When should we abandon the product that made us?

The question behind the question

Abandoning a core product is the hardest decision in this collection, because the product is usually also the company’s identity, the source of its leaders’ careers and the work its best people are proudest of. The economics may be clear long before the decision is made. What delays it is psychology: loss aversion, described in Loss Aversion and the Pull of the Status Quo, which makes giving up a known business feel worse than an equal gain feels good, and the pull of sunk costs, described in Sunk Costs and the Escalation of Commitment, the urge to justify what has already been spent.

Andy Grove called the moment when a business’s fundamentals change a strategic inflection point. In Only the Paranoid Survive (1996) he described how Intel left the memory chip business in 1985, after he asked Gordon Moore what a new chief executive would do if the board replaced them. Moore answered that a newcomer would get Intel out of memories. The question works because it strips away identity and leaves the economics.

When leaving early worked

Microsoft shifted its strategic priority from the OS/2 partnership with IBM to Windows after Windows 3.0 succeeded in 1990, abandoning a joint project into which it had put years of work because its own product had momentum. Apple discontinued the Newton handheld in 1998 to concentrate its shrunken resources on the Macintosh; the decision freed engineers and money for products that later succeeded. Intel adopted AMD’s 64-bit extension of x86 for its mainstream processors in 2004, conceding that its own Itanium architecture had lost, though it kept Itanium for high-end servers for years, which the record judges mixed. Satya Nadella’s 2014 memo redefined Microsoft as a company for a world of mobile devices and cloud computing, moving its centre away from Windows, the product that had made it.

Robert Burgelman’s study of Intel’s exit from memory found that the decision had been prepared from below. Middle managers allocating scarce manufacturing capacity by profit per wafer had been shifting it away from memory for years before the leaders made the formal decision. The organisation’s own rules saw the change first, an example of what the site calls The Knowledge No One Holds, which leaders can learn from if they look.

When holding on failed

The makers of Lisp machines, specialised workstations for artificial intelligence research, kept investing in proprietary hardware through the 1980s while general-purpose workstations improved rapidly and soon ran Lisp well enough; the companies collapsed. France Télécom and the French state continued to rely on and protect Minitel, the national videotex service, as the internet spread elsewhere in the mid-1990s; Minitel’s success had made it a reason to delay, and France came late to the internet as a result. Both failures share a shape: the old product was still working, still loved by its users and still central to the organisation’s identity, and the alternative looked inferior by the old product’s standards.

A diagnosis to start from

The diagnosis rests on three questions. Is the market’s centre of gravity moving, and is the evidence coming from customers and the front line rather than from strategy decks? Is there an alternative, inside or outside the company, that already has momentum? And how long can the old business fund the move?

The guiding policy that the successes share is to leave while the old business is still strong enough to pay for the new one, and to redirect people rather than merely cut. The coherent actions include ending investment in the old product’s next generation, moving its best people to the new effort, being honest with customers about the timeline, and treating those who lose most with care, because how an exit is handled shapes how the next one will be received.

The strongest counter-case

The strongest objection is that many companies have abandoned their core too early, chasing a fashionable alternative that never arrived. Netflix’s 2011 attempt to split off its DVD business as Qwikster, which the record judges to have backfired, moved faster than customers wanted. And some companies that held on to an old product did well for decades by serving a loyal niche. The question “what would a new chief executive do?” can also be answered badly by a new chief executive who misreads the market.

The counter-case argues for evidence over fashion. The successful exits in the record moved to alternatives that were already working, such as Windows after version 3.0 and x86-64 after AMD shipped it, not to visions.

What the wider research says

Clayton Christensen’s The Innovator’s Dilemma (1997) explained why well-managed companies fail to leave in time: their processes for allocating resources, which serve existing customers well, systematically starve new businesses that begin small and unprofitable. Barry Staw’s research on escalation of commitment showed that people who feel responsible for a failing course of action tend to invest more in it. Together they suggest that the decision to abandon a core product often needs a deliberate break in normal processes: a separate unit, a new leader, or a question that forces the economics into view.