Strategic Planning Without the $500K Price Tag: Using AI to Replicate Top-Tier Consulting Frameworks
The consulting industry's disruption moment that never quite happened
Back in 2013, Harvard Business Review predicted imminent disruption in the global consulting space, citing price as one of the main vulnerabilities. The argument was sound: if you're a small-to-midsize business facing a strategic challenge, you're stuck between options that don't quite work. Large consulting firms bring proven frameworks and deep expertise, but they cost millions. Smaller agencies or independent practitioners offer more accessible pricing, but projects drag on for months—and in today's rapidly moving markets, a strategic plan that takes six months to develop has often lost relevance before implementation even begins.
More than a decade later, that disruption still hasn't materialized. But the conditions that make it inevitable have only intensified. Strategic planning remains one of the most valuable capabilities a business can access—the consulting industry was valued at $327.65 billion in 2023 and is projected to reach $494.36 billion by 2033. Deloitte and McKinsey have both seen steady growth over the past decade, with Deloitte doubling its revenue to $65 billion in just ten years. Meanwhile, companies from Bayer to Google have built formidable in-house consulting teams, recognizing that this capability is too valuable to outsource entirely.
The value is clear. The accessibility remains the problem.
What AI actually changes (and what it doesn't)
Here's what AI can't do: it can't replace the strategic frameworks that make consulting valuable in the first place. The TOSCA problem definition framework, the MECE principle for issue trees, hypothesis-driven analysis—these aren't artifacts of the pre-AI era. They're structured approaches to complex problem-solving that remain as relevant as ever.
What AI can do—when applied thoughtfully—is dramatically reduce the operational costs of using these frameworks effectively. A well-prompted LLM can guide you through a TOSCA analysis by asking the right questions in sequence. It can generate an initial issue tree following MECE principles. It can help prioritize solutions using benefit-to-effort analysis. It can create detailed work breakdown structures in minutes rather than days.
The catch? You need to understand the frameworks deeply enough to guide the AI, evaluate its outputs critically, and refine them into actually usable strategic plans. This isn't about prompt engineering tricks. It's about knowing what good strategic thinking looks like, then using AI to accelerate the process of getting there.
That's the hypothesis I wanted to test.
Building a replicable methodology
I've studied many strategic problem-solving frameworks from a former McKinsey consultant, and spent years implementing these approaches in real business contexts. When I started experimenting with using LLMs to streamline this work, I wasn't trying to replace human strategic thinking—I was trying to compress the timeline and reduce the overhead of applying proven frameworks.
I actually used an early version of this methodology during the strategic planning phase of the Celigo rebrand project. Before diving into visual design, I needed to structure a complex organizational problem: how to execute an enterprise-level rebrand with a lean team, aggressive timeline, and zero room for the typical iterative back-and-forth that bogs down brand projects. Using the 4S framework helped me identify the real constraint (stakeholder alignment, not creative direction), structure the solution around a single comprehensive presentation rather than phased reveals, and sequence the work to build foundational systems first. That strategic clarity is what enabled us to deliver an award-winning rebrand in a fraction of the typical timeline—and it's what convinced me this approach had broader applications worth systematizing.
The methodology I developed centers on the 4S process: State, Structure, Solve, and Set in Motion. This isn't my invention—it's an established approach used by top consulting firms. But the workflow I built around it, leveraging AI at specific points while maintaining human oversight at critical junctures, is designed to deliver comparable quality in a fraction of the time.
Here's how it works in practice:
Phase 1: STATE – defining the core problem
Most strategic initiatives fail not because of poor execution, but because teams never properly defined what problem they were actually trying to solve. The TOSCA framework addresses this by forcing clarity across five dimensions:
Trouble: What's the actual symptom? Not assumptions about root causes—observable evidence of the gap between current state and desired state.
Owner: Whose problem is this? Who decides whether a solution succeeds?
Success: How will we know the problem is solved? What specific, measurable outcomes define success?
Constraints: What limits must any solution respect? Time, budget, resources, existing commitments?
Actors: Who else cares about how we solve this problem, and what do they want?
The traditional approach involves a consultant asking these questions through iterative conversations, documenting responses, and synthesizing a problem statement. With an LLM, you can replicate this process using a carefully structured prompt that instructs the AI to act as a strategy consultant, asking one question at a time, moving through each dimension of TOSCA systematically.
