# ============================================================================= # Published - The Puberty Stage of Change Management (verbatim) # ============================================================================= # An article authored by Cedric Ancellin, first published on this site. # Text is preserved as written (including original wording and any typos). # Governed by ../LAWS. Verbatim files are exempt from the em-dash rule. # # Added on 2026-11-12. # ============================================================================= title: "The Puberty Stage of Change Management" author: "Cedric Ancellin" published: "2026-11-15" platform: "zc8.com" linkedin_url: "" # set once cross-posted to LinkedIn cover_image: "/assets/articles/the_puberty_stage_of_change_management.jpg" topics: ["change management", "scaling", "Series A", "Series B", "organizational growth", "fair process", "early employees", "career ladder", "pay equity", "process", "time tracking", "DORA", "SPACE", "psychological safety", "founders", "board management", "AI coding agents"] summary: > A growing company reaches a stage where it outgrows the way it works while its people still behave as they did around one table. Three researched principles, fair process, participation, and endings before beginnings, applied to early stars, process, measurement, fear and founders. body: | On the way to adulthood, there is a moment when body and mind change faster than the person inside them. For a company, it comes when headcount doubles, a new C-suite arrives, and growth goals widen a roadmap already full of promises. The company is suddenly too big for the way it works, but its people still behave as they did when everyone fit around one table. That is puberty. It is awkward, emotional and harder than most engineering decisions. It is also well studied. In 1972, [Larry Greiner](https://hbr.org/1998/05/evolution-and-revolution-as-organizations-grow) described how companies grow through calm evolution broken by crises. Founder creativity leads to a crisis of leadership, the direction that solves it to a crisis of autonomy, and the controls that solve that to a crisis of red tape. In [Growing Engineering from Startup to Series B](https://zc8.com/published/growing_engineering_from_startup_to_series_b), I described what the organization should look like after a Series A, and warned that the steps are extremely hard to implement. This article is about the hard part. Almost every technique below applies one of three principles. **Fair process.** [W. Chan Kim and Renée Mauborgne](https://store.hbr.org/product/fair-process-managing-in-the-knowledge-economy-hbr-classic/R0301K) showed that people commit to decisions they disagree with when the process behind them is fair. That takes engagement, explanation and expectation clarity. Ask the people affected for input, tell them why, even when their idea lost, and make the new rules explicit. Fair process is not consensus. You still decide. But skip any part, and even a good decision will be resisted. **Participation.** In 1948, [Lester Coch and John French](https://doi.org/10.1177/001872674800100408) introduced the same change to textile workers in Virginia with different levels of involvement. The groups that helped design it recovered far better than the group simply told. People resist change done to them much more than change they helped shape. **Endings before beginnings.** [William Bridges](https://wmbridges.com/about/what-is-transition/) separated change, the new org chart that happens in a day, from transition, what people go through to accept it. Transition starts with an ending, passes through a confusing in-between, then reaches a new beginning. Most leaders announce the beginning and skip the ending. People do not resist the change, they resist the loss. Name what is ending before you sell what is starting, and expect performance to dip before it rises. So do not change everything at once. In my experience, stacked changes merge their dips into one long valley that people read as the company failing. Sequence them. I watched the new management of a healthy, growing company ignore these three principles and impose their vision and rules as decrees, and within nine months they destroyed the company beyond salvation. They were not incompetent, just certain of their playbook. These principles matter most to new leaders. If you are the new executive, learn before you act. [Michael Watkins](https://hbswk.hbs.edu/archive/hbsp-audioconference-the-first-90-days-mastering-successful-management-transitions) observes that new leaders often learn the business but fail to learn the organization. Ask the people who built the systems why things are the way they are before deciding they are wrong. Bring your experience, not your last company's methods. Earn credibility with [early wins](https://hbr.org/2009/01/new-leader-get-early-wins) the team already wanted, then spend it on harder changes. The first people to feel the change are the ones who built the company. Your early hires knew the whole stack, fixed anything at any hour, and became stars. Then you specialized the teams, one on the database, another on the frontend. On paper it was a promotion. They took it as an attack, because their value was knowing everything. [Molly Graham](https://review.firstround.com/give-away-your-legos-and-other-commandments-for-scaling-startups/) calls the answer giving away your Legos. To grow as fast as the company, you give away your job every couple of months and build the next thing. Say it before the reorganization, and name the next thing for each person. Build a real technical ladder, with staff