Research-Driven Opportunity Sizing

“How big is this opportunity?” sounds like a simple question until you’re standing in front of your leadership team with three different answers from three different models, none of which feel particularly convincing.

Opportunity sizing isn’t really about arriving at a perfectly accurate number, it’s about building confidence that you’re solving a problem worth the investment.

The market’s littered with products built on optimistic Excel models that fell apart on contact with reality. The teams that get opportunity sizing right aren’t the ones with the fanciest spreadsheets. They’re the ones who know what they don’t know and design research to reduce that uncertainty systematically.

A Practical Framework

Step-by-step approach

The opportunity sizing framework that’s served me best has three layers: market reality check, user segment validation, and economics proof. Miss any layer and your sizing falls apart.

Start with market reality. Not TAM/SAM/SOM slides borrowed from analyst reports, actual validation that customers are experiencing the problem you think they’re experiencing.

The market reality check needs to answer: Are enough businesses experiencing this problem acutely enough to pay for a solution? You validate this through problem interviews, not solution pitches. I aim for double-digit conversations with the target segment before I’ll stake a sizing claim to leadership.

Second layer is user segment validation. Even if the market opportunity is large, can you realistically capture a meaningful portion? This is where most sizing models fall apart. They assume uniform distribution when reality is lumpy.

Segment validation requires: Can you reach these customers efficiently? Do they have budget authority and willingness to switch? Are there structural barriers (regulation, lock-in, integration complexity) that make capturing this segment harder than it looks?

Third layer is economics proof. Revenue opportunity means nothing if unit economics don’t work. I’ve seen too many “million-euro opportunities” where customer acquisition cost exceeded lifetime value by 3x.

The framework looks like this:

Market reality (weeks 1-3): Problem validation interviews, market structure mapping
Segment validation (weeks 3-5): Customer reachability assessment, willingness-to-pay research
Economics proof (weeks 5-6): Unit economics modelling, payback period analysis

Don’t run these sequentially unless you have to. Overlap them. The goal is getting to conviction fast, not conducting the perfect study.

Real examples from product teams

Time for a story: an enterprise software company wanted to size an opportunity in healthcare. Their initial model: 5000 hospitals × 50k EUR ACV = 250M EUR opportunity. Looked great on paper.

Research revealed a different story. Only 800 hospitals had the technical infrastructure to use their product. Of those, 200 were already contracted to competitors with multi-year deals. Of the remaining 600, only about 300 had budget allocated for this category of tooling in the next 18 months.

The actual addressable market? About 300 customers worth 15M EUR in annual bookings potential. Still a good opportunity, but requiring completely different resource allocation than a £250M TAM would suggest.

Here’s what they did right: they didn’t stop at top-down modelling. They mapped out specific hospitals, spoke to procurement teams, understood budget cycles and decision-making processes. The sizing was grounded in reality, not just multiplying big numbers.

Compare that to another company that sized their opportunity by taking “millennials with debt” (millions of people) × “average debt burden” (thousands of pounds) × “5% conversion rate” (made up). The model was nonsense. They had no validation of conversion rates, no understanding of whether their target segment would actually use the product, no sense of competitive dynamics.

Six months after launch, they’d acquired 2000 customers instead of the projected 50000. The opportunity was real, but an order of magnitude smaller than their model suggested. Investors were not amused.

The lesson? Bottom-up beats top-down every time. Find ten real customers who have the problem, will pay for a solution, and can be reached efficiently. Then model how many more customers like those ten exist and how you’ll reach them. That’s your opportunity size.

Common Pitfalls and How to Avoid Them

Mistakes to watch for

Confusing market size with opportunity size. The UK mattress market might be worth 1B EUR annually, but if you’re a new direct-to-consumer brand, your realistic opportunity in year one is probably closer to 5M EUR. Market size tells you the total prize; opportunity sizing tells you what you can actually capture.

I see this constantly with startups. They’ll pitch a “10B EUR market” when what investors actually want to know is: can you build a 50M EUR revenue business in three years? That’s the opportunity question.

Optimistic assumptions compounded. This is how you get opportunity models that assume 10% conversion × 20% adoption × 30% upgrade rate and arrive at numbers that never materialise. Each assumption introduces error, and multiplying optimistic assumptions compounds that error exponentially.

At one company, we modelled an opportunity assuming 5% of target businesses would trial the product, 40% of those would convert to paid, and 25% would upgrade within six months. Reality? 2% trial rate, 15% conversion, 10% upgrade. Our opportunity projection was off by 12x.

Ignoring structural barriers. A market can be huge, but if it’s protected by regulation, high switching costs, or network effects, your opportunity to capture share might be minuscule.

Like in insurance - imagine wanting to disrupt a 40B EUR market. What you would underestimate: insurance is heavily regulated, existing players have massive balance sheet advantages, and customers are extraordinarily sticky. The realistic opportunity isn’t disrupting incumbents; it is capturing growth in underserved segments. Completely different sizing.

Prevention strategies

Work backwards from specific customers. Don’t start with market stats. Start by identifying 10-20 specific organisations or people who you believe represent your target segment. Validate that they have the problem and would pay to solve it. Then, and only then, estimate how many similar customers exist.

