All companies claim that their strategies are customer driven. But the term “customer” is among the most elastic in management theory. A working definition might be that your customers are the people or entities that buy your products and services and supply your revenue. That includes any number of actors in a company’s value chain: consumers, whole­salers, retailers, purchasing departments, and so forth. Some companies go as far as to label internal units as customers: Manufacturing is a customer of R&D, for instance, and both are customers of HR.

Other definitions don’t even require that a customer supply revenue. Pharma giant Merck’s most important customers are not the patients who use its drugs or the physicians who prescribe them. Instead, Merck has chosen research scientists in labs and universities around the world as its primary customer. Accordingly, its business model relies on encouraging its own world-class researchers to act like university scientists by conducting basic research, publishing papers, and presenting results at conferences, all with the intent of discovering groundbreaking compounds that can then be commercialized by Merck’s marketing and sales group. The business is even configured like a research university—a simple functional structure in which a powerful, centralized R&D unit receives the bulk of organizational resources.

Unsurprisingly, perhaps, many executives are reluctant to define their customers as narrowly as Merck has. By not singling out any group as the primary customer, executives can sidestep difficult choices that might turn out badly—a temptation that’s particularly strong in new, rapidly evolving markets. What’s more, many business leaders believe that treating all value chain partners as customers improves internal coordination and responsiveness.

But by not identifying one primary customer, companies that consider themselves “customer focused” soon become anything but. Consider the contrasting fortunes of Yahoo and Google. Yahoo began as a broad-based internet portal supported by proprietary editorial content. To attract users, it hired journalists to write entertainment stories and created utilities such as Yahoo finance, Yahoo movies, and Yahoo sports. Over time, Yahoo executives began to spread resources among many additional initiatives, including social networks, products, media, and advertising. As a result, they under­invested in search, and the website became messy and confusing.

Then Google entered the field. From the outset, Google focused on users who appreciated technology and its ability to unlock new opportunities and applications. Like Merck, Google allocated the lion’s share of its resources (and prestige) to its technologists and engineers, who were given freedom to innovate. The aim throughout the business was to build the best technology in the world—whether in search, Android, or maps. With that sharply focused value proposition and business model, Google quickly leapfrogged Yahoo in the competitive marketplace.

The bottom line is this: The strategic choice of primary customer—with special emphasis on “primary”—defines the business. This is certainly true at Amazon, which serves four very different types of customer: consumers, sellers, enterprises, and content providers. You might think that it considers all four customer groups to be equally important. But the company’s choice of primary customer is reflected clearly in its well-known mission “to be the world’s most consumer-centric company.” Amazon devotes maximum resources to pleasing consumers, even if that means sellers or content providers sometimes feel shortchanged (sellers whose storefronts are hosted on the Amazon platform have been known to sue Amazon for more resources). This unwavering focus on consumers has created innovations such as prime free shipping, detailed product reviews (including negative ones), look-inside-this-book, and the listing of lower-priced products from off-site competitors. These practices have often been criticized as inherently unprofitable or injurious to Amazon’s other constituents. But the main results of the company’s choice are the ones that count most: unparalleled customer loyalty and stratospheric stock valuations.

In the following pages I’ll present a truly customer-driven framework that can help executives build winning business models for their companies. The framework lays out four steps: identifying the best primary customer for your business, creating processes to learn what that customer values, allocating resources accordingly, and building an interactive control process to monitor the assumptions that underlie your choice.

Step 1: Identify Your Primary Customer

As the cases of Merck, Google, and Amazon illustrate, your most important customers are not those that generate the most revenue but those that can unlock the most value in your business. For some businesses, the primary customer will be the end user or consumer of the product or service. For others, an intermediary (such as a reseller or a broker) will be the critical customer to which organizational resources should be devoted.

But how can executives be confident that they’re making the right choice? Identifying the best primary customer for your firm involves assessing each group of customers along three dimensions: perspective, capabilities, and profit potential. Let’s look briefly at each.

