A company can answer every customer question faster and still give customers more reasons to leave. It can automate its service desk, improve its app, and launch new features while overlooking the problem customers wanted solved in the first place.
That is why Theodore Levitt’s Marketing Myopia remains relevant more than six decades after its original publication in 1960. His central argument challenges leaders to define their business around the customer’s need rather than the product they currently sell. Today, that distinction should guide decisions about artificial intelligence, investment, and growth.
My argument is straightforward: leaders should use new technology to improve the outcomes customers value, and build their strategy around delivering those outcomes profitably. Making the existing business more efficient matters. Understanding whether customers still need it in its current form matters more.
Levitt’s argument has three immediate implications:
- Customer needs reveal competitors that conventional industry definitions overlook.
- Technology creates value when it improves a meaningful customer outcome.
- Customer understanding changes strategy only when leaders change investment decisions and incentives.
Together, these ideas turn a classic marketing article into a practical test of corporate strategy.
Customers define the competitive market
Levitt’s most familiar example concerned railroads. He argued that executives defined their business around rail transportation rather than the broader need to move people and goods. That narrow definition obscured opportunities as customers gained other ways to travel and ship freight.
The enduring lesson concerns how leaders draw the boundaries of competition. An executive who studies only companies selling similar products can miss an alternative that solves the customer’s problem differently.
Consider a hypothetical professional training company. Its leaders might compare course prices, instructor credentials, and completion rates with those of rival providers. Customers, however, may be comparing a course with an AI tutor, advice from a colleague, or software that eliminates the task they needed to learn. The competitive question extends beyond who teaches the best course. It includes whether taking a course remains the best way to become capable.
The same reasoning applies to customer research. Asking buyers which features they want tends to produce improvements within an existing product category. Investigating what they are trying to accomplish can expose problems the category handles poorly.
That difference changes investment decisions. A training provider might discover that customers struggle less with understanding concepts than with applying them at work. More course content would address the wrong constraint. Practice, feedback, or support during an actual assignment might create greater value.
A broader definition still requires discipline. Calling a training company a “human potential business” offers little guidance about where to compete. A useful definition identifies a particular customer, a consequential problem, and an outcome the company can credibly deliver.
Levitt’s insight becomes actionable when a broader understanding of demand leads to sharper choices.
AI makes the distinction more consequential
AI gives leaders another opportunity to confuse technical progress with customer progress.
A service operation can generate responses faster without resolving difficult problems. A marketing team can produce more messages without making its offer more relevant. A software company can release an AI feature without establishing whether customers would pay for the result.
These improvements may have operational value. Their strategic value depends on what changes for the customer.
Evidence shows that meaningful gains are possible. In research published in the Quarterly Journal of Economics in May 2025, Erik Brynjolfsson, Danielle Li, and Lindsey Raymond examined an AI assistant introduced among 5,172 customer support agents. Access increased productivity, measured as issues resolved per hour, by 15% on average. Effects varied across workers. The result comes from one company and setting, rather than a universal estimate of AI’s benefits. Source: Generative AI at Work.
The choice of measure matters. Issues resolved per hour connects productivity to the purpose of customer support more closely than messages generated or tools deployed. Even that measure needs companions: whether the resolution was correct, whether customers had to return, and what the service ultimately cost.
Commercial models also reflect this shift. Intercom currently lists pricing for its Fin AI Agent at $0.99 per outcome, with other charges depending on the arrangement. That approach connects part of the bill to a defined result. It also makes the definition of an “outcome” commercially important: the supplier and customer need to understand what qualifies and how performance is verified. Source: Intercom pricing.
These examples point toward a contemporary reading of Levitt. The strategic question begins with the result customers need. Technology selection follows.
The most dangerous myopia can look like success
The harder case is a company whose dashboard looks excellent while its relevance weakens.
Imagine a hypothetical service provider that reduces handling time, increases automated responses, and lowers cost per interaction. Management celebrates. Yet customers still contact the company repeatedly because their underlying problems remain unresolved.
Every internal measure can improve while the customer’s total effort rises.
This is where Levitt’s argument reaches beyond marketing. Companies allocate money, reward managers, and organize work around established products and functions. Those arrangements can make an existing business model appear indispensable because so many internal targets depend on preserving it.
A customer may benefit from fewer interactions, fewer software licences, or less need for expert assistance. The supplier may earn revenue from those very activities. Recognizing the customer’s need therefore raises a difficult question: will management support an improvement that reduces demand for part of today’s offer?
Customer orientation becomes strategically meaningful at that point. It requires leaders to examine how the company will earn attractive returns as the way it creates value changes.
Levitt’s argument also needs a boundary. Customer focus cannot overcome every disadvantage. Capital requirements, regulation, capabilities, and competitive economics still constrain what a company can do. Nor does identifying a broader need justify entering every adjacent market. The value of the article lies in improving the questions leaders ask, rather than providing a complete explanation for every industry’s success or decline.
Three decisions leaders should make now
1. Define one customer outcome precisely enough to measure.
Within the next month, select an important customer group and describe the problem it pays the company to solve. Examine recent purchases, abandoned purchases, complaints, and departures. Include customers who chose an alternative outside the conventional competitor set.
The immediate deliverable should be a clear outcome statement supported by observed behaviour. A software provider might focus on helping a finance team close its books accurately and on time. That definition gives product development a more useful target than increasing feature adoption alone.
Over time, the outcome should guide decisions about which capabilities to build, partner for, or discontinue.
2. Test technology against the customer’s entire task.
Choose one recurring problem and establish a baseline before introducing AI or another intervention. Measure successful completion, accuracy, total customer effort, repeat contact, and cost.
A pilot should reveal whether the intervention improves the task from beginning to end. Faster handling at one stage may simply transfer work to another employee or to the customer. Predetermined success criteria help leaders distinguish a promising demonstration from an improvement worth expanding.
The longer term opportunity is to redesign the service around what the test reveals, including the work customers should no longer have to perform.
3. Give customer outcomes a place in budgets and incentives.
At the next planning review, assess major investments against the outcomes they are expected to improve. Assign an accountable leader where delivery crosses departments, and pair financial targets with a small number of customer measures.
This requires explicit trade-offs. A proposal that protects current revenue may deserve less funding than one that serves customers better and establishes a credible future source of profit. Leaders need evidence for that decision, together with a transition plan.
Over time, incentives should reward the profitable delivery of customer value, including improvements that reduce dependence on an established product.
Levitt’s article remains relevant because success still encourages companies to treat their current offer as permanent. AI raises the stakes by allowing them to improve that offer at greater speed.
At the next strategy meeting, put one question beside the investment plan: If customers could achieve the same result without our current product, what would they still pay us to do?
The answer may reveal where the next business begins.