The finding that complicates this entire series
Dan Lovallo and Olivier Sibony examined 1,048 major strategic decisions made across five years — new product launches, restructurings, market expansions. For each one, they asked managers how far they had applied seventeen specific practices. Eight of those practices concerned the quantity and detail of the analysis. Nine concerned the process by which the decision was made.
Then they compared both against outcomes: revenue, profit, market share.
Process mattered more than analysis by a factor of six.
Their summary of why is the part worth holding onto. A good process often improved the analysis, by exposing faulty logic. The reverse did not hold. Superb analysis is useless unless the decision process gives it a fair hearing.
This series has spent nine articles on frameworks. Frameworks are analysis. The research says analysis is the smaller half.
That is not a reason to abandon them. It is the reason the preceding nine articles kept ending with diagnostics, decision rules, and questions about falsifiability rather than with templates. Those are process. This final piece makes the argument explicit.
Three ways a framework produces a confident wrong answer
1. Selection failure
A framework constrains attention. That is its mechanism and its value — it tells a team which questions to answer and which considerations to set aside.
Which means applying one built for a different question produces an answer that is structurally sound and about the wrong thing. Nothing in the process signals the error. The framework performed exactly as designed.
Structural analysis will tell you what a category permits and will never tell you whether customers want the thing. Customer outcome research will rank underserved needs and will never tell you whether the category can support your margins. An evolution map will show what is about to commoditise and has nothing to say about whether the market is worth serving. Positioning will make a choice legible and cannot make a choice that was never made.
The tell is when a team names the framework before naming the decision. "We should do a SWOT" precedes any statement of what is actually in question. The output will look like rigour.
2. Evidence inflation
The strategy field runs on numbers that have been repeated until they sound like findings.
This series found one in almost every article. Not because the frameworks are bad — most are useful — but because a practitioner guideline stated confidently enough eventually gets cited as research, and then a board decision rests on it.
The catalogue from the preceding nine pieces:
95% of new products fail, attributed to Clayton Christensen. Very hard to trace to a primary study. A 2022 academic paper examining the claim found no empirical support for the widely repeated 90% figure.
86% success rate for Outcome-Driven Innovation, against a 17% industry average. Originates from Strategyn's own track-record study. Never independently replicated.
40% as the product-market fit threshold. A pattern Sean Ellis observed across roughly 100 startups. Not independently validated, and routinely run at sample sizes where the confidence interval spans the threshold in both directions.
3:1 LTV to CAC. A guideline published by David Skok around 2010, derived from mature public SaaS companies at steady state. Now applied to seed-stage companies and to business models with entirely different cash profiles.
1% price increase produces 11% profit growth. Arithmetic, not a market experiment. It holds volume constant, which real price increases do not.
The P&G turnaround as proof of the strategy cascade. A single case, reported by its own participants, not repeated in the same leader's second tenure.
Each of those is worth knowing. None of them is worth citing as evidence. The distinction is the difference between using a framework and being used by one.
3. Confirmation by construction
Some frameworks cannot produce a disconfirming result.
The narrative branch of Jobs to be Done is the clearest example. If a hypothesised job fails to predict behaviour, the analyst re-narrates until it fits. The theory absorbs every outcome. Blue Ocean Strategy has a related problem: its research examines successful market creations and derives shared traits, without examining the businesses that did the same things and failed.
A framework that cannot be wrong will confirm whatever the team already believed, in new vocabulary. That is worse than no framework, because it arrives with the feeling of having tested something.
What process actually means here
Not governance theatre. Five specific practices, all cheap, all drawn from what the preceding articles kept arriving at independently.
Name the decision before the framework. Write the question in one sentence, with the options. Then select the tool that answers that question. Reversing this is the most common error in the series.
State the disconfirming result in advance. What would we see if we are wrong. If nobody can answer, the exercise will confirm rather than test. This is the same discipline as writing a pilot's decision rule before launch, applied to analysis.
Grade the evidence out loud. Empirically grounded, documented practitioner method, or thinly evidenced. Say which, in the room, when a number is introduced. It costs one sentence and changes how much weight the number carries.
Require an exclusion. A strategic conclusion that rules nothing out has not concluded. This applies to every framework in the set, not just the strategy cascade.
