Two competent people, one bad meeting
He has been at the company eleven weeks. He was hired through a six-stage process, came with a defined remit, and is good at his job in the specific way the company said it needed him to be good.
His manager has been at the company four years. She joined when the team was twelve people. She is also good at her job, in a way that is harder to describe. She has caught two near-misses in the last quarter that nobody else saw coming. From a three-line message she can read whether an engineer is stuck or stalling, and she has a feel for which customers will tolerate a delay and which will not. None of this is in any document. Most of it is not visible from outside her head.
They are in a 1:1. He has brought a proposal, a small change to how the team intakes work, the sort of thing he has done twice before at larger companies and is confident is correct. She listens, asks two questions about timing, says she isn’t sure it’s the right time, that “this isn’t really how we do things,” and that they should come back to it. What she does not say is that the proposed change would route around a customer whose last three escalations came through exactly the channel he wants to formalise away, a customer the company cannot afford to surprise this quarter. She has held that context for so long she has stopped noticing no one else has it. The conversation moves on.
Two weeks later, in a different meeting, someone who has been at the company since the second year raises the same intake question with a different framing: what would we have to be sure of about that customer before we could change this? The manager engages immediately, and the two of them work through the constraint together. His change ends up piloted on a different segment first, while the customer-impact question gets worked through separately.
He goes home that night and opens LinkedIn for the first time since he joined. She goes home thinking she handled it well and that he is, on reflection, a bit junior.
This scene reads as a fit issue. It is not. Both of them are right: her call protected a customer she alone knew about, his proposal was the kind of formalising he was hired to do. They are living in two realities inside one company: one where context lives in people’s heads and one where it has to be written down before anyone can act. The operating structure was built for the first and never adjusted for a team that now holds both. Unless someone names that, the scene keeps producing itself for the next eighteen months, until he leaves, she leaves, or both do and the company decides it just hasn’t found the right people.
What is actually happening
Underneath the friction in the meeting is a structural fact: the company is running two operating models at once. In founder-led scale-ups, this usually shows up somewhere between the first fifty and the next hundred hires. The people on each side are competent inside their own model and look incompetent inside the other. It is the engine of everything that follows.
The first model is the one that built the company. The first few dozen hires were selected, sometimes deliberately and more often by gut, for the traits that made the early stage work: ambiguity tolerance, ownership, willingness to act on incomplete information, comfort with the founder personally directing their work, speed over precision. For them this was not an ideology but how they were required and learned to operate; anything else would have killed the company in its first two years.
The second model is the one the company now needs to keep growing. The hires who come in once the founder can no longer be in every conversation are selected through a defined process for capability inside a defined role. They come with the institutional habits the company needs in order to scale: documentation, coordination across teams, structured planning, written decisions. They expect the rest of the company to work the same way, because that is the model they were hired into and need around them to be useful.
The original employee who decides in the moment on the strength of context they alone hold is exercising a real and load-bearing skill, and looks to a new hire like someone refusing to operate in the open. The new hire who writes a one-pager, takes stakeholder comment and only then proceeds is also exercising a real and load-bearing skill, and looks to an original like someone who cannot make a decision without a meeting. Each side judges the other by its own model’s standards and concludes the other is the problem. The misreading needs no bad faith; it is hardest to dislodge precisely when both sides act in good faith, because each side’s honest effort to do its job well reads to the other as incompetence.
The growth-stage literature has named this transition for half a century. Greiner’s “crisis of leadership” at the end of the creativity phase is the canonical statement. What that literature names is the symptoms. What it does not name is the feature that makes the conflict so resistant to fixing from inside the company: both populations are right about themselves and wrong about each other, and the wrongness is symmetric. Every fix that targets one side validates the other’s misperception.
A note on the binary. The two populations correlate with tenure but are not identical to it. There will be four-year hires in your company who operate fluently in the new model, and six-month hires who have gone native and operate like originals. “Originals” and “new hires” below are shorthand for two operating models, not for years on the badge.
A second note. Some originals are not carriers of valuable tacit knowledge at all; their advantage depends on the ambiguity the transition is about to remove, and using the structural frame to protect them from a performance conversation they are actually due is a misuse of it. The same caveat applies in reverse for institutional hires who are importing process the company does not need rather than the operating discipline it does. The frame is a way to see the structural misreading before it gets converted into individual judgement, not a way to avoid the judgement when it is needed.
The misdiagnoses
Every framing the company tries gets this wrong in the same direction. It treats a structural condition as a moral or developmental failing of one side or the other.
