Market Sizing for International Expansion

Most expansion decisions I have seen go wrong do not fail at execution. They fail at the sizing stage, months before anyone books a flight or hires a local lead. Someone builds a slide with a very large number on it, the room nods, and the company commits budget to a market it never actually understood. The number was real in the sense that it existed in a spreadsheet. It was not real in the sense that anyone could ever capture it.

I work as a Growth Product Manager in Paris, and a good part of my job is figuring out which markets are worth entering and in what order. Sitting in Europe gives you a useful vantage point on this, because Europe is not one market. It is a dense cluster of countries that look similar on a map and behave completely differently once you try to sell into them. Payment habits, price tolerance, language, regulation, and the local competition change every time you cross a border. That reality forces a discipline that a single large domestic market lets you skip: you cannot hide behind one big number, because the big number is a fiction stitched together from many smaller ones.

This post is about doing market sizing honestly. Not the version that produces a comforting headline figure, but the version that tells you whether to go, where to go first, and how much you can realistically expect to earn if you do. Sizing well is not about precision. It is about being wrong in useful, bounded ways instead of confidently wrong in one direction.

Why bad market sizing kills expansion bets

The most common failure is over-optimistic TAM. Someone multiplies a population by an average revenue figure, arrives at billions, and treats that as the prize. The problem is that this number describes a market that would only exist if you had no competitors, no distribution costs, perfect localization, and every potential buyer deciding to pay you. None of those things are true. The gap between that headline and what you can actually reach is where expansion budgets go to die.

The second failure is ignoring the reachable market entirely. A market can be genuinely huge and still be closed to you. It may be locked up by an incumbent with exclusive distribution, gated behind a licensing regime you cannot satisfy for two years, or served by a local product so well adapted that your version feels foreign. Size on its own says nothing about winnability. Treating a large TAM as evidence that entry is a good idea is the single most expensive mistake I see, and it is expensive precisely because the number feels so authoritative.

Bad sizing does not just produce a wrong answer. It produces a confident wrong answer, which is worse, because it shuts down the questions that would have saved you. When the headline is big enough, nobody wants to be the person asking whether it is real.

TAM, SAM, and SOM explained honestly

The TAM, SAM, SOM framework is useful when you treat it as a set of honest reductions rather than a funnel of increasingly optimistic guesses. Here is how I actually use it.

TAM, the total addressable market, is the entire spending on the problem you solve, assuming everyone who could conceivably buy does. It is a ceiling, not a target. Its only real job is to tell you whether the category is large enough to bother with. Beyond that, quoting TAM as if it were achievable is where most decks go wrong.

SAM, the serviceable addressable market, is the slice you could serve given your actual product, business model, and the segments you can support. If you only sell to mid-market companies, or your product only works in a specific regulatory context, or you can only bill in certain currencies, SAM is what remains after you strip out everyone you cannot actually serve today. This is usually a fraction of TAM, and being honest here is uncomfortable, which is exactly why people avoid it.

SOM, the serviceable obtainable market, is what you can realistically win in a defined period given competition, your distribution reach, and how fast you can move. This is the number that should drive the decision, and it is almost always far smaller than people expect. A SOM that is a low single-digit percentage of a fragmented market in year one or two is a normal, healthy estimate, not a pessimistic one.

The discipline is to move down the ladder honestly at each step and to name the assumption that shrinks the number. If you cannot explain why SAM is smaller than TAM in one plain sentence, you have not done the work.

Top-down versus bottom-up, and why bottom-up wins

There are two ways to build a market estimate. Top-down starts from a big published figure, often an analyst report on the category, and carves out your share with a series of percentages. Bottom-up starts from the individual unit of demand, a customer or a segment, and builds up from what each one is worth and how many of them exist.

Top-down is fast and almost always misleading. The starting figure is someone else’s estimate, built for a different purpose, with definitions you cannot inspect. Then you apply a chain of percentages, and because each one feels reasonable on its own, the compounded guess feels defensible. It is not. You have layered your assumptions on top of a stranger’s assumptions, and you cannot pull the result apart to see where it breaks.

Bottom-up is slower and far more trustworthy. Every input is something you can name, defend, and pressure-test. How many businesses of the relevant type exist in this country? What share have the problem acutely enough to pay? What would they realistically pay, given local price levels? When you build the estimate this way, someone can challenge any single input, and you can respond with evidence or a clearly labelled assumption. That is what makes it useful for a real decision rather than a slide.

