App Market Expansion: How to Choose New Countries and Enter Them Profitably

App market expansion is the process of selecting new countries for a mobile app, entering them with localized store presence, creative and pricing, and proving the unit economics in each one before scaling spend. It is a sequencing problem, not a translation problem. Most teams treat it as the latter, switch on a handful of new geos inside an existing campaign, watch the blended numbers get worse, and conclude that the markets were bad. The markets were usually fine. The entry sequence was wrong.

Admiral Media has run this sequence for apps expanding out of a single home market and into eight-language operations, across brain training, meditation, dating and mobility categories. This guide sets out how the Admiral Media team decides which markets are worth entering, in what order to localize, and what evidence to demand before a new geo gets a real budget. Every metric cited below is drawn from a published Admiral Media case study or a named third-party source, and linked so you can check it.

Why app market expansion usually destroys efficiency before it creates growth

Expansion destroys efficiency when a new market is added to an existing campaign rather than treated as a separate business with its own economics. The mechanism is straightforward and algorithmic. A campaign optimizing toward a conversion or value target learns from the aggregate signal it receives. When a low-cost, low-value geography is added to a campaign that was previously trained on a high-value one, the auction system does exactly what it was told to do: it buys the cheapest units that satisfy the event definition. Install volume rises, cost per install falls, and revenue per user collapses. The dashboard looks like a win for about three weeks.

The second failure mode is partial localization. An app that translates its store listing but not its onboarding, or its ad copy but not its paywall, creates a break in the experience precisely where intent is highest. Google’s Play Console documentation is explicit that if you add text translations without localized graphic assets, the app’s graphic assets show from the default language, which means a Spanish-language store listing can still be selling with English screenshots unless you deliberately fix it.

The third failure mode is measurement. A new market that has been live for four weeks has almost no cohort maturity. Judging it on day-zero return on ad spend punishes exactly the markets where the payback curve is longest, which are frequently the markets with the best long-term value. Admiral Media treats a new geo as unjudgeable on revenue until it has enough cohort depth to be compared against the home market at the same cohort age, a discipline covered in more depth in the Admiral Media guide to CAC payback period for mobile apps.

The scale of the opportunity is real, which is why the sloppiness is expensive. Apple reported that the App Store saw over 850 million average weekly users spanning 175 countries and regions. Almost no app monetizes more than a handful of them properly.

The Admiral Media Market Entry Ladder

The Admiral Media Market Entry Ladder is a six-rung sequence that moves an app from an unproven market to a fully localized, separately budgeted operation, with a stop-or-continue decision at every rung. The point of the ladder is that each rung costs more than the last, so no team should climb a rung until the previous one has produced evidence. Admiral Media applies this ladder whether the target is a single adjacent market or a multi-country rollout.

The Admiral Media Market Entry Ladder

1. Qualify on economics, not on size. Screen candidate markets on the ratio between acquisition cost and realistic lifetime value in that market, not on population or total app revenue. A market where cost per install is a fifth of your home market is only attractive if value per user clears the cost. Total addressable market is a vanity input at this stage. Disqualify any market where you cannot legally take payment, support the language, or meet local data requirements, before you look at any performance number.
2. Mirror before you localize. Run a deliberately small, isolated campaign in the candidate market using your existing proven creative, in its existing language, with your existing store listing. This is a read, not a growth attempt. It answers one question: does demand exist here at any price, and how far off is the cost curve. Isolate it in its own campaign so it cannot contaminate the learning signal of the home market.
3. Localize the surface. Fix the store listing first, because it is the cheapest lever with the widest reach and it improves paid and organic acquisition simultaneously. Localized metadata, localized first screenshot, localized app name and subtitle. This rung reliably moves organic install volume before a single euro of extra media spend is committed.
4. Localize the message. Move the ad creative into the local language. This is where most budget decisions are actually made, and where the choice between adapting proven creative and commissioning native production has the largest cost implications. Rung four is covered in detail below.
5. Localize the offer. Local currency, local price points, local payment methods, and a paywall written in the local language. An app that has climbed rungs three and four but not five is spending efficiently to deliver users to a checkout they do not trust. The mechanics of that checkout are the subject of the Admiral Media guide to app paywall optimization.
6. Institutionalize. Give the market its own campaign structure, its own target return on ad spend, its own creative pipeline and its own line in reporting. A market that shares a budget with the home market will always lose the budget fight during a bad week, regardless of its long-term value.

The rungs are ordered by cost per unit of information. Rung two costs almost nothing and rules out most candidate markets. Rung five costs engineering time and finance approvals. Teams that invert this order spend heavily localizing an app for a market that a cheap, isolated mirror test would have eliminated in a fortnight.

