Week 15, 2026—Issue #408
🤝 Organize Around the Scenario
File this under #customervalue
Last week I argued that AI is pushing organizations toward smaller, more autonomous units (#407). Large firms will decentralize into networks of lean teams. The long tail of small businesses will keep growing. Either way: more units, less headcount per unit.
That argument ends with an assumption we didn’t examine. Namely, that all these small, independent units will somehow work together.
They won’t. Not automatically.
The coordination problem
Traditional coordination structures are optimized for large organizations. Approval chains, governance layers, program offices; these are tools for managing complexity at scale. They are expensive and slow, and small autonomous units have neither the patience nor the resources for them.
But customer problems don’t get simpler just because your teams do. A customer who wants a smart home doesn’t want to deal with separate vendors for each appliance, platform, and service. They want a coherent experience. Delivering that requires multiple units working together. The more autonomous those units are, the harder that becomes.
This is the problem that fragmentation leads toward. Nobody has an obvious answer for it yet.
An unlikely source
A Chinese appliance company might not be the first place you’d look for organizational innovation. But Haier — the world’s largest home appliance maker — has been running one of the most radical management experiments of the past two decades.
Starting around 2012, they restructured into thousands of small, independent microenterprises, each operating as its own P&L center. Rapid entrepreneurship, strong accountability, and a company that kept growing while staying nimble. It all made sense.
Get Andreas Holmer’s stories in your inbox
Join Medium for free to get updates from this writer.
But it created a problem. Because when these units got real independence, they optimized for themselves. Haier’s microenterprises were working on their own terms, but they also stopped cooperating. Meanwhile, customers were asking for things no single unit could deliver alone.
The EMC model
Haier’s response was to stop organizing around products and functions and start organizing around customer scenarios. They called these groupings Ecosystem Micro-Communities, or EMCs.
An EMC is a temporary, voluntary, contractual alliance of units assembled around a specific customer need. What holds it together is not a manager but a shared financial stake. Each unit has a contractually defined share of the outcomes. If one unit fails, everyone feels it. And if the scenario succeeds, everyone benefits. The incentive shifts from “protect your own unit” to “make the whole thing work.”
…
This isn’t to say that the coordination issue is solved. On the contrary. Making EMCs work at scale requires significant infrastructure investment, and replicating what Haier has built is not trivial. But the underlying idea — that independent units can align around customer scenarios, bound together through shared outcomes — is worth paying attention to as organizational fragmentation continues.
/Andreas
PS. One additional thing worth noting: Haier regularly invites external firms (i.e., companies not owned by Haier) into their EMCs. Its enterprise boundary is porous by design.
Sidenotes
🏢 Institutional AI
A16Z makes a useful distinction between individual AI and institutional AI. The first makes individuals more productive. The second ensures that added capacity is aligned and used well across the organization. As organizations fragment into smaller units, the institutional question (i.e., how collective capacity gets directed toward shared outcomes) becomes the billion-dollar question on everyone’s mind.
🤝 Coordination Layers
AI coding tools may speed up writing code, but not necessarily shipping software. As Eran Stiller writes in InfoQ, the bottleneck has shifted toward clearer specifications and stronger verification. When coding gets cheaper, coordination matters more. The same logic applies one level up: as AI makes individual units more capable, the challenge shifts to how those units combine.
How can we build better organizations? That’s the question I’ve been trying to answer for the past 10 years. Each week, I share some of what I’ve learned in an email newsletter called WorkMatters. Back issues are marinated for three months before being published to Medium. This article was originally published on April 10, 2026. If you are reading this, you are missing out. Subscribe to email updates at www.workmatters.studio.
