Walmart entered Germany in 1997, ran 85 stores, and walked out in 2006 having lost an estimated billion dollars. It wasn't beaten on price or logistics, it was beaten by the gap between how Bentonville thought retail worked and how Germans actually shopped. That gap is the whole subject of market-entry strategy: not whether to expand, but how to cross a border without assuming the other side is a copy of home.

The quick version

  • Two decisions, not one. Pick an entry mode (how much you own and commit, from exporting to a wholly owned subsidiary) and a localisation level (how much you adapt the product, brand and operating model). They're separate calls and you make both.
  • Distance is the diagnosis. Ghemawat's CAGE framework, Cultural, Administrative, Geographic, Economic distance, tells you where a market is genuinely far from home, and far is where plans quietly die.
  • Commitment should track knowledge. The Uppsala model says firms internationalise in steps because experience, not capital, is the scarce input. Big bets into markets you barely understand are how you become a case study.
  • Localise where it changes the buying decision; standardise everywhere else. Adapt for real local differences, not to feel busy. The cost of adapting the wrong things is as fatal as failing to adapt the right ones.

The idea in depth

Most market-entry failures aren't strategy failures in the boardroom sense. They're a refusal to believe that the new market is different in ways that matter. So the discipline starts with measuring difference honestly, then choosing a mode and a degree of localisation that match it.

Distance still matters, measure it before you map it

The instinct in the 2000s was that the world had gone "flat", that technology and trade had erased the friction between countries. Strategy researcher Pankaj Ghemawat pushed back hard. In "Distance Still Matters" (Harvard Business Review, 2001) he introduced the CAGE framework: the distance between any two countries has four dimensions, Cultural (language, values, norms), Administrative (legal systems, trade policy, currency, corruption), Geographic (physical distance, climate, infrastructure), and Economic (wealth, consumer behaviour, cost of inputs). His later data, summarised in his 2012 TED talk, found that most economic activity is far more domestic than the "flat world" story assumes, only a low single-digit percentage of phone-call minutes, immigration and direct investment is genuinely cross-border.

So before building a market-entry plan, score the target market on all four CAGE dimensions against your home base and find the one or two where the distance is large. That's where your assumptions will break. A British firm entering Australia faces low cultural and administrative distance but real geographic distance; the same firm entering Japan faces low geographic distance (by air freight, comparable to the US) but high cultural and administrative distance. Spend the plan's attention where the distance is, not evenly across the board.

The honest limitation: CAGE diagnoses where difference lives; it does not tell you what to do about it, and it can give a false sense of precision. There is no agreed formula that converts "high cultural distance" into a decision. Treat it as a structured way to ask better questions, not a score that makes the choice for you.

flowchart TD
    A(["Target market"]) --> B(["Score CAGE distance vs home"])
    B --> C(["Cultural"])
    B --> D(["Administrative"])
    B --> E(["Geographic"])
    B --> F(["Economic"])
    C --> G(["Where is distance largest?"])
    D --> G
    E --> G
    F --> G
    G --> H(["Localise & commit cautiously there"])
    G --> I(["Standardise & move faster elsewhere"])
CAGE turns "this market feels different" into four testable dimensions, then points your effort at the one or two that are genuinely far. Leaders Loop

Choose an entry mode by what you're willing to lose

Entry modes sit on a ladder of commitment, control and risk. At the low end, exporting, you make at home and sell abroad, cheap and reversible, but you're at arm's length from the customer. Next, contractual modes: licensing and franchising let a local partner use your brand or IP for a fee, fast and capital-light, but you hand over day-to-day control and risk diluting the brand. At the high end, equity modes: a joint venture shares cost, risk and local knowledge with a partner (sometimes the only legal route in), while a wholly owned subsidiary, built greenfield or bought through acquisition, gives you full control and full exposure. This export-to-equity ladder is the standard spine of the international-business literature (see, for example, the International Business open text on expansion entry modes).

The deeper logic for climbing that ladder gradually comes from Jan Johanson and Jan-Erik Vahlne's Uppsala internationalisation model (first published 1977; revisited in the Journal of International Business Studies, 2009). Their claim: firms expand in increments because the binding constraint is experiential knowledge of the market, which can only be acquired by operating in it. So companies start in psychically close markets (those that feel familiar) with low-commitment modes, then deepen commitment as knowledge grows. The 2009 revision reframed the real barrier as the "liability of outsidership", being outside the relevant local network matters more than raw cultural distance.

The practical rule: match your entry mode to how much you actually know about the market, not to how excited the board is. If your read on local demand, regulation and distribution is thin, buy yourself an option, export, license, or take a minority JV stake, and earn the right to commit harder. If you already operate next door and understand the customer, a fuller-control mode can be the cheaper long-run choice, because licensing leaks both margin and control.