The founder calls it a culture problem and commissions a values refresh. New posters, a short list of behaviours written in the language of every other tech company’s values page. The originals shrug because they wrote the actual culture by doing it, and the posters do not describe what they did. The new hires shrug because the posters do not describe anything they can act on either. Nothing changes.
The two populations diagnose each other. The originals say the new people do not understand what got the company here. The new hires say the old guard is defensive, threatened, unable to grow up. Both diagnoses are partially correct and either side could write the other in its sleep. The result, at best, is an offsite that accelerates nothing except the participants’ certainty about who is to blame.
The interventions that follow share a single defect. Values workshops, performance review rewrites, town halls, all-hands speeches about being one team: they are attempts to fix the operating structure by adjusting its artefacts. Artefacts are downstream of the structure: they cannot move it, only encode it once it has changed. That conceptual error sits inside every fix the company has tried.
The seam: where the conflict actually lives
What looks like a culture clash is most legible in the manager-report relationship where an original-cohort manager has new-cohort reports, or the reverse. Both configurations exist in the same company, often at the same time, and the seam runs through each of them.
This is not an accident of org design. It is structurally close to inevitable, and it forms in two ways. The early hires were promoted into management on the strength of tenure and trust — they were there, they could be relied on, they had the founder's ear — and new hires were brought in beneath them, often because the company needed work the early cohort could not do at scale or rigour. At the same time, as the company professionalises, institutional managers are hired in from outside or promoted from the new cohort and placed above original-cohort reports who hold deep context but not the new model.
What happens at the seam is performance evaluation and opportunity distribution both get corrupted, the disfavoured side feels both shifts before they can name them, and they respond by closing ranks. None of this is sequential; it reinforces itself.
Adverse selection in feedback. The manager evaluates the report's work against the standards of the manager's own operating model, and the report is, by definition, working in the other one. When the manager is an original, the yardstick is the old model — move fast, do not ask, figure it out — and the new-cohort report is coded as slow, process-heavy, lacking ownership; their real contribution, making the team's work durable and scalable beyond any one person's attention, is invisible because it was never required when the manager learned what good looked like. When the manager is institutional, the yardstick is the new model — document it, communicate it, get the decision in writing — and the original-cohort report is coded as a cowboy, undisciplined, a key-person risk who will not follow process; their real contribution, the load-bearing judgement calls and the disasters quietly averted, is invisible because the saves never show up in any system, and the manager never had to operate without the process that would have caught them. In both directions the report ends up formally underperforming against criteria the company never hired them to meet.
In-group preference at the resource boundary. This is the most corrosive of the mechanisms, because it destroys value-creation directly rather than indirectly. Opportunity flows toward whichever cohort the manager recognises as competent — which is the manager's own. Under an original manager, the plum work goes to the originals: greenfield projects, high-visibility accounts, the meaty problems that build reputation inside the company, while the new hires get the maintenance and the cleanup. Under an institutional manager it inverts: the visible, career-making work goes to those who present fluently in the new model — other new hires, or originals who have gone native — while the original-cohort reports are parked as custodians of legacy knowledge, valued for what they know and passed over for what they cannot yet show. Promotions follow the same gradient either way. So does public defence. When the favoured cohort's bet goes wrong it is framed as a learning opportunity and the team rallies; when the out-group's does, it is framed as evidence they were never a fit — the improviser's miss proves the company needs process, the process-bound hire's slowness proves they lack drive. None of this is necessarily conscious; most managers would sincerely deny it. But from the disfavoured side the pattern is clearly visible, and they are correct about what they see.
The disfavoured cohort does the maths. Under an original manager, the new hires were brought in to enable growth the company could not produce without them, and they watch it reward the people whose work needs them in order to scale; the most capable, the ones with the most options, leave first, and the company loses exactly the people it spent the most to acquire. Under an institutional manager, the originals see their judgement recoded as recklessness and their context treated as a liability, and the ones who can leave take with them the tacit knowledge the company has been running on and cannot quickly rebuild. Either way the manager's side, watching the departures, finds its priors confirmed — they were never really a fit — and the loop closes.