engineers paid like managers at the same level and given real authority over standards, architecture and the hardest cross-team problems. A title without decisions is a consolation prize. And ask your stars where team boundaries should be. It turns the likeliest resisters into authors of the change. Then diagnose honestly. [Elad Gil](https://growth.eladgil.com/book/recruiting/old-timer-syndrome-early-employees/) warns that CEOs often look for a place where an early employee will not do much harm, instead of a place where they will excel, and that this is usually the sign to part ways. Parting with respect and the right package is kinder than a role that slowly diminishes them. I learned this the hard way. Some of my right hands from the bootstrapped years stepped up to the challenge, while others felt threatened by the change, and I could not get them to embrace it. We were like family and had survived so many storms together, so I tried too hard to find them a place, and we stayed together. Years later, they still had not found it. On paper their careers looked stagnant, and opportunities elsewhere had become hard to find. In reality, they are incredibly talented. Had we parted then, in their prime and with only successes to their name, their careers could have gone much further. Not having the strength to accept reality at the time, and to support them toward a bright future elsewhere, is one of my biggest regrets. Rapid growth splits a company into those who came before and those who came after. Old-timers feel newcomers do not understand the culture, and newcomers feel shut out. Mix them on purpose. Pair every new hire with an early employee, and never staff a project from only one group. Keep a living record of why decisions were made, as I described in [From a Novel Helper to an IT Oracle](https://zc8.com/published/from_a_novel_helper_to_an_it_oracle), so newcomers stop depending on old-timers and old-timers stop feeling ignored. The career ladder will also expose money. To hire fast, you paid newcomers market salaries, often above what early employees earn, while early employees hold more equity and remember working for less. Each side sees only its own disadvantage, and once levels are written down, every gap feels personal. Audit pay against the new levels before you publish them, fix the worst gaps first, and explain the rule going forward. I have seen companies lose their best people this way. A respected, valuable employee asks for a raise and gets a minimal, defensive one. They leave, and their job is posted at a higher salary. That is insulting to the person who left, but the real damage is that everyone in the company learns about it. They conclude that management values newcomers over proven veterans, and your key people start looking elsewhere, because they assume there is no growth for them inside. The same divide runs between business and engineering. In a startup, sales walks over to a developer and it is done by the afternoon. Later, everything goes through product and a ticket. That protects the architecture, but the business sees a wall and goes around it, through a favorite developer or the CEO. I watched this happen more than once. A customer advocate needs an urgent fix for a marquee customer, and the request lands behind dozens of others with no date. So they call the developer who built the original integration and ask for a favor. The fix ships in a few days and breaks unrelated modules the following week. Nobody on the product team knows the change exists, so the outage drags on while they hunt for the cause, and other customers pay for one customer's favor. The developer is not the main culprit for helping, and neither is the advocate for asking. The process is, because it offers a wall instead of a date, and no path to escalate. The wall holds only if the business can live with it. Make the queue visible, with each request's position and target date, and commit to a response time. At first, keep an escalation path for critical work and quietly write the tickets the business should have written. Give business leaders a seat at roadmap reviews. They will not win every argument, but they will stop going around the process. Inside engineering, every new process will be called unproductive, and at first it is. The engineer who once pushed a fix in ten minutes now waits two days for a review, and tells everyone at lunch. This is Greiner's crisis of red tape arriving early. **Explain what it protects.** A ticket gives finance the auditable record it needs for capitalization and GAAP reporting, as I described in [Capex and Opex for Engineering](https://zc8.com/published/capex_and_opex_for_engineering). A code review keeps the next [Ugly Duckling](https://zc8.com/published/growing_engineering_from_startup_to_series_b) out of production. A process nobody can explain deserves to go. **Let teams design it.** Set the outcome, such as an audit trail for SOC 2, and let one team decide how. Pilot it with them, then roll it out with that team as its advocates. **Review and remove.