How to score and prioritize candidate markets

Market prioritization should be a weighted score across economics, operational cost and strategic fit, resolved into a single ranked list. Admiral Media uses a scoring matrix rather than intuition because expansion decisions are usually made by whoever argues most confidently in the meeting, and confidence is uncorrelated with market quality. The matrix below is the decision framework the Admiral Media team applies before any spend is committed.

Two notes on using this matrix. First, weight is not fixed. For a subscription app, payment readiness dominates. For an ad-monetized app, it is nearly irrelevant and channel availability moves up. Second, the disqualifier column matters more than the score. A market that scores well on six criteria and trips one disqualifier is not a good market, it is a project.

Localized ASO is the cheapest rung on the ladder

Localized app store optimization is the highest-return step in app market expansion because it changes organic acquisition and paid conversion rate at the same time, for a fixed one-time cost. Store metadata is not media spend. You pay for the keyword research and the translation once, and the ranking improvement compounds for as long as the metadata stays competitive.

The mechanics are worth understanding precisely, because they are not intuitive. Apple’s App Store Connect documentation explains that if you localize app metadata in French, that localization displays to users whose language setting is French, and also to users in countries or regions where the App Store supports French but not English, and that users can search using localized keywords in all countries or regions where the App Store supports French. In other words, a single language localization does not serve a single country. It serves a language footprint. Apple also documents that when you add a language, screenshots and other properties default to the primary language except for the description and keywords, which is the exact trap that produces half-localized listings.

Admiral Media managed localized ASO for NeuroNation in the South Korean market, rebuilding the app’s Korean metadata around the structure of the Hangul writing system rather than translating English keywords, and achieved a 93.10% increase in search downloads and a 43.95% increase in search impressions. The keyword ranking distribution shifted at the same time, which is the leading indicator that explains the download result.

Two details from that engagement generalize. First, Admiral Media had no native Korean speaker on the team and still produced the result, by studying how Hangul composes characters into search terms and validating candidate phrases systematically rather than translating word for word. Localized ASO is a research discipline before it is a language discipline. Second, the same case study notes that the South Korean app market grew by 1.5 billion dollars in revenue year on year and recorded an estimated 1.9 billion app downloads in 2023, up from 1.8 billion the year before, which is the kind of market-level evidence that belongs in rung one of the ladder rather than rung four.

Teams running this rung should also decide their translation procurement route deliberately. Google Play offers a paid human translation service for store listings and in-app products costing as little as USD 0.07 per word with translations completed within seven days, alongside a free machine translation option that Google explicitly flags as not reviewed or approved by humans. For metadata that determines your ranking, the seven-day human route is almost always the correct spend. Admiral Media’s ASO team treats machine translation as a research input, never as shipped metadata.

Creative localization: adapt what works or commission native production

Creative localization is the rung where most expansion budgets are won or lost, and the default assumption that a native local creator will outperform an adapted version of your proven creative is not reliably true. This matters because native production is slower and more expensive per asset, so the assumption sets the entire cost structure of the expansion.

Admiral Media tested this directly. Inshallah, a dating app for the Muslim community with over five million users worldwide, had proven creative in the United States and wanted to enter the French market. The Admiral Media team took the winning US creator videos, cloned the creator’s voice and translated the speech with lip sync using AI video tooling, adapted the ad copy under character and meaning constraints, and ran the result head to head against a creative made by a native French creator on Facebook Ads over a ten-day test. Converting one video from English to French took under thirty minutes.

Admiral Media translated Inshallah’s proven US creator videos into French using AI voice cloning and lip sync, and the translated creative delivered installs at a cost per install 15% cheaper than the native French creator ad, which recorded a CPI of 2.52 euros. The translated creative absorbed 77% of total ad spend and produced 75% of total installs as the algorithm allocated toward it.

The reason this works is not that AI creative is inherently better. It is that a creative concept that has already survived thousands of impressions in one market carries proven structural information: the hook timing, the objection order, the proof sequence. A new native production starts from zero on all of it. Translation preserves the proven structure and localizes only the surface. Commissioning a native creator resets the structure and gambles on a new one.

The opposite case is equally instructive. Admiral Media scaled Petit BamBou, a meditation app used by more than nine million people with over two thousand guided sessions in six languages, on TikTok using creator-led Spark Ads, reducing CPI by 57% and increasing the conversion rate from clicks to app installs by 170%, with the winning creative running just ten seconds and the campaign accumulating over 40.6 million views. The team then took the pattern that worked in France and rebuilt it in Spain with a local Spanish creator on an app-events optimization campaign.

Admiral Media rebuilt Petit BamBou’s proven French creative pattern in Spain with a native Spanish creator, and the Spanish execution delivered a 16% decrease in cost per click, a 21% increase in conversion rate and a 14% decrease in cost per acquisition compared with the account’s other Spanish-language ads.