Give dissent a structural home. Lovallo and Sibony's finding is largely about whether contrary views get a hearing. Analysis that is never challenged does not improve. Assigning someone to argue the opposite case is a process intervention, not a personality one.
The same research found that organisations working to reduce bias in their decision-making achieved returns several percentage points higher. The mechanism is unglamorous and it is available to any leadership team without buying anything.
This is not an argument against frameworks
It would be easy to read all of this as a case for intuition, and it is the opposite.
The Lovallo and Sibony finding is not that analysis is worthless. It is that analysis is necessary and insufficient, and that most organisations invest heavily in the necessary half while leaving the insufficient half to chance. A separate McKinsey survey of over 2,000 executives found only 28% describing the quality of strategic decisions in their organisation as generally good, with 60% saying bad decisions were as frequent as good ones.
Those organisations were not short of frameworks.
Frameworks earn their place by constraining attention, forcing exclusions, and making inconsistency visible. The ones covered in this series do that well when matched to the right question and read with honest evidence standards. What none of them can do is supply the process around themselves.
Diagnostic: is this analysis or is it justification?
Seven tests, run on any strategy work in progress.
The decision was written down before the framework was chosen.
Someone can state what result would have changed the conclusion.
Every number introduced has a stated evidence grade.
At least one option was excluded by name.
A contrary case was argued by someone whose job it was to argue it.
The conclusion is different from what the team believed at the start of the process — or, if not, someone can say what evidence held it in place.
The framework used is the one that answers the question, not the one the team already knows.
Test six is uncomfortable and the most diagnostic. Analysis that never changes anyone's mind is expensive agreement.
Closing the Define stage
Across ten articles, the pattern that recurred was not that the frameworks were wrong. It was that they were treated as producing answers rather than as constraining questions, and that the numbers inside them were treated as findings rather than as guidelines.
The Define stage closes when a leadership team can state what it has chosen, what it has ruled out, what evidence supports that, and what would prove it wrong. Frameworks help reach that point efficiently. They are not what the team is trying to produce.
A decision specific enough to fund, and specific enough to be wrong. Everything downstream — the structure that gets built, the growth that gets driven, the systems that scale — depends on the quality of that one output.
Frequently asked questions
Are strategy frameworks worth using?
Yes, when matched to the decision at hand. Research on 1,048 major strategic decisions found process mattered six times more than analysis, which means frameworks are necessary and insufficient. The failure mode is treating the framework as the process.
What is framework selection failure?
Applying a framework built for a different question. The output is internally coherent and about the wrong thing, and nothing in the process signals the error. The most reliable tell is a team naming the framework before naming the decision.
How should we treat statistics in business literature?
Grade them. Empirically grounded means independent, replicated research. Documented practitioner method means field-developed and internally consistent, with evidence produced by those who sell it. Thinly evidenced means popular and resting on selected cases. Most widely cited strategy statistics fall into the second and third categories.
Why do some frameworks never produce a negative result?
Because they permit re-interpretation after the fact, or because their supporting research examined only successful cases. Either way the framework absorbs every outcome and confirms whatever was already believed.
What is the cheapest way to improve strategic decisions?
Write the disconfirming result before the analysis starts, and assign someone to argue the opposite case. Both are process interventions, cost nothing, and address the half of decision quality that most organisations leave to chance.
How many frameworks should a decision use?
Two or three per cycle, selected after the decision has been stated. More produces parallel documents that never resolve into a choice.
Does this mean we should trust intuition instead?
No. The finding is that analysis is insufficient without process, not that analysis is unnecessary. Organisations that reduced bias in their decision processes achieved measurably higher returns — an improvement in how analysis was used, not a replacement for it.
Sources
Lovallo, D. and Sibony, O., The Case for Behavioral Strategy, McKinsey Quarterly (2010) — analysis of 1,048 major strategic decisions across five years, and the McKinsey survey of over 2,000 executives on decision quality
Kahneman, D., Lovallo, D. and Sibony, O., Before You Make That Big Decision, Harvard Business Review (2011)
Evidence qualifications drawn from the preceding nine articles in this series, with primary sources cited in each
Structure your next phase
Zerologic works with founders and leadership teams through the Define stage — structural analysis, demand evidence, positioning, validation, and a funded choice specific enough to be proven wrong.
Talk to us: partners@zerologic.io · zerologic.io
