Roughly four out of every five market entries fail. That figure comes from McKinsey’s long-running analysis of entry decisions, and it has held up well enough to become a standard reference point for strategy teams. The pattern behind the number is consistent across industries. Companies collect plenty of data. They aim it at the wrong question.
Entering a new country is not a single decision. It is a sequence of smaller ones: which markets to screen, which signals to trust, which costs to model, and how much to commit before the first real customer shows up. Market data is useful at each of those steps. It becomes dangerous when a team treats a dashboard as a verdict.
The decision gets framed before any data arrives
Most country research starts with a download spree. A researcher pulls population figures, GDP growth, and platform user counts, and a week later has a stack of context without a recommendation. The problem is rarely the data. The scope was never defined, so nobody knows which evidence matters or what action it should support.
A useful entry study starts with a written decision question. Something like: should we open a sales operation in Poland by Q2 next year, given a budget of X and a two-year break-even requirement? That sentence does more work than any report. It tells the team which numbers are load-bearing and which are decoration.
Three questions market data is built to answer
Once the question is framed, the evidence tends to organize itself around three dimensions. A 2026 overseas market entry framework from String Global lays them out cleanly, along with the warning signs that a team is fooling itself.
| Dimension | Core question | Warning signal |
|---|---|---|
| Market attractiveness | Is there sustained demand? | Demand traces to a single source, or growth cannot hold |
| Ability to win | Can we reach customers with a real difference? | Incumbent brands control the channels and customers have no reason to switch |
| Commercial viability | Can we operate compliantly and still earn a margin? | Market access is unclear, contribution margin is negative, or capability gaps cannot close |
A high score on one dimension does not rescue the others. That is the failure mode behind most confident, well-funded expansions that quietly stop growing. The team proved demand and never proved it could reach that demand at a profit.
From a long list to a short list
The screening stage is where macro data earns its keep. It is cheap, it is standardized, and it eliminates obviously bad markets before anyone books a flight.
- Population, income, and consumption: national statistics offices and the World Bank. These set the broad bounds of a market.
- Import volumes by product: UN Comtrade covers about 200 countries and more than 99% of merchandise trade, and the International Trade Centre’s Trade Map breaks flows down across 5,300 product categories down to six-digit HS codes.
- Tariffs and rules of origin: the ITC’s market access tools and the EU’s Access2Markets portal. These decide whether the landed price can compete at all.
The screening mistake I see most often is overweighting market size and underweighting regulatory complexity. A large market that demands a local entity, sector licensing, and domestic data storage can cost several times more to enter than a smaller market with a light compliance load. Size is the headline. Compliance is the invoice.
Demand signals that survive contact with reality
Macro figures describe the setting. They do not show that anyone is actively shopping. For that, teams look at three layers of evidence.
Search signals show what people are typing and comparing. Transaction signals show what is actually selling: prices, review volume, return complaints, rating drift on competitor listings. Customer signals come from interviews with recent buyers and from people who abandoned a purchase. The third layer is the one most teams skip, and it is usually the one that explains why a market that looked busy converts poorly.
Sizing follows the same discipline. Total addressable market, serviceable addressable market, and serviceable obtainable market form a funnel, and each layer should be more grounded than the one above it. Topping out a model with a top-down industry figure and claiming one percent of it is not a forecast. It is a wish with decimal places.
The cost side is where good entries get saved
Demand is the attractive half of the decision. The half that kills projects is the cost chain: landed goods, duties, fulfillment, returns, and customer acquisition in a market where you have no brand recognition. A rule that used to be a footnote has now become a line item.
On 1 July 2026, the EU ended its €150 customs duty exemption and replaced it with a temporary €3 customs duty per item on low-value consignments, set to run until 1 July 2028 under Council Regulation (EU) 2026/382. The European Commission counted almost 5.9 billion low-value parcels entering the bloc in 2025. Brands that built their European math on a duty-free threshold need to rebuild it. The same logic applies to data residency rules such as GDPR in Europe, LGPD in Brazil, and PIPL in China, where compliance shapes both cost and timeline.
What timing looks like in the numbers
Growth is uneven and it moves. FlavorCloud’s 2026 State of Cross-Border report found Switzerland growing 68% year over year in the first quarter of 2026, France 47%, and Singapore 43%, all faster than the highest-volume lane into Canada. A market that looked secondary two years ago can become the reason to move. That is the case for reviewing the data on a schedule rather than once.