The clique as a defensive structure. The originals are not behaving badly out of malice. They are responding to a transition they have not been invited to participate in and whose terms have not been put to them. They have agency: they hold equity and voice, and are not pure subjects of it. But they are working from incomplete information about what is being asked. Their operating model, the thing that made them load-bearing for years, is losing institutional legitimacy. Nobody has told them what the new mode requires of them, what of their behaviour is still wanted, what is being coached against. In the absence of that conversation, they form cliques. The clique is what a population does when it senses that its model is being deprecated and it has no formal voice in the deprecation. “Corporate bullshit” is the name they give to a transition they were never invited to. The worst version of this, and it is not how most of these end, is when one side pushes the other out, and the company falls back into the model it had outgrown or hardens past the founder into a generic professional shell.
The IC peer layer carries the same dynamics in quieter form. The structural multiplier of a manager is absent, which makes each instance less acute, but there are more ICs than managers and the contact between original-cohort and new-cohort is more sustained. The same mechanisms simply run through informal channels rather than formal ones: who gets pulled into important work, whose reasoning is adopted in a Slack thread when there is a disagreement, where original-cohort attention and information actually flow. The dynamics are no less real for being informal.
Why the standard fixes fail (and which ones make it worse)
The wrong diagnosis produces the wrong intervention, and these split into two kinds. Some are merely useless: they consume effort and produce nothing. Others actively make the conflict worse. They ratify one operating model as legitimate without saying what the other was for. The side whose model has just been demoted reads this accurately and radicalises against it. It matters which kind is which, because the active accelerants tend to be the interventions the company is most proud of.
The performance review system is the classic accelerant. It codifies institutional-mode behaviour as the legitimate standard and codes original-mode behaviour as deviance, without ever saying what the new operating model is or why. The originals read it accurately as an indictment of how they have been operating for years, and the new hires read it just as accurately as overdue ratification of how they already work. Neither group has been told what the company is doing, and the system has done damage on both sides while solving nothing.
The org chart change works the same way when it happens before the underlying transition has been named. Moving original-cohort managers into IC or advisory roles, inserting institutional managers above or beside them, professionalising the management layer: all of it reads to the originals as a purge dressed up in the language of org design, and to the new hires as long-overdue normalisation. The change is treating the symptom while the problem compounds.
The values refresh is the most common and the most useless. It produces new language that pleases no one and changes nothing, because what is being fought over is not values. Both groups would describe their values in roughly the same words. They are fighting over what the words mean in practice, and a poster cannot resolve that.
The all-hands intervention almost always backfires. The founder stands up and says something heartfelt about being one team, and the room gives two responses simultaneously: the originals hear confirmation that nothing is supposed to change, the new hires hear denial of what they are experiencing daily. The founder feels they have addressed it. They have not.
The actual work: making the transition explicit
The work is not a workshop. It is an act of leadership, performed mostly by the founder, sustained over quarters until internalised by the team, and embedded in how the company subsequently makes decisions about people. It begins with articulation.
The first move is to name what the company was, honestly and without condescension. Not as a phase to be outgrown, but as a real operating model that worked, produced the company that now exists, and is still legible inside it. The originals have to hear, from the founder, on the record, that what they did was real, that it required skills the company depended on, and that those skills are respected. This is the precondition for them being willing to grow and adapt. A population that believes its history is being erased will not cooperate with its successor. A population that believes its history is being honoured might. The naming has to be specific. Not “we used to move fast.” The actual behaviours: deciding alone on the strength of context you alone held, shipping before the documentation existed, telling the customer yes and figuring out delivery afterwards.
The second move is to name what the company is becoming, and why. This is the heaviest, because it has to land the entire frame. The new operating model exists because the old one cannot produce capabilities the company now needs: coordination across teams that have stopped sharing a single conversation, decisions that route without escalating to the founder, predictability for customers and partners who now depend on the company, and continuity that survives any individual leaving. These are capabilities the old model genuinely could not produce. Not because the original cohort lacked virtue, but because the old model was optimised for a different problem. The old model is not bad and the new one is not better; the company simply has a different problem now, and a different problem needs a different operating model. The originals can hear this. The originals cannot hear “we have grown past the way you operate.” The difference between these two framings is the difference between the originals adapting and the originals forming a faction.
The third move is to name what is kept across the transition. The small number of behaviours that are load-bearing for the company’s identity and that survive the change untouched. Customer obsession. Directness. Ownership in the sense of seeing a problem through to resolution. Whatever they are for this particular company. These are the behaviours the originals taught the company by doing them, and the new hires are required to learn them. This is the part of the early company the originals are not being asked to give up.
