Strategy Trendslop: Why AI Needs Decision Discipline

AI can now produce strategy language in seconds. For busy leaders, that can feel like progress: polished recommendations, balanced options and apparently thoughtful reasoning.

But there is a harder question beneath the fluent output: is the AI helping you make a better strategic choice, or simply giving you an answer that sounds strategically credible?

A Harvard Business Review article on “strategy trendslop” raises exactly this concern. Its central argument is unsettling: large language models (LLMs) can favour fashionable management ideas over the approach that is strategically right for a particular organisation, market and moment.

That matters because strategy has never been about selecting the most socially acceptable answer.

Strategy is about choice.

And choice means trade-offs.

What is strategy trendslop?

Strategy trendslop describes the tendency for AI systems to reproduce dominant ideas in management discourse rather than assess what genuinely fits the situation.

For example, an LLM may naturally favour recommendations such as:

• Collaboration over competition

• Long-term thinking over short-term focus

• Differentiation over cost leadership

• Human augmentation over automation

None of these positions is inherently wrong. In the right circumstances, each could be an excellent strategic choice.

The problem emerges when they become default answers rather than conclusions reached through evidence, comparison and judgement.

A recommendation can sound intelligent and still be strategically weak if it ignores market conditions, competitive realities, timing constraints, financial pressures or execution capability.

The danger for leaders is subtle because polished advice is persuasive. But persuasive is not the same as correct.

Strategy has always been about difficult choices

Many strategic failures happen not because organisations lacked ideas, but because they avoided making choices.

Michael Porter famously argued that strategy requires trade-offs. Trying to pursue incompatible positions simultaneously can leave an organisation “stuck in the middle”: neither distinctive enough to command a premium nor efficient enough to win on cost.

Yet AI models can be inclined towards compromise. The HBR article highlighted what might be called the “hybrid trap”.

When models were not forced to choose, they often tried to combine strategic alternatives:

• Differentiate while pursuing cost leadership.
• Invest in both radical innovation and incremental optimisation.
• Optimise for short-term gains while maximising long-term transformation.

At first glance, these recommendations can appear sophisticated. They seem balanced, flexible and difficult to disagree with.

But experienced executives recognise the problem immediately. Resources are finite. Attention is finite. Capabilities take time to build. Many strategic tensions cannot simply be merged.

Sometimes choosing one path means explicitly declining another.

That is uncomfortable. Saying “both” feels safer than making a commitment. AI can unintentionally reinforce that instinct by presenting compromise as wisdom, even where it obscures the real decision.

Why better prompting is not enough

It is tempting to assume that AI bias can be fixed through prompt engineering:

• Provide more context.
• Ask better questions.
• Use more detailed instructions.

These steps matter. Clearer inputs can produce more useful outputs. But they do not, by themselves, create reliable strategic judgement.

The article found that adding context helped only modestly and that changes in prompting reduced bias inconsistently. More concerningly, the order in which options appeared could influence recommendations.

In other words, AI can appear to reason while remaining surprisingly sensitive to how a choice is presented.

For organisations building AI-enabled strategy processes, the lesson is clear: prompt design matters, but prompt design alone is not enough.

You cannot engineer sound strategic judgement purely through better wording. You need a process that tests assumptions, compares alternatives, exposes uncertainty and records the reasoning behind decisions.

The difference between AI assistance and AI theatre

Many AI strategy tools risk becoming “ChatGPT in a strategy wrapper”.

The experience feels impressive:

• Ask a question.
• Receive polished recommendations.
• Generate attractive outputs.
• Move on.

But if recommendations remain unaudited, unchallenged and disconnected from evidence, organisations may simply accelerate poor decisions.

The issue is not that AI lacks useful capabilities. It can support research, structure thinking, identify patterns, generate options and improve the quality of strategic discussion.

The issue is that strategy requires accountability.

• Who owns the recommendation?
• What assumptions drove it?
• What evidence supports it?
• What alternatives were rejected?
• What trade-offs were considered?
• What would need to be true for this decision to succeed?

Without answers to these questions, strategic outputs become difficult to trust.

And trust matters. Strategy shapes investment decisions, organisational priorities and the effort people put into execution. A convincing answer is not enough when the consequences are real.

Why StratNav was built differently

This challenge reinforces the direction we have taken with StratNav.

The objective was never to build AI that acts as “the strategist”. The objective is to build AI that strengthens strategic thinking.

That distinction matters.

StratNav is designed around the idea that better strategy comes from structured thinking, evidence and decision discipline. AI can assist the process, but it should not hide the reasoning or remove human responsibility for the choice.

In practice, that means helping organisations to:

Surface strategic alternatives

Rather than assuming there is one obvious answer, good strategy examines meaningful competing options. It makes the choices visible before leaders commit resources.

Capture rationale and debate

Decisions rarely emerge fully formed. Good strategy often develops through discussion, disagreement and iteration. Recording the rationale helps teams understand not only what was decided, but why.

Link recommendations to evidence

Recommendations should not simply appear as polished statements. Leaders need to understand the evidence, assumptions and analysis that support them.

Maintain human accountability

AI can support analysis. Humans remain responsible for judgement, priorities and commitments.

Support execution as much as planning

A strategy creates value only if it changes behaviour. Execution discipline matters as much as strategic insight.

These principles become more important as AI becomes more capable. The better AI becomes at sounding convincing, the more important it is to have a structured process for deciding what to believe, what to test and what to do next.

The missing layer: decision discipline

The issue goes deeper than model bias.

LLMs are not weak because they lack eloquence. Their limitation is a lack of groundedness. They generate outputs from patterns in data; they do not carry the responsibility for committing your organisation to a course of action.

Strategy, by contrast, requires commitment.

Those are different activities.

This creates an opportunity for systems such as StratNav to go beyond recommendation generation. Useful AI-enabled strategy workflows should include:

Evidence trails

Users should be able to see the sources, observations and supporting information linked directly to insights and recommendations.

Forced comparisons

Alternatives should be considered explicitly, rather than allowing one appealing answer to crowd out the rest.

Counter-argument generation

Systems should help teams consider the strongest case against a proposed direction. This does not eliminate judgement, but it can make weak assumptions easier to spot.

Human checkpoints

Leaders should actively accept or reject recommendations and explain why. This keeps accountability where it belongs.

Decision histories

Organisations should be able to track how their strategic thinking changes over time, including shifts in evidence, assumptions and AI-supported outputs.

Strategic judgement does not improve by hiding uncertainty.

It improves by exposing it, discussing it and making informed choices despite it.

The question leaders should ask

The real question may not be: “Can AI create strategy?”

Increasingly, AI can generate elements of strategy: analyses, options, narratives, frameworks and recommendations.

The better question is: “How do we stop AI-generated strategy from becoming polished nonsense?”

If strategy is fundamentally a set of decisions, organisations need systems that make difficult decisions visible and unavoidable. They need to see the alternatives, challenge the assumptions, weigh the evidence and own the trade-offs.

The future probably does not belong to AI replacing strategists.

It belongs to AI helping leaders think more rigorously, challenge assumptions and make stronger choices.

That is a very different proposition.

And perhaps a much more valuable one.

Interested in exploring how AI and structured strategy can work together?

Schedule a demo of StratNav or try StratNav for free.

You can also book a discussion at Chris C Fox Consulting if you would like to explore how strategy decision systems could strengthen your organisation’s planning and execution process.