For a long time, the Silicon Valley funding model has been hailed as a powerful alternative to the stifling way corporate America works. Many are betting on the new generation of technology firms to unsettle the old guard. Today the number of unicorns—startups that valued at $1 billion or greater—is staggering. Fortune counted more than 170 of the mythical creatures, with an average of one unicorn born every week during 2015. Back in 2009, there were just four companies that fit the bill.

Despite this surge, the reality is that disruption doesn’t always happen as quickly as people assume. More than 40% of the unicorns that went public since 2011 saw their valuation stay flat or dropped. Some observers have compared the situation to the dot-com bubble of the late ‘90s.

Meanwhile, tech giants like Amazon, Google, and Facebook have continued to grow impressively, especially considering their large size. What is the secret that allows these incumbents to fend off the startups aiming to displace them?

The answer is deceptively simple: embracing self-cannibalization. Self-cannibalization occurs when a company chooses to proactively replace one product or process with another that is potentially worth less. Forward-looking incumbents recognize the need to cannibalize their own products, rather than leaving it to other startups, who are more than happy to take on the challenge.

Embracing this approach isn’t easy – it doesn’t always seem natural to talk about how to replace profitable businesses. But there are four rules which can help managers of all walks of an organization instil the principle in their day-to-day work, in order to make self-cannibalization successful in the long run.

Rule #1: Get into the habit of setting up new business units that compete with the old
Consider China’s Tencent. Founded in November 1998, the company has since grown into China’s largest and most-used Internet service portal, and in the process, has also become the world’s largest gaming company.

Back when desktop computers still reigned supreme, Tencent dominated online instant messaging with its QQ service (similar to ICQ). With the rise of the smartphone and mobile technology, QQ failed to capture new users. Top management swiftly established a new team in Guangdong, away from the Shenzhen headquarters, and tasked a small group of engineers to reimagine a different social media platform, giving them carte blanche to cannibalize the existing QQ product. That was the beginning of WeChat.

Since then, WeChat has been constantly launching new services, from mobile payments to booking doctor appointments, from police reporting to taxi hailing, from video conferencing to mobile banking services. WeChat now has more than 690 million active users and is still growing. In August, WeChat started rolling out an impressive international series of new games—another move aimed at disrupting Tencent’s core business.

The willingness to cannibalize a company’s existing business before its decline was also a major focus of Apple under Steve Jobs. In 2005, when the demand for the iPod Mini remained huge, the Nano was launched, effectively destroying the revenue stream of an existing product. And while iPod sales were still going through the roof, Jobs launched the iPhone which combined iPod, cell phone, and Internet access into a single device. Three years after the iPhone’s launch, the iPad made its debut despite the risk that it might one day cut into Mac desktop computer sales. So resolute was Apple’s determination in trading a highly profitable business for an unknown future that Jobs reportedly said, “If you don’t cannibalize yourself, someone else will.”

Rule #2: Find a balance between derivative products, platform upgrades, and breakthrough innovation
Self-cannibalization occurs at different levels. It can mean both replacing existing products and platforms with incrementally better ones and replacing them with something completely different. Most companies find the latter more challenging, but the best companies pursue both.

Recruit Holdings from Japan is a publishing and classified advertisement company, founded in 1963. Its classified advertising has evolved into numerous publications in which merchants, ranging from gourmet restaurants, beauty salons, and wedding banquet venues advertise, and Recruit editorial teams provide magazine content. (Similar to Cosmo, Glamour, and GQ.) Since the early 2000s, many of the company’s publications have gone digital.

Instead of merely digitizing magazine content, managers built a number of web-based platforms. Operating managers were empowered to develop unique business models and launch derivative products by themselves to best serve each unique segment; and in the process, disrupted and competed with one another.

At the corporate level, however, it became all too apparent that without a commonly shared platform upgrade, the collection of individual businesses would have limited potential. Senior management at Recruit began to push for a unified backbone platform, a platform upgrade that would better serve customers’ needs. Doing so did require getting rid of many overlapping offerings, cut-short a few cash-cow businesses, but the end result was a more cohesive overall product portfolio.

Although launching derivative products and upgrading platforms are both important, Recruit didn’t let them serve as substitutes for more ambitious innovation. Lately, the company realized the importance of machine learning and artificial intelligence (AI), and proceeded to set up a mini AI lab, dedicated to breakthrough innovation. In 2015, Alon Halevy, a former Google research scientist, joined the company to help guide this effort. Top management saw AI as having the potential of redefining Recruit’s corporate mission—the ultimate expression of self-cannibalization.

Rule #3: Create a bypass mechanism to pitch ideas to the top
One challenge for companies looking to self-cannibalize is the fact that game-changing ideas can easily be filtered out as business proposals move up the corporate ladder. To counter-balance this tendency, Recruit hosts several idea pitch days when operating managers can showcase unconventional proposals to senior executives directly, bypassing the management hierarchy entirely.

A persistent manager at Recruit had repeatedly lobbied for an online education business, which his supervisors couldn’t reconcile with the existing business model. But when shown to the board members at one pitch night, the business case was so compelling that it resulted in seed money and the establishment of an independent unit. The new unit has since focused on online “cram school” delivery, providing less affluent students a way to compete for university admission.

Rule #4: Create a corporate goal with a percent of revenue earmarked to new products
Edward Deming, the father of the quality movement in the 1950s, believed that what can’t be measured doesn’t get managed. The same holds true for growth. 3M is famous for employing the thirty percent rule, with 30% of each division’s revenue coming from products introduced in the last four years. At companies like Tencent and Recruit, similar metrics are tracked rigorously and employee bonuses are based on the successful achievement of this goal.

Adding such a metric makes it easier to follow Rule 1, since a culture that encourages new products will be more likely to consider cannibalizing its own successes. But even with such a metric, the balance between derivative products and breakthrough innovation remains important.

The fundamental advantage of large companies is in their ability to integrate and reconfigure offerings and services based on their prior capabilities. Startups may move fast, but they lack experience. Big companies, by contrast, possess a wealth of knowledge and know-how. What they lack is the bandwidth to commercialize the next growth engine when time is still on their side. But the two don’t need to be exclusive. By embracing self-cannibalization as a principle and following these few simple rules, an incumbent can fend off the unicorns at its gates.


Howard Yu is professor of strategic management and innovation at IMD.

Thomas Malnight is professor of strategy and general management at IMD.

This article was originally published on HBR.org.