Does Your Business Need a New Strategy—or Better Execution?

The quarterly review lasts eleven minutes before someone says it: “The strategy was right. We just failed to execute.” Nobody asks what result a successful execution was supposed to produce.

That missing question is the whole problem.

A disappointing result can come from at least three places. The strategic theory may be wrong. The organization may not have enacted that theory. Or the conditions that made the theory plausible may have changed. The result alone cannot tell you which one happened.

A bad result is an alarm. It is not a diagnosis. To decide whether a business needs a new strategy or better execution, reconstruct three records: what the strategy predicted, what the organization actually did, and whether the world stayed inside the strategy's assumptions. Then ask which counterfactual still looks credible.

A counterfactual is a disciplined statement about what would probably have happened under a different condition. Here, the useful question is not “Who failed?” It is: What would this result have looked like if the same strategy had been enacted faithfully under the conditions it assumed?

That question is harder than assigning blame. It is also much cheaper than changing the wrong thing.

A poor result does not name its cause

Management language encourages false certainty. Strategy sits on one side of a slide. Execution sits on the other. When performance falls short, leaders pick a side.

The research does not support such a clean contest. Cândido and Santos audited the familiar claim that 50–90 percent of strategy implementations fail. They found that the true failure rate remains unknown because many estimates are outdated, fragmentary, fragile, or unsupported. The famous number survives because it is memorable, not because it is a dependable diagnostic benchmark.

The word failure also hides several different events. A strategy can fail because its theory of value is false. An implementation can fail because critical actions or conditions never existed. A company can achieve a short-term target yet fail to learn whether the strategy remains useful. These are not synonyms.

Consider a business-to-business software company that wants to shorten a long sales cycle. Its strategy is to let qualified mid-market buyers evaluate price, fit, and security without waiting for a sales representative. The company launches transparent pricing, a configurator, a security portal, and a guided trial.

Six months later, the sales cycle has not moved.

The dashboard contains one fact. It supports several explanations:

• Buyers may still need a person because the main constraint is perceived procurement risk, not missing information.
• The configurator may be inaccurate, security documents may be incomplete, and sales incentives may pull every buyer back into the old process.
• The company may have moved upmarket during the test, where procurement is slower and self-service has a different role.

Each explanation leads to a different decision. The same number cannot select among them.

Reconstruct the strategy's prediction

Before asking whether a strategy worked, identify what the strategy claimed.

Felin and Zenger describe strategy as a firm-specific theory: a view about which problems to solve, which activities and assets to assemble, and how the organization will create value. This definition matters because a theory makes predictions. A direction such as “become more customer-centric” does not.

Write the prediction as a causal sentence:

If we make this choice for this customer, then this mechanism should change their behavior, which should create this form of value, provided these conditions remain true.

For the software company, the strategic record might say:

If qualified mid-market buyers can establish price, fit, and security without waiting for a representative, then they will reach a confident buying decision sooner because information delay is the main source of friction.

Now the strategy has load-bearing parts. The target is qualified mid-market buyers. The mechanism is earlier confidence from better information. The predicted behavior is faster progression. The intended value is a shorter sales cycle without a lower-quality customer. The key assumption is that information delay—not organizational risk, legal review, or internal consensus—is the binding constraint.

If leadership cannot reconstruct this chain, the company does not yet have evidence against a strategy. It has evidence against an aspiration, a project bundle, or a sentence that was too vague to fail.

This is where strategic and technical competence can part company. Tripsas and Gavetti's study of Polaroid found that the company developed relevant capabilities in digital imaging, but managerial cognition still shaped how leaders understood the new market and directed those capabilities. An organization can therefore build real capability and execute serious work while its commercial theory remains tied to an older world.

That is not sloppy execution. It is competent movement inside the wrong map.

Reconstruct what the organization enacted

The strategy document records intention. The implementation record must show what reached customers, employees, operations, and the market.

This distinction makes execution more than delivery. Lee and Puranam use a computational model to show why precise implementation can improve learning about strategy. When the enacted work closely represents the strategic intention, the result tells the organization more about the theory. When implementation varies in uncontrolled ways, success and failure become harder to interpret.

Implementation researchers call this fidelity: the degree to which an intervention was delivered as intended. Carroll and colleagues argue that measuring fidelity is necessary to understand how and why an intervention produced its result. Their framework comes from implementation science rather than corporate strategy, but the inferential problem is the same. You cannot judge an intended treatment by observing a different treatment.

For the software company, the implementation record cannot stop at “the portal shipped.” It needs to show whether:

• the target buyers could see complete, accurate prices;
• the configurator produced reliable answers for the intended use cases;
• the security portal resolved common objections;
• the trial created the experience the strategy promised;
• marketing attracted the intended segment;
• sales incentives allowed buyers to remain in the guided path; and
• analytics could distinguish qualified journeys from casual visits.

This is not a checklist for its own sake. Each item belongs to the predicted mechanism. If the mechanism was earlier confidence, then accuracy, completeness, continuity, and measurement are part of the test.