The key is preparing properly. Before engaging the AI, document your situation using this framework:
General information on the situation
Background information on the business
The challenge you're facing
What you've tried so far to overcome the challenge
The results of your attempts so far
This context lets the AI ask more intelligent follow-up questions. But here's the critical part: the AI's role is to prompt discussion and synthesis, not to provide final answers. Every output should be reviewed, refined, and iterated on with your team until you have genuine alignment on the problem statement.
When you're finished with this phase, you should have a clear, actionable problem statement—often formatted as a "How might we...?" question that addresses the trouble, is written from the owner's perspective, leads to your defined success, respects your constraints, and acknowledges your various actors' concerns.
Phase 2: STRUCTURE – breaking down the problem
With a clear problem statement, the next challenge is making the problem manageable. This is where most teams either get paralyzed by complexity or jump to premature solutions.
The issue-driven approach uses an issue tree following the MECE principle: Mutually Exclusive (each issue is distinct, with no overlaps) and Collectively Exhaustive (together, the issues represent all possible approaches without gaps). Building a good issue tree is one of those skills that separates strong strategic thinkers from everyone else—it requires both analytical rigor and creative problem decomposition.
An LLM can generate an initial issue tree remarkably well if properly prompted. The key is instructing it to:
Systematically break down the primary problem into sub-problems and further detailed branches (2-4 levels deep)
Ensure no overlaps exist between branches
Comprehensively cover all potential areas relevant to the primary problem
Optimize the structure for strategic discussion
But an AI-generated issue tree is never the final output. It's a starting point for team discussion. I've found the most effective workflow is:
Generate initial tree in ChatGPT (using GPT-4 with Canvas for better control)
Review and identify branches that need expansion or clarification
Use the AI to expand specific branches (you can highlight text and ask for more detail on just that section)
Export the refined tree to Miro and visualize it using their AI-powered mindmap tool
Use Miro's AI to expand branches further with questions and ideas
Review and refine as a team until the tree satisfies MECE
This phase takes longer than the initial problem definition, but it's time well spent. A solid issue tree becomes the foundation for everything that follows.
Phase 3: SOLVE – analyze and prioritize
With a comprehensive issue tree, you're facing a new problem: you've identified dozens of potential approaches, but you can't pursue them all simultaneously. This is where prioritization becomes critical.
The traditional consulting approach uses a benefit-to-effort matrix. For each option, estimate the benefit (1-10) and the effort required (1-10), then prioritize options with high benefit-to-effort ratios. Simple in concept, tedious in execution—especially when you're evaluating 20+ options.
An LLM can accelerate this dramatically. You provide your issue tree and ask it to evaluate each issue across both dimensions, providing supporting arguments for each score. The prompt instructs the AI to:
Consider the benefit in terms of contribution to your defined success factors
Measure effort in terms of time and cost
Create a table with all issues, scores, arguments, and benefit-to-effort ratios
The AI's initial scoring gives you a starting point for team discussion. Some scores will feel obviously right. Others will spark valuable debate: "Wait, they rated this as low effort, but we know our systems can't actually do that easily." That debate is the point. The AI accelerates getting to the meaningful strategic conversations rather than spending hours on individual scoring exercises.
After refining the prioritization as a team, you'll have clarity on which 20% of options are most likely to drive 80% of your results—the classic Pareto principle applied to strategic planning.
Phase 4: SET IN MOTION – create the work breakdown structure
A strategic plan that doesn't include clear execution steps is just an expensive thought exercise. The final phase translates your prioritized solutions into an actionable work breakdown structure (WBS).
A comprehensive WBS includes:
Issues and sub-issues (from your tree)
Tasks required to address each issue
Concrete to-dos for each task
Proposed start dates
Duration estimates
Proposed end dates
Dependencies between tasks
Building this manually is time-intensive and error-prone. You're juggling dozens of tasks, trying to sequence them logically, identifying dependencies, estimating timelines based on resource constraints—it's the kind of detailed project management work that often gets rushed because teams are exhausted after the strategic analysis.
An LLM can generate a comprehensive WBS remarkably well, using all the context from previous phases. The prompt instructs it to work through each issue systematically, considering the entire scope of the issue tree while prioritizing tasks appropriately. It outputs a detailed table with all necessary columns.
But again: review and refinement are essential. The AI might suggest unrealistic timelines, miss critical dependencies, or propose task sequences that don't account for your specific resource constraints. The team needs to go through the WBS systematically, adjusting based on operational reality.
Once refined, you can export the WBS as a CSV and import it into Google Sheets or TeamGantt to visualize as a Gantt chart. This gives you a professional project plan that clearly communicates the work ahead to stakeholders.
