Nobody in your organisation knows who can say yes. Decisions get escalated, deferred, or made by whoever cares most and pushes hardest. The org chart says one thing. The actual decision flow says another. There’s a shadow hierarchy — an informal power structure that determines what actually gets decided and by whom — and it’s invisible from the outside and exhausting from the inside. The conventional fix is “empower people to make decisions.” But empowerment without architecture is chaos. The problem isn’t that people lack permission to decide. It’s that the decision rights, the information needed to decide well, and the accountability for outcomes aren’t in the same place.
Decision rights, information, and accountability have to sit in the same place
Decision architecture has three components that must be co-located: the right to decide, the information to decide well, and the accountability for the outcome. In most growing startups, these three things are in different places. The right to decide is formally distributed — the org chart says the VP of Product owns product decisions — but informally concentrated, because the founder overrides when they disagree. The information to decide well is scattered across teams, tools, and people’s heads — nobody has the full picture without assembling it from multiple sources. The accountability for outcomes is assigned after the fact, usually to whoever is most visible when things go wrong. When these three things aren’t co-located, decisions either don’t get made (the decider lacks information or confidence), get made badly (the decider lacks information), or get overridden (the person with informal authority disagrees). The result is a decision system that produces high latency, low quality, and zero accountability — and in a climate tech company running four operating models simultaneously, the latency compounds across every cross-functional decision.
The shadow hierarchy serves the people who hold power within it
The founder who overrides decisions isn’t doing it maliciously. They have better context, better judgment, and the outcome matters to them. The problem is that their informal authority destroys the formal system without providing a scalable alternative. But giving up that informal authority feels like accepting worse decisions, which feels irresponsible. At ICEYE, every significant product decision that required coordination across hardware, software, government, and commercial operating models had no clean authority structure. The founder’s involvement wasn’t ego — it was the only way to get the cross-functional context. The informal architecture was the only architecture.
Meanwhile, the people who’ve learned to navigate the shadow hierarchy have an advantage. They know who really decides. They know which meetings matter and which are theatre. This navigational knowledge is a form of organisational capital they don’t want to lose. Both the people with informal power and the people who know how to access it have incentives to maintain the current system. The people who suffer are the ones who take the org chart at face value — often the newest hires and the most capable senior leaders, who expect the formal authority they were given to mean something. This is why senior hires keep leaving: the gap between their formal authority and the informal reality makes the role structurally unworkable.
The diagnostic test takes one question asked three ways
Ask the founder how a specific product decision got made last quarter — who proposed it, who approved it, who executed it, how long each step took. Then ask two other senior leaders the same question independently. If you get three compatible accounts, the decision graph is legible. If you get three different stories and each person believes theirs is definitive, the architecture is improvised daily. Capella Space’s October 2023 CEO replacement is the pattern’s endpoint: when the informal decision architecture can no longer support the company’s procurement cadence, the person at the centre of the informal system gets replaced — not because they’re bad, but because the architecture failed first and the person got the blame.
In climate tech specifically, decision architecture failure is amplified by the number of competing decision domains. Earth observation companies make decisions across satellite operations, analytics product, commercial sales, and government contracts — and the authority rarely matches the expertise. Carbon capture companies split between R&D decisions and deployment decisions with different time horizons. Energy markets companies layer regulatory, technical, and commercial decisions that interact in ways the org chart doesn’t reflect. Each additional operating model multiplies the decision seams, and each seam without an explicit authority structure defaults to the founder.
The fix is making the informal formal, not adding more empowerment
Decision architecture has to be designed explicitly, not inherited from the org chart. For every major decision type — product priorities, hiring, pricing, resource allocation, customer commitments — the architecture needs to specify three things: who decides (one person, clearly identified), what information they need (and where it comes from), and who is accountable for the outcome (which should be the decider, or the system is broken). The most effective decision architectures are simple — a one-page document that maps the top twenty decision types to specific roles, with clear escalation triggers. The key is making the informal formal. If the founder is going to override product decisions when technical architecture is at stake, build that into the architecture explicitly. A decision architecture that reflects reality — including the founder’s involvement — is infinitely better than one that pretends the org chart governs decisions when it clearly doesn’t.
For investors, decision architecture is one of the most diagnostic elements of organisational due diligence — and the hardest to assess from the board seat. A portfolio diagnosis that maps the actual decision flow almost always reveals why execution feels slow when the strategy and the team are both sound.
If a product decision that should take a day is taking three weeks, the bottleneck isn’t the decision — it’s the architecture around it. Map the top twenty decisions: who actually decides, with what information, at what cadence.