Where it breaks down: the staged Uppsala path can be too slow. In digital and platform businesses, waiting to accumulate local experience can simply hand the market to a faster rival, and "born-global" startups routinely skip the ladder. Use the model as a default that assumes ignorance is costly, then override it deliberately, with eyes open, when speed genuinely beats caution.

flowchart LR
    A(["Exporting
low cost · low control"]) --> B(["Licensing /
Franchising"])
    B --> C(["Joint venture
shared risk & knowledge"])
    C --> D(["Wholly owned subsidiary
greenfield or acquisition"])
    D --> E(["Full control ·
full exposure"])
The entry-mode ladder: each rung trades more commitment and risk for more control and a larger share of the upside. Leaders Loop

Localisation: adapt, aggregate, or arbitrage

Once you're in, the running question is how much to change. The cleanest map is Ghemawat's AAA framework from Redefining Global Strategy (2007), which names three ways to create value across borders. Adaptation adjusts the offer to local tastes and rules (the localisation most people mean). Aggregation groups markets to capture scale and standardise, a regional product, a shared platform. Arbitrage exploits the differences themselves, making where it's cheap, selling where it's dear. Most firms can't max all three at once; the art is choosing the dominant posture per market.

This sits on an older idea: C.K. Prahalad and Yves Doz's integration–responsiveness framework (The Multinational Mission, 1987), which holds that every international business is pulled between pressure for global integration (efficiency, consistency) and pressure for local responsiveness (fit to local demand). Where you land on that tension is, in practice, your localisation decision.

Here the rule flips around: localise where local difference changes the buying decision, and standardise relentlessly where it doesn't. Language, payment methods, legal compliance, and anything tied to a strong local norm usually need adaptation; back-end systems, core engineering and brand fundamentals usually don't. Walmart got the customer-facing layer wrong in Germany, greeters, store format, labour relations, while Aldi won on a stripped-back model that fit what German shoppers expected. The point isn't "always localise." It's localise the few things that decide the sale.

The honest limitation: over-localising is a real and underrated failure mode. Every adaptation multiplies cost and complexity, fragments your supply chain, and erodes the scale advantage that justified going global at all. "Adapt everything" is not caution, it's a slow way to lose money.

A worked example

A mid-sized UK software company, call it Northwind, an illustrative composite, all figures invented to show the method, sells HR-and-payroll software to small firms. It wants its first overseas market. The team shortlists Ireland, Germany and Singapore and runs a rough CAGE read.

Ireland scores low on every CAGE dimension, same language, similar law, next door, comparable economy. Germany is geographically close but administratively distant: payroll software is wired into national tax and employment law, so the product needs deep local compliance work, and cultural distance around data privacy is real. Singapore is far on every axis and small.

Using Uppsala as a default, Northwind starts where knowledge is cheapest to acquire: Ireland, via a light-touch mode, a small sales team selling a near-identical product (high integration, low adaptation), because the buying decision barely differs from home. Say this reaches an illustrative £400k of recurring revenue in year one at modest cost. That success would not justify copy-pasting the playbook into Germany, where the CAGE read says administrative distance is the whole game: payroll rules are the product. So the German move is a higher-adaptation entry, likely a joint venture or local hire with payroll-compliance expertise, accepting slower revenue (an illustrative £150k in year one) to get the one thing right that decides every sale. Singapore waits until there's an experienced international team to send.

The lesson the numbers dramatise: the right entry mode and localisation level are different for each market on the same shortlist, and the deciding factor is where the distance is, not which market looks biggest on a spreadsheet.

Frequently asked questions

What's the difference between entry mode and localisation?

Entry mode is the legal and financial structure of how you go in, export, license, franchise, joint venture, or wholly owned subsidiary, and it sets your level of control, commitment and risk. Localisation is how much you change the product, brand, pricing and operating model once you're there. You decide both, and they don't move together: you can run a wholly owned subsidiary selling a near-identical global product, or license a heavily localised one.

What's the lowest-risk way to enter a new country?

Exporting and licensing are the lowest-commitment, most reversible options, you risk less capital and can withdraw cheaply, which is exactly why the Uppsala model treats them as the natural first rung when your local knowledge is thin. The trade-off is less control and a smaller share of the upside. "Lowest risk" and "highest return" sit at opposite ends of the ladder; the skill is matching the rung to what you actually know.

How do I decide how much to localise?

Adapt where local difference changes the buying decision, language, legal compliance, payment methods, anything tied to a strong local norm, and standardise everywhere the customer can't tell or doesn't care. Ghemawat's AAA framework is a useful filter: ask whether each market is mainly an Adaptation play (fit local taste), an Aggregation play (scale across similar markets), or an Arbitrage play (exploit cost differences). Over-localising quietly destroys the scale advantage that made global expansion worth doing.

Is "go global fast" ever right, given the staged Uppsala model?

Yes. Uppsala describes a sensible default, commitment should track knowledge, but it isn't a law. Digital, platform and network-effect businesses often face a land-grab where a slow, incremental entry hands the market to a faster competitor, and "born-global" startups internationalise from day one. Treat staged entry as the prudent baseline and override it deliberately when winner-takes-most dynamics make speed the bigger risk to manage.

Why do well-run companies still fail abroad?

Usually because competence at home becomes overconfidence abroad, the assumption that what worked here will work there. Walmart's exit from Germany is the textbook case: a world-leading operator lost roughly a billion dollars partly by importing US store formats, labour practices and customer-service rituals into a market that read them as strange. The failure wasn't capability; it was refusing to believe the distance was real.

Related in the Toolkit

Where to go next