The sunk cost is the strategy
For years, digital transformation meant consolidating every business unit onto a single platform. The promised end state: one source of truth, one upgrade path, one vendor to hold accountable. In practice, I've watched this playbook produce the opposite. A global manufacturing client spent two years mapping 14 regional order types into a "standard" model, only to discover the model broke every local pricing rule that made the business profitable. I saw the same pattern in a distribution company where a single-vendor mandate forced three business units into one order entry workflow, adding 45 seconds to every order and silently eroding margin. The customization debt didn't vanish; it moved into the integration layer.
The composable alternative is quieter but harder
The current shift is not about microservices buzzwords. It's about recognizing that some capabilities are commoditized — finance, HR, core record keeping — while others are genuinely differentiating: pricing, supply planning, field service. Attempting to standardize the latter through a single ERP is where transformation programs die. I've started advising a different sequence: keep the commodity layer boring, and rebuild the differentiating layer as composable services with clear data contracts. This is slower in the first six months because it demands real domain modeling. But it survives executive turnover because each service has an owner, a client, and a P&L link.
The lesson
The specific development I'm seeing is not a vendor announcement — it's the unwinding of failed big-bang programs. Boards are no longer rewarding "transformation theater." They're funding smaller, reversible cuts. If you're asked to justify a multi-year platform migration today, the strongest proof isn't a roadmap slide. It's a working service you can show in production inside a quarter. Start there.