I use top-down only as a sanity check on a bottom-up number. If the two are within the same order of magnitude, I have some confidence. If they diverge wildly, one of them is built on a bad assumption, and finding out which is time well spent. The bottom-up build is the one I actually believe.

Building a bottom-up estimate

A bottom-up estimate is a chain of four honest questions. Get each one roughly right and the result will be more useful than any polished top-down figure.

Start with the population of the segment. Not the whole country, but the specific set of buyers who match your product. If you sell to independent retailers, count independent retailers, not all businesses. Government statistics offices, industry associations, and business registries usually get you a defensible base here, and this is the input you can pin down most firmly.

Next comes penetration, the share of that segment that has the problem badly enough to pay for a solution and is reachable given how the market buys. This is where honesty matters most, because it is tempting to assume everyone is a prospect. They are not. Some are served, some do not feel the pain, and some you cannot reach through any channel you can afford. A realistic penetration rate is often far lower than the first instinct.

Then willingness to pay, which is not your home-market price converted at the exchange rate. It is what this segment, in this country, will actually pay given local income levels and what they currently spend on alternatives. Anchoring to a substitute they already buy is far more reliable than anchoring to your existing pricing.

Finally price, the amount you would actually charge locally, which may differ from your headline price after discounts, local packaging, and the payment model the market expects. Multiply population by penetration by price, adjusted for how often they pay, and you have an obtainable revenue estimate you can defend line by line. The value is not the final figure. It is that every step is visible and arguable.

A big market is not a winnable market

This is the distinction that separates sizing that informs a decision from sizing that just decorates one. A market being large tells you the prize exists. It says nothing about whether you can take any of it.

Winnability depends on things that never show up in a TAM calculation. Competition is the obvious one: a large market defended by a strong local incumbent with better distribution and deeper local knowledge can be far harder to enter than a smaller market with no clear leader. I lean heavily on structured competitive intelligence that moves decisions here, because the shape of the competition is often the single biggest factor in whether a large market is actually open to you.

Distribution is the next constraint. If the way this market buys runs through channels you do not have and cannot easily build, the addressable market on paper is not addressable in practice. Regulation can close a market for years, or impose a cost of compliance that quietly destroys the unit economics. And localization is rarely just translation. It is payment methods, support in the local language, adapted onboarding, and sometimes a materially different product. All of that costs money and time, and it eats directly into the return.

I treat these as filters applied after the raw size, not footnotes. A smaller market you can genuinely win beats a larger one you cannot touch. The sizing exercise should end with a winnability view sitting next to the number, not a number standing alone.

Adjusting for local realities

The inputs that most often wreck an expansion estimate are the local ones, because they are easy to assume away from a desk in another country. This is the part where a European vantage point earns its keep, because the differences between neighbouring markets are impossible to ignore here.

Payment methods come first. In some markets cards dominate, in others bank transfers, direct debit, or local wallets carry most volume. If your billing does not support how a market actually pays, your conversion collapses regardless of demand, and the cost of adding local payment support belongs in the estimate. This connects directly to how you package and charge, which is why I think carefully about SaaS monetization models like credit and pay-as-you-go when the local buying pattern does not match a flat subscription.

Price sensitivity is next. The same product can command very different prices across countries with similar headline wealth, because of what local buyers are used to paying and what substitutes cost them. Assuming home-market pricing travels is a fast way to overstate revenue. Language and channel round it out. A market where you need local-language support and local content, sold through partners rather than direct, carries a cost to serve that has to be netted against the revenue, or the estimate flatters a market that is actually expensive to operate in.

The point is not to model every detail with precision. It is to make sure the obvious local realities are represented in the numbers rather than silently assumed to match home.

Data sources and how much to trust them

International sizing lives or dies on data, and the data is uneven. Knowing how far to trust each source is part of the craft.

National statistics offices and official business registries are the most reliable inputs for population and segment counts. They are slow to update and sometimes coarse, but they are grounded in real records, and I anchor the base of every estimate to them where I can. Industry associations and trade bodies are useful for segment structure and adoption context, though they carry a bias toward showing their industry as large and healthy, so I read their figures with that pull in mind.

Analyst reports and market research are useful for direction and for the top-down sanity check, but their absolute figures are built for a different purpose and rarely survive contact with a bottom-up build. I use them to check my order of magnitude, not to set my number. Primary research, actually talking to prospects, partners, and local operators, is the highest-value source for the softer inputs like willingness to pay and how the market really buys. It does not scale, but a handful of honest conversations will correct more bad assumptions than any report.

