After American airline deregulation, Continental's planning model still calculated what fares ought to be. It added an allowed return to estimated route costs, producing prices that had made sense to negotiate with a regulator. In competition, those calculations could not establish what customers would pay. Yet the familiar model directed attention toward which aircraft to buy, leaving the assumed fares largely unquestioned.
Richard Rumelt calls this inertia of routine. A trusted method carried the old economic world into new decisions. The first task was to replace that way of planning, with suitable information and expertise. His broader answer is to rebuild whatever prevents an effective response. The problem might instead be culture, an incentive to wait, or simple neglect.
First establish what the change does. Smarter components can perform tasks that once required central coordination, allowing specialists to displace integrated producers. Rising development costs can push the other way, favoring firms with enough sales to cover them. The relevant costs and coordination needs determine which structure has an advantage. Such openings arrive unevenly; a company cannot schedule them to satisfy its growth plan. Understanding the economics can guide preparation without settling when a transition will happen or who will win.
Different reasons for standing still
Cultural inertia concerns what people consider worthwhile work. A team admired for research may lack the norms needed for fast product engineering. A new market plan cannot assume that capability into existence. Rumelt proposes simplifying activities so work and responsibility become visible, separating units where they can function independently, and deciding what to close or repair. A financially strong unit can still spread damaging norms, so those decisions must consider behavior as well as profit. Changing influential group members or leaders can establish new work habits before coordination is rebuilt. Fragmentation would be harmful where activities need to work closely together.
An apparently inert competitor may have a different reason to wait. Customers who switch slowly can keep its old business profitable. Offering everyone a new deal might sacrifice more income than gradual departures cost. Rumelt calls this inertia by proxy. The restraint reflects customers' habits. When the balance changes, the incumbent may respond quickly. A rival should not mistake the delay for lasting incapacity.
Entropy is different again. Focus, order, and efficiency can decay even without an external shock. Profitable old resources can hide multiplying initiatives and weakening coordination. Maintaining useful boundaries is therefore continuing work. Those resources can also finance the next capability, even if that capability threatens established jobs or income. The task is to preserve what still contributes while confronting what now obstructs progress.
Rebuilding a working system
Nvidia's historical graphics-chip strategy shows how a response becomes a working system. After an unsuccessful multimedia product, the company concentrated on three-dimensional graphics for desktop computers. It used the established triangle-based approach to graphics and hired others to manufacture its chips. The graphics pipeline broke a scene into triangles, processed them, and assembled the image. Moving more of those stages onto one chip could improve performance. As chips could hold more transistors, Nvidia could also process more work simultaneously. Manufacturing advances could therefore buy better graphics instead of merely cheaper chips. Gamers playing against other people had a reason to value those gains. Delayed or unclear graphics could put them at a disadvantage.
To deliver substantial improvements every six months, Nvidia used three overlapping development teams. Each project still took eighteen months from start to market. Staggering the work increased the share of time Nvidia could offer the higher-performing product against slower rivals. A two-month delay could consume much of the advantage, so meeting that rhythm required changing more than deadlines.
Mistakes discovered after chips were manufactured could require another costly production run. Nvidia invested in checking designs earlier. Simulation represented a chip in software; emulation configured special chips to behave like the planned product. Emulation was faster than simulation, though slower than the final chip. It also let programmers begin the driver software before the actual chips arrived. Drivers let other software communicate with the graphics hardware, so this work reduced another potential delay.
Nvidia took control of drivers previously written by the makers of the boards carrying its chips. A common driver could detect a chip's capabilities and adjust its commands, reducing compatibility problems. Commercial relationships had to support that technical change. Board makers could resist losing margins, so Nvidia worked with Dell and contract manufacturers to bypass the obstruction. The powerful buyer helped bring the improved offering to market. The release policy depended on engineering, software, and commercial arrangements working together. An early product's moderate success financed further development; the full result took successive releases.
Some rivals were constrained by scattered efforts or inflexible processes. Those barriers could change. One rival acquired engineering talent and matched Nvidia's release pace. By two thousand nine, the graphics pipeline was fully implemented on the chip and many gamers felt less urgency about another chip. The earlier technical opportunity was narrowing. Rumelt's twenty eleven account describes moves toward general computation and systems for smaller devices, with both directions still uncertain.
A well-maintained system can still need a new strategy when its technical and customer foundations change. Knowing what made it work helps an organization recognize what it must rebuild next.
Three things to keep
Diagnose the changed economics and the particular obstacle before choosing what to rebuild.
Slow responses can reflect profitable delay; neglected coordination can decay without any external shock.
An effective system needs renewal when its technical opportunity or customers' priorities change.