This article originally appeared on HBR.org.

Given the pace at which digital innovation is disrupting industries globally, it’s not surprising that most CEOs feel pressure to find and deploy the right technology as fast as their budgets will allow. Many are discovering, however, that becoming a digital leader isn’t simply a matter of technological savvy. It’s about creating an agile organization that can detect what type of change is essential and respond quickly with the most competitive solution.

In our experience, most companies are already steeped in technology and learning fast about how it can transform their businesses. Typically, teams in the field are well aware of the digital threats and opportunities within their area of the organization—usually more so than the corporate center. They have launched their own apps, deployed robotics, established partnerships with digital players, or are using data to analyze their business and make better decisions.

The problem is that that these efforts tend to be ad-hoc and uncoordinated. Without the proper framing and orchestration at the overall company level, the best initiatives will fail to get the attention and investment they need. While it is important to encourage local ownership of ideas and projects, turning them into game-changers requires clear, sometimes ruthless direction from the center around which projects to scale and in what order. Only the CEO has the power to provide this kind of direction across the entire enterprise.

To do that effectively, CEOs need a holistic view of the digital threats and opportunities facing key parts of the business, and a way to link them to an overall vision for how digital is reshaping the competitive landscape. This brings order to the chaos of initiatives and provides a clearer basis for narrowing down priorities and managing the cross-functional interdependencies that the best digital solutions often present. Three ways to manage the digital transition are:

Define where change is needed most: Digital technology affects every company differently, but it tends to create or destroy value in four critical areas of the organization: customer engagement, digital products and services, operational performance, and preparing for disruptive new business models. Developing a clear point of view on the opportunities or threats in each area will suggest which capabilities need the most attention and where to concentrate investment.

Consider how General Electric arrived at the decision to develop and launch its Predix cloud-based industrial operation system. The initiative began when GE’s CEO encouraged his organization to explore how the accelerating trend toward value-added services in the industrial sector might eventually affect the company’s growth. In essence, he challenged his team to act as a change leader—to disrupt before being disrupted—by interpreting the weak signals coming from the market. He asked them to pay special attention to how digital native companies were creating shifts in customer behaviors and to diagnose digital solutions for business challenges that had not yet taken a toll on the company’s profit and loss statement.

Launched in August 2015, Predix helps companies see how their machines and infrastructure are performing so that they can improve them constantly. GE has treated the platform as open source, reasoning that it will power the growth of the industrial Internet, which will in turn accrue major benefits to GE. Its early success also highlights the CEO’s critical role in challenging the organization to assess its digital competence and to determine how urgently it needs to respond to threats and opportunities.

Choreograph the change: Even the clearest digital strategy will fail if your people are unprepared to embrace it. As critical as defining where you need change is setting up the capabilities and processes that will enable it. IT, for instance, is very often the tightest digital choke point because it is mired in old processes and needs significant reshaping to link it more closely to strategy, while creating a more agile approach to development. It is also essential to develop key capabilities in data analytics to make better decisions using the flood of new information flowing through the organization.

Ensuring that change sticks involves the hard work of defining new roles, adding new skills and adopting new ways of working. And it is important to carefully choreograph the change, defining who will lead the effort and how it will be sequenced.

Mobilizing for this kind of change inevitably means shaking up the status quo and leaders themselves need to be prepared to manage the company differently. Consider the challenge companies face in the rapidly changing market for power train compressors. Increasingly, competing in this market means equipping compressors with hundreds of sensors that send information back to the manufacturer about power consumption, vibration, wear and output. Manufacturers then analyze the data remotely to predict problems their customers might face and offer solutions proactively. Blending the hardware with digitally enabled services creates measurable customer value. But taking full advantage of it requires significant cultural changes. Product development teams have to work with field maintenance and commercial teams. Data management teams have to develop the predictive algorithms to improve the customer experience in coordination with the customer-facing teams. Marketing, commercial and finance have to work together to develop new pricing models. All of this has to happen fluidly and rapidly. The old system of passing possible solutions across silos, wading through validation loops and meeting threshold tests simply isn’t fast enough.

Empower people: One clear implication of this approach is the central importance of an orchestration model for digital—prototyping, risk-taking, and mobilizing the frontline to push concrete initiatives. Many of the leading digital models to date have been distributed globally throughout an organization via “digital relays” or champions within each geography and business unit. They are centrally orchestrated at a cadence that improves uptake and so the design remains consistent where appropriate.

This “project team”-based approach relies on empowering people at every level of the organization to work together to devise and implement solutions. Again, that requires some critical organizational and cultural changes. Everybody, for instance, needs access to customer data and the analytics and visualization tools used to interpret it—information that is typically hoarded in a particular part of the organization. This “democracy of data” frequently puts pressure on the middle managers in charge of it and passes decision rights to many others.

Only the CEO can manage this process by breaking down the appropriate boundaries, giving teams permission to set new rules, and providing the strategic framework to buttress the new order. It often makes sense for the CEO to delegate to an “orchestrator,” in the form of a Chief Digital Officer. But that person needs to be fully empowered to compel change across the organization in the name of the CEO. There isn’t time for anything else.

Laurent-Pierre Baculard is a partner with Bain & Company in Paris and leads Bain’s Digital practice in EMEA.