Execution is also broader than conformance. A review by Tawse and Tabesh organizes implementation around managerial actions, necessary organizational conditions, and the managerial capabilities that combine them. Friesl, Stensaker, and Colman identify five forms of implementation work: matching structures and processes, matching resources, monitoring, framing, and negotiating. A 2025 systematic review of 160 peer-reviewed papers similarly emphasizes interdependent managerial and organizational levers.

So faithful execution does not mean following every original task without thought. It means preserving the strategy's load-bearing causal commitments while adapting the methods that carry them. If legal review changes, the team may change the security workflow. It may not quietly remove the evidence that buyers need and still call the test faithful.

The useful distinction is core versus adaptable:

• Core elements express the customer, value claim, mechanism, and trade-off that make this strategy this strategy.
• Adaptable elements are the local methods, sequence, tools, and coordination choices that can change without replacing the theory.

Without this distinction, “fidelity” becomes bureaucratic obedience. With no fidelity at all, “adaptation” becomes a polite word for testing something else.

Check whether the conditions changed

Even a sound theory and faithful enactment can produce a misleading result when the test environment changes.

The Medical Research Council's guidance for evaluating complex interventions separates implementation, mechanisms, and context. Although the guidance addresses public-health research, its causal discipline is useful for business. An intervention does not operate in a vacuum. Conditions can alter delivery, activate a different mechanism, or change the meaning of the outcome.

For the software company, a move from mid-market buyers to large enterprises changes the test. The strategy may still improve access to information, but enterprise procurement can add legal review, security committees, budget cycles, and consensus among more stakeholders. A flat sales-cycle metric no longer tests the original claim cleanly.

Context is not a universal excuse. It must name a condition that was load-bearing for the theory or the enactment. “The market was difficult” explains nothing. “The target account shifted from one economic buyer to a seven-person buying committee, so information delay was no longer the main constraint” is a testable explanation.

Past success can make this harder to see. Audia, Locke, and Smith combined a ten-year archival study of airline and trucking industries with a laboratory study. They found that greater past success increased strategic persistence after radical environmental change, and that persistence contributed to later performance decline. Lant, Milliken, and Batra also found that performance, managerial interpretation, team characteristics, and environmental context affected strategic reorientation.

In other words, numbers arrive inside a story that managers are already telling. A leadership team can treat a bad result as temporary because the old strategy once worked. Another team can treat the same result as proof of disruption because a new strategy is politically attractive. Neither interpretation is neutral.

The context record disciplines that story. It asks which relevant conditions changed, when they changed, how strongly they affect the predicted mechanism, and whether the strategy anticipated them.

Compare the three records at the point of divergence

You now have three records:

1. Theory: what choices, mechanism, value, and conditions the strategy predicted.
2. Enactment: what people, systems, incentives, processes, and customer experiences actually existed.
3. Conditions: what changed in the market, customer, regulation, technology, competition, or organization during the test.

Do not score the records and average them. Find the earliest consequential divergence.

If the enacted work diverged before the predicted mechanism could operate, repair execution. The strategy has not received a fair test.

If the work preserved the mechanism but the expected intermediate behavior did not appear under valid conditions, the strategic theory is in trouble. Do not keep adding delivery effort to protect it.

If a load-bearing condition changed, revise the relevant strategic assumption. The old theory may have been reasonable and still be wrong now.

If several records diverged together, the result is confounded. That word is not an escape hatch. It means the evidence cannot isolate cause. The next move is a smaller test.

Causal record

One result leaves three different records

Compare the strategic theory, enacted work, and operating conditions at the point where expectation and observation first separate.

AStrategic theory

Recorded before the test

Better information creates confident decisions sooner

Inspected after the result

Did confidence change before the final outcome?

BEnacted work

Recorded before the test

Price, fit, security, incentives, and measurement all operate

Inspected after the result

Which load-bearing commitment reached practice?

COperating conditions

Recorded before the test

The target buyer and binding constraint remain stable

Inspected after the result

What changed before the result appeared?

Decision ruleFind the earliest consequential divergence.

Repair that layer first. If several records moved together, isolate one with a smaller test.

Return to the software company. Suppose the team discovers that the configurator was accurate, security evidence was complete, and buyers used both. Qualified mid-market buyers moved through evaluation faster, but the company shifted half its acquisition budget to enterprise accounts during the same period. The aggregate sales cycle stayed flat.

The result does not show that self-service failed. It shows that the company mixed two populations with different buying mechanics. The strategic claim may survive for the original segment, while the portfolio needs a different theory for enterprise accounts.

Now change one fact. Suppose the segment remained stable, enactment was faithful, buyers used the tools, and interviews show that internal approval—not information delay—caused most waiting. The strategy's mechanism failed. Better execution would only produce a more polished answer to the wrong problem.

