Why Business Middle‑Platform Projects Often Fail: Lessons from Alibaba’s Reorganization

Hello everyone, I am Chen (a humble person).

Recently, during an internal meeting, Alibaba’s “Xiaoyaozi” introduced the company’s new “1+6+N” organizational reform. While the specific new units are not the focus, a key point was the significant staff reduction in departments responsible for building and providing middle‑platform capabilities.

The concept of a middle platform has been discussed at Alibaba for several years; it was not originally created by Alibaba. Jack Ma visited the Finnish game company Supercell, researched the idea, and brought the middle‑platform concept back to Alibaba, sparking a wave of digital middle‑platform enthusiasm.

Consequently, many firms such as JinDie, Yonyou, and former Alibaba employees have started ventures offering digital or data middle‑platform solutions. Alibaba’s latest reform, however, seems to suggest that after persuading all software vendors, the company is now stepping back.

I have previously explained the middle‑platform idea, including the core principle of sinking common capabilities to quickly build upper‑layer applications; these ideas themselves are not problematic.

Nevertheless, in most real‑world projects for enterprise clients, digital middle‑platform initiatives either collapse, fail, or do not succeed. The internet‑originated middle‑platform concept does not fit well in traditional enterprises, and data middle‑platforms often become a confusing mix of BI, master data, data exchange, and other capabilities, failing to deliver consolidated data assets as services.

The business middle‑platform concept is even harder to articulate. If you look at Alibaba’s overall digital middle‑platform architecture, you will see a collection of micro‑service centers—user, product, order, settlement, member, etc.—exposed as interfaces, which Alibaba refers to as its business middle platform.

For enterprises that already have years of IT systems, asking them to build a business middle platform from scratch is unrealistic. Instead, when you have a monolithic system, exposing reusable capabilities as shared services (SOA) is often more effective and faster for constructing upper‑layer business and processes.

We should also reflect on why advancing a business middle platform is so difficult. It is not merely about building a platform or applications; it also involves business and organizational transformation. Even with an independent middle‑platform department and an innovative application team, many issues persist.

First perspective: Adding new capabilities to the middle platform hampers agility. From a business viewpoint, an innovation team expects to reuse existing middle‑platform services, but when those services are insufficient, they must develop new ones. Deciding whether to build the capability themselves or discuss adding it to the middle platform leads to prolonged debates, eroding the agility of application development.

Second perspective: The middle‑platform team becomes detached from the business. After organizational changes, the dedicated middle‑platform department does not own any front‑line business, so its members spend time pondering which new capabilities to add, without practical insight from actual business usage.

In short, middle‑platform capabilities cannot be imagined out of thin air; they must be derived from the massive IT systems built over time. If you continue to rely on large‑scale IT system construction to accumulate capabilities, you will find that there is no standard “business middle platform” product. It is essentially an aggregation of existing business systems and micro‑service capabilities, similar to the traditional SOA shared‑service platform.

Therefore, when guiding traditional enterprise IT transformation, building a data middle platform is acceptable, but avoid claiming to build a business middle platform because the concept is vague and hard to implement. Instead, discuss constructing a capability‑sharing or capability‑aggregation platform, which is perfectly reasonable.

That’s all for today’s brief sharing; I hope it inspires you. Goodbye!

Finally, one more sentence (please follow me).

Each of my articles is carefully crafted. If this article helped or inspired you, please like, view, share, and bookmark—it’s the greatest motivation for me to keep going!

Additionally, my Knowledge Planet is now open. Reply with the keyword “Knowledge Planet” to the public account to get a 30‑yuan coupon; joining costs only 99 yuan, a small price for the huge value, including Spring full‑stack practice series, billion‑scale data sharding practice, DDD micro‑service columns, and many other premium content.