** Retrospect every process and delete those that do not pay for themselves. Nothing earns more credibility than a leader who deletes a process they introduced. AI tools now land in the middle of puberty, and they change the work engineers were proudest of. Writing code by hand was the craft, and agents now do much of it. Treat AI like any process change. Name what is ending, let each team decide how agents fit its workflow, and pilot first. Give your best engineers the new craft, meaning the architecture, the review and the knowledge base the agents work from. I wrote about that shift in [The Junior Developer Is Gone. The Junior Product Engineer Isn't.](https://zc8.com/published/the_junior_developer_is_gone) Measurement is where process feels most personal. The day a dashboard shows hours per engineer, someone will screenshot it and ask who it is for. A [meta-analysis of 94 samples](https://onlinelibrary.wiley.com/doi/abs/10.1111/peps.12514) found that monitoring raises stress with no evidence of better performance, and that transparent, less invasive monitoring is received better. Measure the work, not the person. Hours go against tickets to value the work, not to rank people, and delivery metrics stay at team level. [DORA](https://dora.dev/guides/how-to-empower-software-delivery-teams/) states that delivery performance is not an individual measure, and [SPACE](https://dl.acm.org/doi/10.1145/3454122.3454124) that productivity is never one metric. Turn one into a personal target and Goodhart's law takes over. Design the tracking with the people tracked, and share the data. In a [survey at a Danish engineering company](https://papers.academic-conferences.org/index.php/eckm/article/download/2902/2377/9748), even employees in a high-trust culture viewed monitoring as negative to neutral, and the researchers concluded it works only when everyone affected helps design it and knows exactly what is collected and why. When engineers see the numbers defend their estimates, tracking becomes evidence on their side. Under it all sits unspoken fear. [Job insecurity](https://pubmed.ncbi.nlm.nih.gov/12148956/) damages attitudes, health and commitment. [Amy Edmondson](https://dash.harvard.edu/entities/publication/13a7b031-0fdd-45ec-a7e0-2b80e2bc679f) calls the opposite of that fear psychological safety, and her research links it to team learning and performance. During puberty, every closed-door meeting reads as a layoff list, so safety drops exactly when you need it most. Put expectations in writing, with the ladder and a growth plan for each person. The first time a number looks bad, fix the process, not the person, because everyone is watching. Give the first promotion in the new structure to someone who was there before it. If layoffs are coming, say so and how decisions will be made. If not, say that too. In a small company, culture spreads by proximity. As it grows, the founders' culture stops reaching the edges, and the hallway version of a decision travels faster than the official one. Repeat the reason for each change until you are tired of hearing yourself, because that is when the edges start hearing it. Brief managers before the team, so they can answer questions. That works only if managers can carry it. Most will be engineers who have never managed anyone. Train them before they start, give them a coach, and make it safe to return to engineering. Puberty is hard on the parents too. Leading through structure asks founders three unnatural things. **Stop being the entry point.** Early employees will walk into your office to escalate decisions their new manager already made. [Julia Austin at Harvard Business School](https://www.library.hbs.edu/working-knowledge/feeling-separation-anxiety-at-your-startup-five-tips) advises founders to listen, redirect them to their manager, and coach them to raise it there rather than commiserate. Every time you overrule a manager for an old friend, the management layer loses authority. The reverse holds too. Your ideas become tickets, and when the team sees you wait in the same queue, the rules become real. **Give away your own Legos.** Keep what only you can do, and give everything else an owner, with an outcome rather than your method. Judge the result, coach in private, and never take the work back in public, or nobody will want that job again. **Prepare the board.** Investors funded growth, and the dip will look like a miss. A board needs a carefully laid-out vision, with clearly defined outcomes, good and bad, and metrics it can plug into its models, not speculation. You are the executive. The board approves the plan and holds you to it. Present a solid case it can sign off against a roadmap and budget, including what will slow down and for how long. Then bring the evidence as execution develops, and report any deviation together with the plan to steer back on course. An announced dip reads as a plan. A surprise reads as a problem. A good board gives you the means to succeed and the rope to hang yourself. A bad board involves itself in executive decisions, and I have only seen that end in failure. In the early stages, everyone had access to everyone, and we debated everything, often starting from speculation. It worked with my team, so I brought the board and investors into the same debates to brainstorm with them. That was a mistake. I brought them confusion when all they needed was a plan written in their language. A few signals tell you whether it is working. Regretted attrition among early employees should stay low. Requests that bypass the queue should fall. New hires should ship sooner, and delivery metrics should recover after the dip. If they move the wrong way for a quarter, slow down and find what you skipped. When you are too close to see clearly, bring in a neutral advisor who will say what nobody inside will. Puberty ends. Led well, it leaves an adult company, where knowledge does not live in three heads, business and engineering trust a process instead of a hallway, and people grow inside the structure instead of fighting it. The best adults keep a teenager's energy and add judgment. The best companies keep a startup's spirit and add structure. transcription: "As written by Cedric Ancellin, first published here 2026-11-15."