The fourth move is to name what is let go, explicitly. This is the hardest move. The behaviours being retired are behaviours the founder modelled, sometimes still models, and the founder has to publicly retire instincts they personally embodied. Shipping without writing things down used to be heroic; it is now a cost the company can no longer afford. Making the call by yourself used to be ownership; in a company with twelve other teams that depend on the decision, it is now a coordination failure. Telling the customer yes without checking used to be hustle; it is now an unpriced commitment. Until this is said, the originals cannot know which of their instincts are still wanted and which are not, and they will read every correction as a personal indictment. The originals watch the founder for the signal of what is actually wanted. They will believe the signal long before they believe any document.
What this looks like on the page, in a real founder’s hand, is two short lists naming what is to be kept and what is to be let go. Specific behaviours, not values. Not “directness” but “raising the disagreement in the room, on the day, not in the hallway afterwards.” Not “we communicate better” but “we do not commit a delivery date to a customer without confirming it with the team that will deliver.” Values posters fail because they describe what the company would like to be true. Behaviour lists succeed because they describe what people are expected to do differently from yesterday.
The new operating model has its own pathologies. Imported operating discipline can degrade into ceremony in its own way, with ownership flattened into status reviews about ownership and speed converted into theatre about speed without the thing itself. The work is not to replace the old model wholesale with imported process, but to build the minimum operating discipline the new complexity actually requires while protecting the few behaviours that still make the company itself.
The work at the seam: managers and ICs
Once the naming exists, the manager-report seam still has to be operated on directly, and the IC peer layer alongside it. The articulation creates the air cover. The structural intervention happens in who is managing whom, and in how the IC layer is structured to keep operating expertise in the company while it externalises what the originals know.
Some of the early-cohort managers will make the transition. They will learn the new operating model, develop genuine fluency in evaluating people whose competence lies in skills they themselves did not learn, and become bridge figures the company desperately needs. These managers are extraordinarily valuable and extraordinarily rare. They are worth investing in heavily: coaching, peer mentoring, explicit role design, sometimes just protected time to figure out what good looks like in the new model.
Some of the early-cohort managers will not make the transition. This is not a moral judgement. It is an observation about the difficulty of evaluating, coaching and developing people whose competence is in a model the manager has spent years operating against. The managers who cannot make this shift are not bad managers in the abstract. They were good managers in the model they were promoted in. They are now structurally mismatched to the role they hold.
Leaving an unwilling original in a management seat over new-cohort reports is among the most expensive decisions the company can make during this transition, and probably the costliest one whose cost is hidden. It metastasises. The reports leave. The manager’s clique hardens. The next round of hiring beneath that manager fails for reasons that look mysterious from the outside but are mechanically obvious from the inside, and the company spends a year recruiting into a role that the structure is actively spoiling. The cost is paid in lost hires and lost work, in market position eroded without a clear cause, and in the manager’s own slow demoralisation as their team turns over without their understanding why.
The redeployment options are real. A strong individual-contributor role can turn the manager’s tacit knowledge from a coaching liability into an asset. An advisory or principal seat institutionalises the knowledge transfer the company needs anyway. Sometimes a customer-facing role is exactly where the early operating model is still required. None of these involve a demotion in any meaningful sense, but they require the founder to invest in designing them. Most companies default to two options instead: keep the manager in role until it explodes, or manage them out. Both are worse than the third path that fits.
The redeployment, where it is possible, is the better outcome by every measure. Where it is not possible, the exit has to be managed well. The person needs the dignity and the kind of internal story that honours what they built rather than framing the departure as failure. The originals are watching. How the company treats the first manager who is moved or exited will determine what every other original concludes about how their own years will be valued when their turn comes.
The same work has to happen at the IC layer, with different mechanics and at least as much weight.
The IC originals are the larger half of the population the founder is navigating through this transition. There are more of them than there are early-cohort managers, and individually their work is often where the company’s deepest operational expertise sits. The IC originals carry the reasoning behind dozens of choices the company has been running on for years: why a contract clause exists, why two functions came to report separately, why one pricing tier was created for a single customer and then became standard. Each was a thoughtful answer to a specific problem at the time it was made, and the reasoning was never written down. That reasoning has to be externalised before its holders move on, or the company keeps running on choices nobody can audit. The principal, staff, or advisory IC role is where the conversion most cleanly happens. The new model has to produce knowledge in encoded form as it is generated, not reconstructed from memory two quarters later. This encoding is part of what the new model is for, and has to be designed deliberately or it will not happen.