Perspective refers to the culture, mission, and folklore of a business, often revealed in stories about important events or people in the company’s history. It is the lens through which executives consider opportunities and strategic direction. Steve Jobs’s obsession with perfection in product design created a legacy that frames the opportunities Apple managers will (and will not) consider. Walmart’s Sam Walton was famously frugal in his own life. And Amazon founder Jeff Bezos is a zealot about delivering a superior experience to shoppers. “When [executives of other companies] are in the shower in the morning, they’re thinking about how they’re going to get ahead of one of their top competitors,” he told Fortune in 2012. “Here in the shower, we’re thinking about how we are going to invent something on behalf of a customer.” Clearly, the choice of primary customer must reflect a company’s perspective; otherwise the company will be unable to leverage the energy and creativity of its people in service to the customer.

Capabilities refers to the embedded resources of the firm. Some firms excel at technology (Apple, Google, Airbus), some at logistics (Walmart, Amazon, Dell). Others provide superior brand marketing (Ralph Lauren, Nestlé, P&G) or have industry-­specific capabilities (original content production at HBO and Netflix, mining at BPH Billiton). Such capabilities, which are built up over time and are often difficult to copy, position a business to serve the needs of certain customers better than others. Dell in its early years built a formidable low-cost logistics operation to support its direct-to-consumer sales model. Today, the company is attempting to change its primary customer by refocusing on CIOs of large enterprises. This pivot has proved difficult for Dell because CIOs look for a set of capabilities—integrated hardware, software, and services solutions—very different from what end consumers need.

Profit potential refers to a customer’s ability to deliver profits. Techniques such as Michael Porter’s five forces analysis can provide insight into the relative profitability of various customer types—and help weed out those that would be a poor choice for primary customer. Consider HBO. Cable operators that purchase HBO’s content might seem to be the obvious choice. But cable operators have low switching costs—they can easily buy content from a variety of producers. Thus HBO would have little market power and would be unable to extract high margins from cable operators. But by targeting filmmakers as the primary customer and devoting significant resources to their needs, HBO can create the unique products that viewers demand, allowing it to charge premium prices that cable operators cannot negotiate. Of course, profit potential isn’t always about customers who can pay premium prices; becoming the preferred destination for cost-conscious customers can deliver substantial profits through volume, as Walmart has demonstrated.

LinkedIn is one successful company whose primary customer clearly fits all three dimensions. For more on how it settled on individuals (rather than job recruiters or advertisers), see the exhibit “How LinkedIn Chose Its Primary Customer.”

How LinkedIn Chose Its Primary Customer

Identifying your primary customer involves assessing each group of customers along three dimensions: perspective (your company’s culture, mission, and folklore), capabilities (distinctive embedded resources), and profit potential (the customer’s ability to deliver profits). The group that best satisfies all three conditions should be the primary customer.

Consider LinkedIn. In deciding where to focus, executives had to choose among three kinds of customers that all had the potential to generate substantial revenue: job recruiters, advertisers, and individual members.

Step 2: Understand What Your Primary Customer Values

Once you’ve determined who your primary customer is, the next step is to identify which product and service attributes the customer values. Within the same market and industry, different primary customers may value different things: Some demand the lowest possible price, others want a dedicated service relationship, and still others are looking for the best technology or brand or other specific attribute. To complicate matters, customers often don’t know exactly what it is they value. Uncovering the full truth about their needs requires systematic research at multiple levels.

Let’s take the easy part first. Assume you have already chosen the best primary customer and have a good working idea of what the customer wants. There’s still plenty of room for improvement. You can refine your understanding by leveraging today’s easy and cheap access to data on customer buying habits, preferences, and search activities. Data analytics is an important tool in uncovering and rapidly responding to changing customer needs. At Google, separate analytics teams for display, search, and maps spend untold hours in their labs with customers studying eye movement and other variables to gauge their reactions to subtle product modifications such as changes in color. Nestlé has a war room where analysts monitor social media to track chatter that relates to or affects the acceptance of its products. The analysts use the intelligence to inform product research and marketing decisions and to evaluate in real time how well their value propositions are meeting the needs of the primary customer.