Read the two cases together and the rule falls out. What transfers across markets is the creative pattern, not the asset and not the person in it. Sometimes the cheapest way to move a pattern is to translate the existing asset. Sometimes it is to brief a local creator against the pattern. The mistake is to move neither, and to let a new market invent its own creative strategy from scratch. Admiral Media’s approach to building and reusing those patterns is set out in the Admiral Media creative testing framework.

Choosing how deep to localize, and when

Localization depth should escalate with proven demand, not with ambition. Localizing everything on day one is the most common way to spend a quarter’s budget on a market that was never going to work. The table below is the depth ladder Admiral Media uses, mapped to the rung of the Market Entry Ladder that triggers it.

The most common error is jumping from level 0 to level 4 because a stakeholder wants the market to work. The second most common is stopping at level 1 and concluding the market is weak, when the store listing was converting fine and the paywall was the problem.

Structuring campaigns and budgets for a multi-market operation

Every new market needs isolated campaign structure, its own conversion signal and its own target, or it will be optimized against by the markets it shares a campaign with. This is the operational heart of app market expansion and it is where the algorithmic reasoning matters most.

Modern bidding systems, whether target return on ad spend, target cost per action or value-based bidding, resolve a single objective across whatever inventory they can reach. Give a campaign a mixed geography with a wide spread in cost per install and an even wider spread in value per user, and the system will find the cheap end unless the value signal it receives is accurate enough to stop it. In practice the value signal in a new market is the least reliable signal you have, because cohort volume is thin, modeled conversion values are barely trained, and on iOS the privacy-preserving measurement layer needs meaningful conversion volume before postbacks carry usable information. Isolation is not tidiness. It is the only way to prevent a young market from being priced by an old one.

Budget sequencing follows the same logic. Admiral Media managed TIER Mobility’s expansion from a local operation to a multi-market one, auditing the existing Facebook campaigns, building a shared ad copy database in eight languages with a client-side brand approval process, adding two new channels alongside Facebook, and moving optimization from installs down-funnel to sign-ups. The result was a 297% increase in new customers, two new channels added alongside the existing one, and acquisition budget scaled 5x in less than three months.

Three things in that sequence are worth copying. The multi-language copy database is the underrated one: eight languages of approved ad copy held centrally means a new creative concept ships into every language in a single pass, and the brand approval bottleneck is solved once rather than renegotiated per campaign. Moving optimization down-funnel from installs to sign-ups is the second: in a new market, install optimization buys you the local users least likely to convert, and the correction is to give the system a deeper event as soon as volume permits. Adding channels only after the first channel was working is the third. Channel diversification is a scaling tactic, not an entry tactic.

Channel efficiency itself varies enormously by market and inventory source, which is why the mirror test at rung two should not be treated as a verdict on the market if it ran on a single channel. Admiral Media ran a head-to-head channel test for the dating app PURE on US Android user acquisition, splitting distinct budgets between an established self-attributing network and a programmatic demand-side platform with creative tailored to each, and the programmatic route delivered a CPI of 2.44 dollars against the self-attributing network’s 9.43 dollars, a 74% reduction in CPI that led to expanded market launches on the strength of the performance. The same spread can exist between channels inside a new geography, and a market written off on one channel is frequently a market that was never properly tested.

Measuring a new market without fooling yourself

A new market must be judged on cohort-aged unit economics compared against the home market at the same cohort age, never on blended account metrics or on early return on ad spend. Blended metrics are the specific enemy of expansion, because a new market is by definition a small, inefficient share of a large, efficient account, and it will always look like a drag in the aggregate.

The practical measurement stack for a new market has four layers. Cost per install answers whether you can buy attention here. Install-to-activation rate answers whether your localization is working, and it is the fastest signal you have that levels 1 through 3 of the depth table are adequate. Activation-to-purchase rate answers whether the commercial layer is right, and a weak number here usually indicts the paywall and pricing rather than the traffic. Cohort revenue at a fixed age, compared against the home market at the same age, answers whether the market is worth institutionalizing.

Two disciplines make this stack usable. First, hold the comparison age constant. A day-seven number for a new market compared against a day-ninety number for the home market is not analysis. Second, report new markets separately from the moment they launch, alongside blended figures rather than inside them. The Admiral Media guide to blended ROAS and marketing efficiency ratio covers how to hold both views at once without letting the blended figure make decisions it is not qualified to make. For category-level reference points on where costs and retention typically land, the Admiral Media mobile app marketing benchmarks are the starting point, though a benchmark is context for a decision and never a substitute for your own cohort data.

The mistakes that cost the most

The most expensive app market expansion mistakes are structural rather than tactical, which is why they survive so long inside otherwise competent teams. Five recur consistently in accounts Admiral Media inherits.

