assumed to exist, and the question is who reaches the person who can secure one.
Two Systems, One of Them Unstated
The visible system carries the reporting; the real one carries the succession
Most enterprises of any size run both at once. A centrally administered scheme exists, with enrolment and a cycle. Alongside it, the executives actually growing their successors do so as they always have, keeping those relationships out of the scheme, because entering them would attach reporting to what they regard as personal judgement.
The two systems have different populations, intensities and consequences, and only one appears in the board pack. When succession comes, the names surface from the unreported system, and everyone senior understands the reported one to be a different activity.
The trade is not uniform across the population
The most common error after failing to state the trade is stating it once, for everyone. Broad access matters most where whole categories never come to senior notice — early-career entrants, non-traditional backgrounds, remote sites, shift work. Depth matters most where judgement is being transferred in a few consequential roles and a quarterly conversation is worth little.
An enterprise can therefore run the formal instrument where distribution is the binding problem and protect informality where transfer of judgement is — provided it says so, funds both, and does not report one as the other.
What the trade does not cover
The relationship, once formed, becomes something the enterprise depends on and does not record — a separate asset with its own failure mode: capability residing in a specific pair, not transferable to either party's replacement, written off by a restructure in which nobody resigns. That is the subject of [Related article: The Capability That Lives in Pairs]. The trade examined here is about who is granted access; that article is about what is destroyed when access already granted is severed.
Nor does it examine whether the reward architecture pays senior people for developing anyone, or what a mix of collective and individual reward reveals about what an enterprise believes a team to be — the subject of [Related article: The Reward System Tells You What You Think a Team Is]. The reward system is taken here as given.
Decision Framework: The Access-Versus-Return Statement
The access-versus-return statement is a single page, signed by the executive accountable for capability and tabled at the board, carrying five items and nothing else.
One: the declared trade, by population. A sentence for each population stating which instrument the enterprise relies on and why — that at the first two levels of supervision it buys access and accepts a lower return, say, while for the successor pool it buys return and accepts narrow access. A statement naming one instrument for the whole enterprise is almost certainly untrue.
Two: the access count, taken by name. For each population, how many people have a developmental relationship with someone two or more levels senior — counted by naming individuals, never by counting enrolments. The gap between that count and the enrolment number is the size of the enterprise's self-deception.
Three: the return measure, expressed as an event. What each relationship should produce, stated observably: an assignment secured, a decision the junior party was present for, a promotion argued in a room they were not in. Satisfaction scores do not qualify; they measure the meeting, not its consequence.
Four: the named owner of the unreached. Whoever is accountable for the population the chosen instrument does not reach, and what they will do about it. Every trade leaves somebody out, and the statement is honest only if it names who.
Five: the review trigger. The date, and the event that forces early review — typically a restructure, a site closure or a change in the senior cohort.
The statement is verified with one exercise, runnable in a fortnight. Build two lists: everyone enrolled in the scheme, and everyone named when senior leaders are asked privately whose careers they are advancing. A name on the second list alone shows succession happening outside the reported system; a name on the first alone is a relationship the enterprise reports but does not receive. If the lists barely intersect, it runs two systems and has never said which one it relies on — the condition this instrument exists to end.
From Strategy to Execution
Immediate. Build the two lists. Do not announce a programme change first; the second list is obtainable only while people believe the answer carries no administrative consequence. Present both unaltered to the executive team, including the names on neither.
Medium term. Write and sign the statement for each population, and change the reporting so the board sees the access count by name rather than the enrolment number. Fund the instrument the enterprise says it relies on, including the unglamorous part: senior time protected in calendars that are already full.
Long term. Treat the composition of the senior cohort as the control variable it is. An enterprise whose informal channel allocates by resemblance changes its output only by changing who does the recognising — a decade-scale decision about appointments, not a programme decision about mentoring. One unwilling to make it should stop calling that channel merit-based and name it accurately: a proximity channel it has chosen to keep.
Signals to Monitor
Watch for a succession slate matching current leadership more closely than the internal candidate pool does — the signature of allocation by resemblance operating upstream of selection. Watch for the formal scheme's reported coverage rising while the names discussed in succession conversations stay constant.
Watch for senior leaders who decline to enter their real developmental relationships into the scheme; that is information, not non-compliance, telling the enterprise which instrument they consider consequential. Watch for a site, shift or intake cohort that has never produced an internal candidate, and for the explanation that suitable people are scarce. And watch for reorganisations that increase how often people change teams, removing the duration informal relationships need and shifting the whole developmental load onto the scheme.
Questions for the Leadership Team
- If our most senior operators privately name the people whose careers they are advancing, how many of those names sit in the formal programme, and how many in populations we say we are trying to reach?
- What proportion of each population has a relationship with someone two or more levels senior, counted by name rather than enrolment?
- In our last three succession decisions, where did the successful candidate's advocacy come from, and was it a relationship the enterprise arranged?
- Which site, shift or intake cohort has produced no internal candidate in five years, and what is the mechanism behind that?
- What does an hour of a senior operator's discretionary attention cost us, and how many such hours a year are allocated through a process anyone approved?
- If our informal channel allocates by resemblance, what must change about who holds senior roles today for its output to differ in ten years?
Closing Perspective
The trade cannot be avoided and can be made deliberately. An enterprise may decide broad access is worth a lower return per relationship, or depth in a few consequential roles worth a narrow distribution, or — most defensibly — that different populations warrant different answers, funded accordingly.
What it cannot do is leave the choice unstated and still claim to manage capability. An unstated trade is not neutrality; it is the informal instrument running at full strength behind a formal one that absorbs the reporting. The board is shown a coverage figure and told it describes how the enterprise develops people, when it describes compliance with a scheme.
The responsibility is narrow and inescapable. Someone must name which trade this enterprise has made, for whom, and who has been left outside it — then defend that answer to the people it excludes.
About EraNorth Insights
EraNorth Insights publishes practical analysis on strategy, projects, operations, transformation and decision intelligence for professional and organisational use. About EraNorth.