The real disruption isn't AI—it's strategic literacy
The consulting industry won't be disrupted by AI alone. It will be disrupted when more business leaders develop strategic literacy: understanding these frameworks well enough to guide AI effectively, evaluate outputs critically, and facilitate the discussions that translate analysis into action.
That's not a threat to strategic expertise—it's an elevation of the standard. The consultants who thrive will be those who can work at a higher level of strategic thinking, using AI to handle the mechanical parts of framework application while focusing their expertise on the judgment calls, the facilitation, the synthesis of conflicting stakeholder needs, and the organizational change management that determines whether strategies actually get executed.
I developed this methodology to test whether established consulting frameworks could be meaningfully accelerated through thoughtful AI integration. The answer is yes—but only if you understand the frameworks first, remain critical of AI outputs, and recognize that strategy is ultimately a human discipline that happens to have some automatable components.
If you're curious to try this approach yourself, I've provided enough detail here to replicate the full process. Start with a real strategic challenge your organization is facing, commit to working through all four phases rigorously, and see what happens. My guess is you'll find what I found: AI dramatically compresses the timeline, but human strategic thinking remains the irreplaceable element that determines whether the output is actually valuable.
And if you do try it, I'd genuinely like to hear how it goes.
Additional resources
For those interested in going deeper, here are the specific prompts I use for each phase:
Phase 1 - TOSCA Problem Definition Prompt:
Phase 2 - Issue Tree Generation Prompt:
You are a seasoned top strategy consultant with expertise in structured problem-solving.
For a strategic discussion, create a problem-breakdown structure in the form of an Issue Tree, adhering to the MECE (Mutually Exclusive, Collectively Exhaustive) principle.
The Issue Tree should:
Systematically break down a primary problem into sub-problems and further detailed branches, with a depth of 2-4 levels
Ensure that no overlaps exist between branches and comprehensively cover all potential areas relevant to the primary problem
Optimize the structure for a strategic discussion setting
Phase 3 - Prioritization Analysis Prompt:
Read each issue from the issue tree and consider the entire conversation we've had so far.
For each issue, estimate the benefit of following this option by assigning a number from 1 (low benefit) to 10 (highest benefit). The benefit is measured by the option's contribution to the defined success factors in the TOSCA framework. The effort is measured by the time and cost of pursuing this option.
Create a final output in a table with the following columns: Top-level issue, sub-issue, sub-sub issue, benefit (number between 1 and 10), arguments for choosing this number, effort (number between 1 and 10), arguments for choosing this number for effort, and the ratio of benefit to effort calculated by dividing the number of the benefit by the number of the effort.
Repeat the following process for each issue:
Read the issue
Consider the benefit. Attribute a number for the benefit
Consider the effort. Attribute a number for the effort
Once you've completed this for all issues, create a table in the format described above as the final output.
Phase 4 - Work Breakdown Structure Prompt:
You are an expert strategy consultant and project lead tasked with assisting in creating a work plan based on an issue tree that breaks down a complex problem into several levels, including a prioritized list of tasks for resolution.
Your objective is to develop a detailed work plan table that organizes the issue, sub-issues, tasks, and their corresponding details in a structured manner.
The work plan must be organized in a table format with the following columns:
Issue
Sub-issue
Task
Concrete To-Dos
Proposed Start Date
Duration
Proposed Finish Date
Dependencies to Other Tasks
Instructions:
Create the work plan step by step, considering each issue and sub-issue in the provided issue tree
Use affirmative directives to ensure that the work plan is detailed and actionable
Think step by step and consider the entire scope of the issue tree while prioritizing tasks appropriately
When working through each sub-issue and task, use a chain-of-thought approach to dive deeper and clearly articulate how tasks interlink with the broader problem
Now, start creating the work plan based on the provided issue tree.
Tools I use
ChatGPT (GPT-4 with Canvas): Primary LLM for all phases. Canvas feature allows precise editing of specific sections without regenerating entire outputs.
Miro: For visualizing and collaboratively refining issue trees. Their AI-powered mindmap generation and expansion features are surprisingly effective.
TeamGantt or Google Sheets: For converting WBS tables into visual Gantt charts that communicate project plans clearly to stakeholders.
The total cost for these tools is minimal compared to consulting fees—and the time investment for a complete strategic planning cycle is typically 2-3 weeks of part-time work rather than 3-6 months of full engagement.