Such data can help you fine-tune a product or a website’s functionality to better meet your customer’s known needs. They’re unlikely, though, to help you identify what your customers want but aren’t getting. For that, you need to actually ask them. Smart companies set up systematic dialogues with their primary customers. Managers at FedEx, for example, hold twice-yearly summits where they bring in a sampling of business customers (the firm’s primary customer) to ask them where FedEx is doing a good job of meeting their needs and where competitors are doing better. At Germany’s Henkel, the world’s leader in adhesives, CEO Kasper Rorsted has created a “tops to tops” program in which all executives are required to meet regularly with their counterparts at major customers to ensure that their needs are understood and the company is responding appropriately. Other companies, especially those with rapid product cycles, manage the dialogue through new-product testing. Google’s Gmail, for example, was released after five years of beta testing by more than 1,000 technology opinion leaders.

Finally, you should set up processes for identifying products or services that customers may not know they need. This can be challenging—and expensive. Smart companies typically rely on ethnographic methods. At P&G, for instance, where consumers are the primary customer, executives ask their managers and market researchers to spend days at a time accompanying consumers on shopping trips and sitting at the family dinner table to more fully understand the extent to which various products meet consumer needs. CEO A.G. Lafley recounts in his book The Game Changer how the experiences of P&G executives living with lower-middle-class families in Mexico City produced Downy Single Rinse, a fabric softener that is simpler to use for markets where water is in short supply.

Most companies assume that their products and services meet the needs of their customers. But surprisingly few actually test this assumption. So ask yourself, What are the processes we use to make sure that we truly understand what our customers value and that we can deliver value better than our competitors do?

Step 3: Allocate Resources to Win

As we saw with Merck and Amazon, your choice of primary customer and your understanding of what the customer values provide all the information you need to make the critically important decision of how to organize your company’s resources—in other words, what kind of business model to adopt. There are five basic configurations you can choose from.

Low price.

If your primary customer is looking for the lowest possible price, centralized operating functions (such as merchandising and distribution) should receive the bulk of organizational resources, in order to create economies of scale and scope. Customer-facing units, such as stores or restaurants, should receive relatively few resources. This is the configuration used by Walmart.

Local value creation.

If your customer values products and services that are customized to local tastes, preferences, and regulations, you should organize like Nestlé. It pushes resources out to regions so that local managers can customize product offerings, while operating core functions are limited to corporate-level support activities.

Global standard of excellence.

If your customers are looking for the best possible technology or brand no matter where they are located, you should organize resources around global business units that are defined by product lines. This configuration allows focus and leverage in R&D, brand marketing, and distribution. Microsoft, for example, has separate business units for Windows, servers, MSN, mobile, and Xbox. Each unit has full revenue and profit responsibility and its own R&D. (Note: Microsoft has recently announced that it intends to change its structure to more of an expert knowledge organization—described below—to emulate Google.)

Dedicated service relationship.

If your customer is looking for an ongoing, deeply embedded service relationship, you should organize like IBM. Customer teams in industry-based “verticals” marshal and coordinate product and service delivery from centralized, product-based “horizontal” units.

Expert knowledge.

Finally, if your primary customer is looking for expert technical knowledge, you should follow the example of Google and Merck, where R&D sits prominently on top of product organizations that receive the lion’s share of the company’s attention and resources, with other functions playing a supporting role. These R&D-led product units, which may be distributed in centers around the world, have no revenue responsibility: They are focused entirely on product development and on creating breakthrough technology. All sales revenue is routed through a centralized, stand-alone sales division that is configured as a distinct function.

Of course, various permutations and combinations of these five basic configurations are possible. Many companies will want to leverage the advantages of several models at once. Some companies experiment with matrix structures that can simultaneously emphasize, say, geography and function or business unit and region. This “split the difference” approach can be appealing if, for example, you are an engineering company like ABB and your primary customer is government purchasers that demand both the best technical features (global standard of excellence) and customized content (local value creation). But it should be noted that matrix organizations are notoriously difficult to manage; all too often, a matrix structure reflects an inherent confusion about who the primary customer is rather than an effective response to the customer’s needs and preferences.