Two companies that had the numbers and ignored the signals
The clearest lessons live in the failures. Target’s Canadian entry collapsed within two years despite an investment that has been estimated at $7 billion, a case documented in a 2026 Harvard Business School working paper on retail expansion. Walmart left Germany in 2006 after eight years and roughly $1 billion in losses, a retreat the New York Times described at the time as a template for how not to expand into a country.
Both companies had research. What they lacked was the discipline to act on it. Retail consultant Mark Satov, quoted in a 2025 case study of Target’s Canadian exit, put it bluntly: “They arrived with a teaching mentality rather than a learning mentality.” Walmart learned the German shopper preferred hard discounters and a shorter aisle, and its management structure kept pushing the American playbook anyway.
My read is that the number inside those case studies is not the loss figure. It is the length of time the company stayed. Target needed two years to leave a decision it could have stress-tested in ten weeks.
What market data cannot tell you
Treating data as an oracle has a real cost. Bart de Langhe and Stefano Puntoni argue in their work on decision-driven data analytics that the standard approach runs backward. Companies start with the data they happen to have and search for a purpose for it. The better method starts with the decision, then asks what evidence would actually discriminate between the options.
That argument matters for country entry because the data you can easily buy describes the past. Purchase intent in a market where you have never sold, the speed at which a partnership will close, and how quickly a local team can build trust are all qualitative. A spreadsheet will not resolve them. I think the honest position is that data narrows the field and prices the downside. It does not pick the winner.
Reading demand before you commit
Sequence is the quiet variable. A company that proves a narrow channel first can widen later, and the data from that first channel is far more valuable than any report. Gabriel Massuh built Bagno by importing bananas from Ecuador into Chile, then expanded the range to mangoes, lemons, and blueberries once the initial distribution channel was proven. The order of those moves carried as much information as the market research behind them.
The practical version is a staged test with a budget cap, a success threshold, and a stop condition set before it starts. A localized landing page, real price quotes, and a small batch of sales will tell you more about conversion in six weeks than a six-month modeling exercise.
What a go, hold, or no-go decision actually looks like
- Go: all three dimensions have strong evidence, no hard stop has been triggered, and the margin survives a stress scenario where costs rise and conversion drops.
- Hold: the opportunity is real, but a critical variable such as certification time, target price, or return rate remains unvalidated. A hold needs a named experiment, an owner, and a deadline.
- No-go: a hard stop on compliance, fulfillment, or margin holds. Walking away is a result, and it protects the capital that would have funded the next attempt.
Whichever way the call goes, the findings should be structured so a board or lender can follow the reasoning. Demand evidence, competitive positioning, regulatory diligence, and a financial model with base, optimistic, and stress cases. That pack is what turns a hunch into a defensible decision.
Frequently asked questions
How do companies decide which country to enter first?
They screen candidates on macro data such as population, income, internet penetration, and import volumes, then narrow the short list using regulatory complexity and cultural distance. The first market should be the one where demand is proven and the cost of entry is manageable, not necessarily the largest.
Is a large market always the best market to enter?
No. A large market can require a local entity, sector licensing, and data residency, which multiplies the cost of entry. A smaller market with lighter compliance can be the faster and more profitable first step.
What data should be checked before entering a new country?
At minimum: category import volumes from UN Comtrade or ITC Trade Map, tariff and rules-of-origin data, search and transaction signals from the target market, competitor pricing, and a landed-cost model that includes duties, fulfillment, returns, and customer acquisition.
How long does a proper market entry study take?
A structured study typically runs six to twelve weeks from defining the decision question to a final recommendation, with the middle weeks spent on interviews and small-scale validation. A desk-only version can finish in three to four weeks and carries less confidence.
Can market data guarantee a successful entry?
No. Research reduces uncertainty. It does not remove it. The goal is to eliminate the markets and the assumptions that are clearly wrong, so the remaining risk is one the company can fund and survive.
When should a company walk away from a market?
When a hard stop on compliance, fulfillment, or contribution margin is triggered, or when a stress scenario shows the business cannot break even within the stated timeframe. Killing an entry early is cheaper than exiting a live operation later.
How this article was put together
This piece set out to explain how companies use market data at the country entry decision, for founders and strategy teams planning cross-border growth. It draws on McKinsey’s entry failure research, a 2026 Harvard Business School working paper on retail expansion, the String Global 2026 market entry framework, the European Commission’s 2026 guidance on the EU low-value import duty, and 2026 cross-border data from FlavorCloud. Trade-data figures come from the International Trade Centre and UN Comtrade. Where a claim reflects judgment rather than research, it is marked as such. Duty rules and growth rates change often, so recheck the EU figures after each revision and treat 2026 growth rates as a snapshot.