The habit that matters is labelling every input with where it came from and how confident you are in it. When the final number is challenged, you want to point straight to the weak link rather than defend the whole chain. The same discipline underpins serious due diligence for acquisitions, where an unverifiable market claim is a red flag rather than a selling point.

Ranges beat false precision

A single-point market estimate is almost always wrong, and its precision is a lie that invites bad decisions. I build every sizing as a range, with a low, expected, and high case, each driven by explicit assumptions on the inputs that matter most.

Sensitivity analysis is the tool here. Take your bottom-up model and flex the two or three inputs you are least sure about, usually penetration and willingness to pay, across a plausible band. Watch what happens to the result. If the market is attractive even in the low case, you have a robust bet. If it only works in the high case, you are betting on your most optimistic assumptions all landing at once, which they rarely do. That is a much more honest way to present a market than a confident single figure that hides where the fragility lives.

Ranges also change the conversation in the room. Instead of arguing about whether the number is right, people argue about whether the assumptions are reasonable, which is the argument you actually want to have. A tracked view of the market and its competitive shape over time, something like a competitive benchmarking dashboard, helps keep those assumptions honest as conditions change rather than freezing them at the moment the slide was made.

From sizing to a go, no-go, and sequence

Sizing is only worth doing if it drives a decision. The output I want is not a number but a recommendation: enter or do not, and if enter, in what order.

The go, no-go rests on the obtainable market in the realistic case set against the cost and risk of entry. If the SOM is large enough to matter even under conservative assumptions, and the winnability filters do not surface a blocker you cannot clear, that is a go. If the market only works in the optimistic case, or a regulatory or distribution wall stands in the way, that is a no-go or a wait, and saying so early saves far more than it costs.

Sequencing is where sizing pays off most, and it is rarely about picking the single biggest market. I look for the market that combines a decent obtainable size with the lowest cost and risk of entry, because the first market you enter teaches you how to enter the next one. A slightly smaller market that is close to home in payment habits, language, and regulation is often the right first move, because you build the playbook cheaply before spending it on a harder, larger target. This is the connective tissue between sizing and a real international market entry plan: the numbers tell you what is possible, the sequence tells you how to get there without betting everything on the hardest market first.

Common mistakes to avoid

A few failure patterns show up again and again, and naming them is the cheapest way to avoid them.

Spreadsheet theater is the first. A model with many tabs and precise-looking outputs feels rigorous, but complexity is not accuracy. If the key assumptions are guesses, more decimal places just make the guess look authoritative. I would rather have a simple model with three honest inputs than an elaborate one built on hopeful ones.

Anchoring on a huge TAM is the second, and the most seductive. Once a big number is in the room, it drags every later estimate upward and makes the obtainable figure feel disappointing by comparison. Lead with SOM, and keep TAM in its place as a ceiling.

Ignoring cost to serve is the third. Revenue is not the prize; margin after the cost of localizing, supporting, and distributing in the market is. A market that looks large on revenue can be unattractive once you account for what it costs to operate there, and that cost belongs in the model from the start, not as an afterthought once you have already committed.

The last mistake is treating the estimate as final. A market sizing is a snapshot of your best current understanding, and it should be revisited as you learn. The first real customers, the first competitive response, the first regulatory surprise all update the picture. Sizing that never gets revised is sizing that has stopped being useful.

The short version

• Bad market sizing kills expansion bets by producing a confident, over-optimistic TAM while ignoring the far smaller market you can actually reach.
• Use TAM as a ceiling, SAM as what you can serve, and SOM as what you can win; let SOM drive the decision.
• Build bottom-up, not top-down: population, penetration, willingness to pay, and price, with every input named and defensible.
• A big market is not a winnable one; filter for competition, distribution, regulation, and localization cost before you get excited.
• Adjust for local realities: payment methods, price sensitivity, language, and channels change the answer across borders.
• Anchor population data to official sources, treat analyst figures as a sanity check, and verify soft inputs through primary conversations.
• Present ranges with sensitivity analysis, not false precision, and check whether the bet holds in the conservative case.
• Turn sizing into a go, no-go and a sequence; enter the market that teaches you the most for the lowest cost first.
• Avoid spreadsheet theater, TAM anchoring, ignoring cost to serve, and treating the estimate as final.

---

I am Deepanshu Grover, a Growth Product Manager in Paris. If you are sizing a new market before betting real money on it, connect on LinkedIn or get in touch.