This approach forces specificity. It’s much harder to handwave when you’re talking about “Sheffield Teaching Hospitals” rather than “the UK healthcare market.”

Build a sizing model with sensitivity analysis. Your model will be wrong. That’s fine. What’s not fine is having a single-point estimate with no sense of the range of outcomes. I always build three scenarios: aggressive (everything goes right), realistic (some things go right), and conservative (lots goes wrong).

Validate willingness to pay separately from problem validation. Just because someone has a problem doesn’t mean they’ll pay to solve it, or that they’ll pay what you need to charge to make economics work.

Early problem interviews should establish: Is this problem acute enough that customers are already spending time/money trying to solve it? If they’re living with the problem and not actively seeking solutions, you don’t have an opportunity. You have a vitamin, not a painkiller.

Then validate pricing explicitly. You can use Van Westendorp price sensitivity analysis: ask customers at what price the product becomes too expensive, too cheap (seems low quality), expensive but worth considering, and a bargain. This gave a realistic price range rooted in customer perception, not just cost structure.

Time-bound your opportunity. A 100M EUR opportunity over three years is very different from a 100M EUR opportunity over ten years. Investors and leadership want to know: what can we capture in the next 18-24 months with realistic resourcing?

I like to size opportunities in waves: Wave 1 (next 12 months, highest confidence), Wave 2 (months 12-24, medium confidence), Wave 3 (24+ months, speculative). This prevents conflating near-term and long-term opportunity, which require different strategies and resources.

Understanding the Fundamentals

Core concepts explained

There are three numbers that matter in opportunity sizing: Total Addressable Market (TAM), Serviceable Addressable Market (SAM), and Serviceable Obtainable Market (SOM).

TAM is the entire market demand for a product or service. If everyone who could possibly use your product did use it, that’s TAM. It’s mostly useful for understanding long-term potential and market dynamics, not for making near-term decisions.

SAM is the portion of TAM you can serve with your specific product and business model. If you’re building SaaS for mid-market companies, enterprise and SMB are outside your SAM even though they’re in your TAM.

SOM is what you can realistically capture given competition, resources, and go-to-market constraints. This is the number that actually matters for planning and investment decisions.

Most opportunity sizing exercises get these confused. They’ll calculate TAM and call it opportunity, or they’ll use SAM as a revenue target without considering competitive dynamics. The only number that’s truly useful for product strategy is SOM, and that requires the most validation to calculate.

Why this matters for PMs

Poor opportunity sizing has real consequences. Oversize the opportunity and you’ll over-invest, building for a market that doesn’t materialise. Undersize it and you’ll under-resource a genuine growth engine.

I’ve seen both. A marketplace platform oversized their opportunity, hired aggressively, burned through capital, and had to lay off 40% of staff when growth didn’t materialise. A SaaS company undersized their opportunity, under-invested in sales, and watched competitors capture the market whilst they were being conservative.

Good opportunity sizing gives you the confidence to invest appropriately. It helps you make the right trade-offs, allocate resources effectively, and set realistic expectations with stakeholders.

It also protects you from bad strategic decisions. If the research-validated opportunity is genuinely small, that’s valuable information. Maybe you pivot to a different segment, or maybe you decide this isn’t worth pursuing. Better to learn that early through research than late through failed execution.

Key Takeaways

Bottom-up beats top-down: Start with specific customers who have validated the problem and willingness to pay, then model how many similar customers exist. Don’t multiply big market numbers by conversion assumptions.
Layer your validation: Market reality check (is the problem real?), segment validation (can you reach and serve these customers?), economics proof (do unit economics work?). Miss any layer and your sizing is suspect.
Build ranges, not point estimates: Your sizing will be wrong. That’s fine if you understand the range of possible outcomes and have stress-tested the business case across scenarios.
Distinguish TAM/SAM/SOM: Total market size isn’t your opportunity. Your obtainable market—accounting for competition, resources, and GTM constraints—is what matters for decision-making.
Validate willingness to pay explicitly: Problem validation isn’t enough. You need evidence that customers will pay at price points that make your economics work.
Time-bound your opportunity: 100M EUR over three years is different from 100M EUR over ten. Size your opportunity in waves with different confidence levels.

Final Thoughts

The best opportunity sizing exercises I’ve been part of weren’t the ones with the most sophisticated models. They were the ones where the team genuinely understood their target customers, had validated the core assumptions with real research, and could articulate what they didn’t know.

Start with one target segment. Not “mid-market businesses” but something much more specific. Find 20 organisations in that segment. Validate that they have the problem and would pay to solve it. Map out how many similar organisations exist and how you’d reach them. That’s your initial opportunity size.

Then test it. Your first customers will tell you whether your sizing was directionally correct. Be prepared to update your model as you learn. The goal isn’t a perfect number at the start. It’s building a model that improves as you gather evidence.

This week, pick one opportunity on your roadmap. Can you name ten specific customers who’d be in your target segment? Have you validated they’d pay? Do you know how to reach them? If not, that’s where your research needs to focus.

Have questions or thoughts? Get in touch - I’d love to hear from you!