For IC originals who can make the transition, the principal-IC role is the most under-used structural move in scale-ups. It lets the person keep doing what they are good at, recognised at a level proportionate to their contribution, while the company benefits from the operational expertise and the knowledge externalisation that follow.
For IC originals who cannot make the transition, the exit has to be managed with the same care as the manager-level exits. They were hired into ambiguity, paid in opportunity, and asked to do whatever the founder needed. The company they are now in does not have that texture and is not going to recover it. Managing their exit well is underrated. Done well, it serves the person, the remaining originals and the company at once: the person leaves whole, the remaining originals see their own contributions will be honoured when their turn comes, and the cultural transmission line stays intact. The founder has to lead this personally. HR cannot do it without the founder’s air cover, because the originals know who actually decides.
What the founder has to do
The fight people are having is not the fight that is actually happening.
The fight people are having is about culture, values, fit, professionalism, whether the new performance review form was designed with adequate consultation. The fight that is actually happening is between two operating models running simultaneously inside one company, unacknowledged. It converts itself daily, at the manager-report seam and through the IC peer layer, into the kind of damage the company will diagnose far too late: work that does not get done, hires it spent the most to acquire and watched leave first, market position that erodes without a cause anyone can name.
The work that has to get done cannot get delegated. It cannot be run by HR, by a new COO, or by a consultant including this one. The founder is the only person at the company with standing in both operating models, and therefore the only person who can name the transition without setting off the war.
What that means specifically: letting the team know, out loud, that the company needs to transform in order to accommodate its new reality. That both operating models are real. That one is being deliberately succeeded by the other for reasons of capability rather than virtue. That the transition is being managed deliberately rather than allowed to happen by attrition and politics. Name what is kept. Name what is let go. Name, by category if not by individual, that some people will not be able to make the transition and that the company will manage those departures with dignity.
Most founders flinch from this; it feels like a betrayal of the originals. The opposite is true: the naming is what protects them. In the absence of the naming, the transition still happens. It happens silently, through performance reviews that fault the originals for behaviours that were celebrated last year, through promotions that route around them, through the new layer of managers above them whose mandate they do not understand. They are eroded daily without ever being told what is being asked. The honest naming is painful in the moment, but it is the only frame in which their contribution is acknowledged on the record, the terms are made legible, and they can choose to adapt, redeploy, or leave whole. The alternative is not preserving their position – it is eroding their position without ever having said so out loud.
Naming, properly understood, is not communication but governance. The first naming is the company’s first official statement of which model it will reward from now on, telling everyone which behaviours stay protected and which lose institutional cover.
There is a deeper version of this risk that the founder has to absorb before any of the work lands. The originals are not merely operating in an old mode out of habit. They are still reading the founder’s revealed preferences and operating accordingly. If the founder stands up and says “we don’t commit to dates without checking” but then celebrates the rep who promises the impossible deadline and somehow pulls it off, the whole transition collapses on the same day. The originals see what gets rewarded in practice, not what gets said on stage, and they are right to. The articulation only lands if the founder is willing to align their revealed preferences with their stated ones, in the moments that matter most: the next time someone goes around the process and it works, the next escalation, the next promotion, the next exception they would once have made because it was the right call in the old model. If the founder cannot do this, the originals are correct to keep operating the way they always have, because that is still the way the company actually rewards. The hardest part of the naming work is not the speech. It is the daily discipline of being the person the speech described.
On timing. The earliest useful moment is when the founder can see two distinct operating models forming inside the company before the performance review system or the org chart has been used to ratify one over the other. The naming gets harder the longer it is deferred. More of the company’s structural artefacts have been pulled into one or the other camp’s reading by the time the founder tries to set the frame. If the conflict at the seam is already producing visible attrition, the naming is overdue and the question is no longer whether but how. The first naming is one event. The work it begins is ongoing. The articulation has to come back, in different shapes, for as long as required: at each all-hands, in each round of promotions, with each significant hire, inside performance reviews. After that it folds into the company’s standard operating language.
One thing to do this week
Not the all-hands, not yet. The all-hands comes after the thinking. The thing to do this week is to write down, in your own words and only for yourself at first, two short lists. The two or three behaviours that defined the early company and that you want kept across the transition. The two or three behaviours that defined the early company and that you are going to coach against from now on. Specific behaviours, not values. That document is the spine of every other conversation you are about to have. Until you can write it, you do not yet have the clarity to name the transition publicly. Once you can write it, the rest of the work follows from it.
If your company is somewhere between founder-led improvisation and institutional operating discipline, this is the transition I help diagnose and design.