Using o1-preview for Multi-Step Strategic Analysis and Business Case Development

What You'll Learn

• How to structure a prompt that activates o1's full reasoning capability
• The specific business case format that produces board-ready output
• When o1-preview outperforms GPT-4o and when it does not

o1-preview is not better than GPT-4o at everything. It is measurably better at one specific category: multi-step problems with dependencies, constraints, and no obvious single correct path. That description fits most executive analysis work.

Market entry decisions, build-versus-buy assessments, organizational restructuring, capital allocation under uncertainty -- all of these require a model that holds multiple variables in tension rather than pattern-matching to a familiar answer shape.

This Pro Tip shows the exact prompt structure that produces board-ready strategic analysis from o1-preview, with a worked example you can adapt immediately.

SUBSCRIBER BREAK -- Premium Content Below

The Prompt Architecture

o1-preview responds to structured input better than open-ended questions. Three-block structure: Context (situation, constraints, knowns), Task (exact output format), Criteria (what a quality answer must satisfy). Explicit criteria are especially important -- o1 checks its own output against them during the reasoning process.

You are a strategic analyst supporting an executive decision. Reason through this problem step by step before producing any output. Context: [Company] is evaluating [decision]. Current state: [size, revenue, team, product, geography]. Key constraints: [budget, timeline, regulatory, organizational]. Relevant knowns: [competitive landscape, market data, prior attempts]. Task: Produce a structured business case in this format: 1. Executive Summary (3 sentences: situation, recommendation, key risk) 2. Market and Competitive Context (key facts, not opinions) 3. Options Analysis (enumerate all viable options, state assumptions for each) 4. Risk Assessment (likelihood and severity for each option's primary risks) 5. Recommendation with Rationale (why the recommended option beats alternatives on the criteria below) 6. Open Questions (what information would change the recommendation) Criteria: All assumptions explicitly stated. Every claim tied to a fact or stated assumption. Recommendation follows logically from the analysis. No jargon. A board member with no context could read the summary and understand the decision.

Worked Example

Situation: B2B SaaS company, $8M ARR, evaluating whether to build a native mobile app or continue with mobile-responsive web only. Engineering team of 12, 18-month runway.

What o1 produced: A structured analysis covering three options (native iOS/Android, progressive web app, status quo), with explicit assumptions about mobile conversion lift, engineering time cost, and opportunity cost. The recommendation favored the progressive web app with a six-week development window, with the primary risk (limited offline functionality) rated medium likelihood and low severity. The open questions section flagged two customer interviews needed before committing.

That output required no editing to share with the board. GPT-4o on the same prompt produced a reasonable framework but omitted the open questions section, stated assumptions implicitly rather than explicitly, and did not cross-reference the options against the stated constraints.

Routing Guide

Use o1-preview for: business cases, scenario modeling, risk frameworks, complex contract analysis, strategic options assessment.

Use GPT-4o for: first-draft writing, summarization, web browsing tasks, image analysis, quick factual questions, anything requiring real-time data.

The models are complements. Route by task type, not by default habit.