As a general proposition, when a business finds that it has more than one primary customer, it should be split into separate units and adopt for each the configuration that best allows it to focus resources on the needs of its primary customer (“the rule of one”). At Nestlé, for example, although most of the business is structured using a local value configuration, the company’s strategy differs for two of its brands: Nespresso and Mövenpick. Customers want a consistent, premium experience from those brands regardless of location. Accordingly, those businesses are managed using a global standard of excellence configuration in which resources are centralized and managed globally.

In reviewing a business model, the key question executives should ask is this: Do the choices we have made about the company’s structure reflect our choice of primary customer? If the answer is no, competitors whose business models are consistent with their chosen primary customer will almost certainly be outplaying you.

Step 4: Make the Control Process Interactive

As good as your business model may be today, it cannot and will not survive forever. Customer tastes will change, new technologies will replace old, unforeseen competitors will enter the market, and regulations and population demographics will evolve over time. That means you must constantly gather information on shifts in your competitive environment, especially those that might affect the behavior of your primary customer. You must be alert to emerging threats and opportunities that will redefine what your customer values and that customer’s profit potential. If the changes are dramatic, you may need to fundamentally reorient your business model—and even, in the most radical situations, select a different primary customer.

The best way to get the information you need is to make sure that your company’s control systems are interactive. Everyone in the organization should be using the same performance measures as the basis for learning and debate. Monitoring changes in customer behavior and the competitive environment, in particular, is not a function to be delegated to a special department. As a technology executive recently told me, “Companies that get it wrong are those that build departments with ‘innovation’ in their titles. We need to have everyone in the business innovating.”

Depending on your business strategy and industry, you can choose to use any of your current management systems interactively—your profit planning system, your brand revenue system, your orders-on-book or new deal system. At HBO, for example, executives constantly track the company’s success rate in bidding for new shows from filmmakers and use that measure to prompt a discussion among managers throughout the business about changes in the competitive marketplace that could affect their strategy. Amazon’s category managers use their Monday morning meetings as a forum to study data about product assortment choices, revenue growth, customer orders, and inventory turnover. Reflecting the firm’s leadership principles (customer obsession; bias for action; earn trust of others; dive deep; and have backbone, disagree, and commit), these meetings are highly interactive as managers from a diverse array of functions work together to interpret the data and come up with action plans. Some of these actions may, over time, plant the seeds of a new strategy.

Companies that hedge their bets usually find themselves looking at the taillights of their more decisive competitors.

Systems that work well interactively—like those at HBO and Amazon—share three essential characteristics: They deliver information about uncertainties that could undermine the assumptions of a current strategy and require attention from the highest levels of management; they are widely used in the organization, receiving frequent and regular attention from operating managers at all levels; and they involve face-to-face meetings that focus on emerging data, assumptions, and action plans. There is no substitute for the energy and creativity that flow from open debate when participants leave their titles at the door.

In using interactive control processes, managers should continually ask three questions: What has changed? Why? and, most important, What are we going to do about it? If you identify changes in your customers’ profit potential, for instance, you might want to rethink your choice of your primary customer. Changes in tastes, regulations, technology, or competition may alter what it is that your primary customer values—resulting in a need to reallocate resources or redesign your business structure. If you have significant first-mover advantage thanks to a new technology—or if competitors are evolving and struggling to find their way—you may be able to duck making a choice of primary customer, opting instead to stay fluid and focus on experimentation. But the entrepreneurial landscape is littered with the carcasses of companies that tried to be everything to everyone. Like Yahoo, they muddled along until they were overtaken by crisis, often bringing in a new leader in a last-ditch effort to impose discipline and focus on a failing business. It is, I believe, ultimately less risky to be proactive and make the key strategic bet of choosing a primary customer. Companies that hedge their bets usually find themselves looking at the taillights of their more decisive and committed competitors.

A version of this article appeared in the March 2014 issue of Harvard Business Review.

Robert Simons is the Charles M. Williams Professor of Business Administration at Harvard Business School. He is the author of Seven Strategy Questions: A Simple Approach for Better Execution (Harvard Business Review Press, 2010).