You walk into a boardroom armed with AI-generated market research that defines your total addressable market at $33.6 billion.

The board’s eyes light up. Investment flows. Strategic decisions get made.

Then reality hits. The actual market? $1.98 billion.

This isn't a hypothetical disaster; it's what happened when we put three leading artificial intelligence (AI) platforms head-to-head against human expertise in a market sizing challenge. The results were both fascinating and sobering.

The Great AI Market Sizing Experiment

At Strategex, we've been watching the AI revolution transform business intelligence with equal parts excitement and skepticism. Everyone's talking about AI's game-changing potential for market analysis, data processing, and strategic insights. But we wanted to know: can AI replace human expertise when it comes to the art and science of market sizing?

So, we designed a test.

We took five recent consulting projects with real clients, real markets, and real stakes. We ran them through three top-tier AI platforms: Gemini Pro, Unchained GPT, and AlphaSense. Each platform received identical prompts and was tasked with conducting market sizing using the industry-standard TAM/SAM/SOM framework.

The question wasn't whether AI could help with market research. We already knew it could. The question was whether it could replace the strategic thinking that separates good consulting from great consulting.

Where AI Shines (And Why We Love Having It on Our Team)

Before we get to the failures, let's give credit where it's due. AI impressed us in three crucial areas:

• Fast Data Aggregation: AI can efficiently scour public information, aggregating scattered data points from a wide range of sources. For our e-commerce software industry analysis, AI scoured dozens of industry reports, synthesized market data, and assembled a comprehensive foundation in minutes. What would have taken a junior analyst days, AI delivered before our coffee got cold.
• Extreme Efficiency: AI doesn't take lunch breaks, doesn't get distracted by emails, and never suffers from Friday afternoon fatigue. It cranks out preliminary TAM/SAM/SOM analyses with machine-like consistency, freeing our consultants to focus on strategic heavy lifting.
• Structured Foundation: AI excels at creating clear, logical frameworks. Every analysis yielded organized assumptions, structured methodologies, and clear presentations. It's like having a very thorough, very fast intern who never forgets to show their work. This provides an excellent starting point that our experts can immediately build upon.

These capabilities make AI a force multiplier for the data collection phase. But as we dug deeper into the analysis, the cracks began to show.

When AI Goes Off the Rails

The further we moved from broad market overviews to client-specific insights, the more apparent AI's limitations became. And nowhere was this more obvious than in the numbers themselves.

The TAM Train Wreck

The following table illustrates how AI’s Total Addressable Market (TAM) estimates were often inconsistent or overstated.

The e-commerce software “disaster” illustrates AI's fundamental flaw: it treats all merchants equally.

AI calculated average software fees and multiplied by total merchant count, ignoring the reality that Amazon pays vastly different rates than Mom's Craft Corner. This basic misunderstanding of market dynamics led to estimates that were off by over $30 billion.

AI’s most significant weakness was its struggle with narrow, specialty markets. For our Niche Packaging project, AI consistently returned a broad estimate for the general food packaging market. No matter how we refined our prompts, AI kept returning broad food packaging market estimates. It required repeated human intervention and clarification to narrow the scope; still, it simply couldn't grasp that our client operated in a highly specialized subset. It's like asking for the market size of artisanal cheese and getting back the entire dairy industry.

The SOM Catastrophe

If TAM discrepancies were concerning, the Serviceable Obtainable Market (SOM) estimates were downright alarming. The following table shows how AI's Serviceable Obtainable Market (SOM) estimates consistently failed to align with the client’s actual capabilities, demonstrating its inability to connect market data to a specific company’s reality. In most cases, AI significantly underestimated the SOM, actual capabilities, and willingness to invest.

Here's what went wrong: AI doesn't understand competitive advantage.

AI couldn't factor in our e-commerce client's proprietary technology, established partnerships, or superior go-to-market strategy. Instead, it assumed average capabilities and conservative growth. It completely missed the strategic opportunities that made the engagement valuable in the first place.

AI models could not incorporate the vital context of a client’s unique strengths, go-to-market strategy, or competitive positioning. As the table shows, for E-Commerce Software and Confectionery, AI’s SOM was less than a quarter of our calculated value. For Niche Packaging, the AI’s highest estimate was nearly nine times greater than the realistic SOM we identified. These disparities underscore AI’s inability to connect generalized market data to a specific company’s reality.

The Human Factor: What AI Can't Replicate

The additional weakness of AI, as highlighted by the estimates of SOM, is the inability to incorporate client capabilities or unique market position. Our experiment revealed four critical areas where human expertise remains irreplaceable:

Market Definition Mastery: While AI drowns in data, experienced consultants know exactly which data matters. We precisely define the market from the outset, ensuring that the TAM, SAM, and SOM are relevant to the client’s specific needs and not a generic, oversimplified estimate

Assumption Archaeology: We don't just accept data sources; we dig into them. When AI cited a market research report, we tracked down the methodology, questioned the sample size, and adjusted for known biases. AI takes sources at face value; we take them with healthy skepticism.

Strategic Translation: Numbers don't exist in a vacuum. Our consultants connect market size to competitive dynamics, client capabilities, and growth strategies. AI sees $357 million as a number; we see it as the difference between a small acquisition and a transformational market opportunity.

Methodological Agility: When top-down data is scarce, smart market researchers pivot to a bottom-up analysis. We conduct expert interviews, analyze channel-specific trends, and build new models. AI hits a wall; humans find another way.

The Future: Hybrid Intelligence, Not AI Replacement

Our experiment definitively answered the question: Can AI replace human expertise in strategic market analysis?

Not even close.

But that's not the right question anyway. The right question is: How can AI amplify human expertise to deliver better, faster insights?

The answer is hybrid intelligence—combining AI's computational power with human strategic thinking. AI is transforming consulting, but it's not replacing consultants. It's making the best ones even better.

The future belongs to firms that master this hybrid approach—leveraging AI's speed while preserving the strategic judgment that turns data into decisions. Ultimately, businesses don't succeed based on market size estimates. They succeed based on the strategic insights that guide their next move.

And that's still very much a human game.