Change another fact. Suppose buyers wanted the tools, but sales representatives withheld pricing because compensation rewarded assisted deals. The organization did not enact the strategy's trade-off. The primary correction belongs in incentives and operating design, not in the customer theory.

The cases look similar on a dashboard. The records make them different.

Choose the smallest test that could change the decision

When evidence is mixed, organizations often respond with a larger transformation. This is emotionally satisfying and diagnostically reckless. More moving parts create more possible causes.

Design the smallest next test that could change the decision. It should hold two records stable while it examines the third.

To test execution, keep the strategic theory and target conditions stable. Define the minimum faithful enactment, remove one known barrier, and observe whether the predicted intermediate behavior appears.

To test strategy, keep enactment stable enough that the mechanism receives a fair trial. Then test the uncertain assumption directly. For the software company, that could mean comparing whether better information changes buyer confidence while procurement risk remains measured separately.

To test context, separate populations, periods, or conditions that activate different mechanisms. Do not bury a segment shift inside one aggregate metric.

Before the test begins, write the revision rule. State which evidence would support execution repair, which evidence would challenge the strategic theory, which condition would limit the conclusion, and when the company will stop. A result cannot remain informative if every outcome is reinterpreted to protect the preferred answer.

The decision rule should also specify a time horizon. Some mechanisms need time to appear. Others should appear early. If the strategic theory predicts faster buyer confidence, the business should not wait a year to see whether buyers can complete a security assessment. If it predicts lower churn from deeper adoption, a two-week test may be meaningless.

This is where rigor becomes practical. The aim is not to turn every executive meeting into a doctoral defense. The aim is to make the next expensive move depend on evidence that can distinguish its alternatives.

The honest answer can be “not yet”

Organizations often prefer a wrong verdict to an incomplete one. “Execution failed” protects the strategy. “The strategy failed” protects the delivery team. “The market changed” protects both. Each label can become an incentive before it becomes an explanation.

Research on organizational learning gives that concern some weight. Beer and Eisenstat studied a self-diagnosis and redesign process used across 20 Alpha Technologies business units from 1988 to 1994. The process met its intended short-term objectives, but it did not appear to improve the organization's underlying capacity for learning. Completing a correction and building a system that can diagnose future corrections are different achievements.

A mature strategy process makes uncertainty visible without making it permanent. It records the theory, observes the enactment, monitors the conditions, and narrows ambiguity with bounded tests. It also accepts that a previously sound strategy can expire and that impressive execution can faithfully deliver evidence against it.

That brings us back to the quarterly review. When someone says, “The strategy was right; we just failed to execute,” the useful response is not immediate agreement or opposition.

Ask what the strategy predicted. Ask what work actually reached the world. Ask which conditions moved. Then ask what would probably have happened if the theory had received a faithful test under valid assumptions.

If that counterfactual still creates value, repair execution. If it no longer does, change the strategy. If the evidence cannot tell, make the next test smaller until it can.

The business does not need a winner in the strategy-versus-execution argument. It needs an attributable reason for its next decision.

References

• Audia, P. G., Locke, E. A., and Smith, K. G. (2000). The paradox of success: An archival and a laboratory study of strategic persistence following radical environmental change. Academy of Management Journal, 43(5), 837–853.
• Beer, M., and Eisenstat, R. A. (1996). Developing an organization capable of implementing strategy and learning. Human Relations, 49(5), 597–619.
• Cândido, C. J. F., and Santos, S. P. (2015). Strategy implementation: What is the failure rate?. Journal of Management & Organization, 21(2), 237–262.
• Carroll, C., Patterson, M., Wood, S., Booth, A., Rick, J., and Balain, S. (2007). A conceptual framework for implementation fidelity. Implementation Science, 2, 40.
• Felin, T., and Zenger, T. R. (2017). The theory-based view: Economic actors as theorists. Strategy Science, 2(4), 258–271.
• Friesl, M., Stensaker, I., and Colman, H. L. (2021). Strategy implementation: Taking stock and moving forward. Long Range Planning, 54(4), 102064.
• Holm, C. G., Kringelum, L., and Anand, A. (2025). Creating effective strategy implementation: A systematic review of managerial and organizational levers. Review of Managerial Science.
• Lant, T. K., Milliken, F. J., and Batra, B. (1992). The role of managerial learning and interpretation in strategic persistence and reorientation. Strategic Management Journal, 13(8), 585–608.
• Lee, E., and Puranam, P. (2016). The implementation imperative: Why one should implement even imperfect strategies perfectly. Strategic Management Journal, 37(8), 1529–1546.
• Moore, G. F., et al. (2015). Process evaluation of complex interventions: Medical Research Council guidance. BMJ, 350, h1258.
• Tawse, A., and Tabesh, P. (2021). Strategy implementation: A review and an introductory framework. European Management Journal, 39(1), 22–33.
• Tripsas, M., and Gavetti, G. (2000). Capabilities, cognition, and inertia: Evidence from digital imaging. Strategic Management Journal, 21(10–11), 1147–1161.