Adding geos to an existing campaign. Covered above, and still the single most common one. It produces a cheaper blended CPI, a worse blended value per user, and a set of conclusions about market quality that are entirely artifacts of the campaign structure.

Chasing low cost per install into markets that cannot monetize. A low acquisition cost is only meaningful relative to local value per user. Expansion decks that rank markets by CPI ascending are ranking markets by how poor the users are.

Localizing the ad but not the destination. Paid traffic in the local language landing on an English onboarding flow converts badly, and the resulting numbers get blamed on the creative or the market rather than on the break in the funnel.

Judging too early. A four-week-old market has no cohort maturity and no trained value signal. Killing it on day-zero return on ad spend systematically eliminates the markets with the longest and most valuable payback curves.

Running three markets at once. Parallel launches fragment creative attention, budget and analytical capacity, and produce three underpowered tests instead of one conclusive one. Admiral Media’s default is sequential entry, with the next market opening only once the previous one has cleared rung five.

How Admiral Media runs app market expansion

Admiral Media runs app market expansion as a staged evidence process rather than a launch project. The Admiral Media team has managed over 500 million euros in ad spend for more than 150 mobile brands, holds a 5.0 rating on Clutch, and maintains a presence across 35 markets, which means the expansion questions above are ones the team answers with account history rather than with theory.

In practice, an Admiral Media expansion engagement starts with the rung-one economic screen and the rung-two mirror test, because those two steps eliminate most candidate markets for a fraction of the cost of finding out later. Localized ASO usually follows, since it moves organic and paid conversion together for a one-time cost. Creative localization is then run as an explicit test between adapting proven assets and commissioning native production, rather than as an assumption. Only once a market clears its own targets does it get institutionalized with its own budget line and its own creative pipeline. Teams that want the same sequence applied to their own account can talk to the Admiral Media team.

Frequently Asked Questions

What is app market expansion?

App market expansion is the process of taking a mobile app into new countries by selecting markets on their economics, localizing the store listing, creative, in-app experience and pricing, and validating unit economics in each market before scaling spend. It is distinct from simply enabling additional geographies in an existing ad campaign, which mixes markets with different costs and different user values into one optimization signal. Proper expansion treats each new market as a separate business with its own campaign structure, its own targets and its own reporting line. Admiral Media sequences this work through a six-rung Market Entry Ladder so that each stage of investment is justified by evidence from the previous one.

How do I choose which country to expand my app into next?

Score candidate markets on the ratio between realistic local value per user and observed local acquisition cost, then adjust for payment readiness, localization depth required, organic search headroom, creative transferability and compliance load. Do not rank markets by population or by low cost per install, because a cheap install in a market that cannot monetize is a loss you pay for twice. The most efficient way to get real inputs is a small isolated mirror test in the candidate market using your existing creative and store listing, which gives you an observed cost curve for a small fraction of a real launch budget. Any market that fails a hard disqualifier, such as unsupported payments or a regulatory requirement you cannot meet, should be removed before scoring.

Is app localization worth it, or is English enough?

Localization is worth it in almost every market where you intend to spend meaningful budget, because it improves organic discovery and paid conversion at the same time for a one-time cost. Apple’s documentation notes that a localization is shown to users whose language setting matches, and also to users in countries or regions where the App Store supports that language, so a single language localization serves a language footprint rather than one country. Admiral Media’s localized ASO work for NeuroNation in Korea produced a 93.10 percent increase in search downloads and a 43.95 percent increase in search impressions. The correct question is not whether to localize but how deep to go, and depth should escalate with proven demand rather than with ambition.

Should I use AI translation or hire a native creator for a new market?

Test both, because the assumption that a native creator wins is not reliably true. In Admiral Media’s ten-day head-to-head test for Inshallah in France, an AI-translated version of the app’s proven US creator video delivered installs 15 percent cheaper than a creative made by a native French creator, which recorded a cost per install of 2.52 euros. The translated asset preserves the proven structure of a creative that already survived thousands of impressions, whereas new native production resets that structure. That said, Admiral Media’s work for Petit BamBou showed a native Spanish creator briefed against a proven French pattern producing a 21 percent higher conversion rate and 14 percent lower cost per acquisition than other Spanish ads, so the right answer depends on how well the pattern survives translation.

How long should I wait before deciding whether a new market is working?

Wait until the market has enough cohort maturity to be compared against your home market at the same cohort age, which for a subscription app usually means at least one full billing cycle plus a renewal window. Judging a four-week-old market on day-zero return on ad spend systematically eliminates markets with long but valuable payback curves. In the interim, use leading indicators that mature faster: install-to-activation rate tells you whether your localization is adequate, and activation-to-purchase rate tells you whether the commercial layer is right. Only cohort revenue at a fixed age should decide whether a market gets institutionalized.

Should I launch in several markets at once to move faster?

Sequential entry produces better results than parallel launches for almost every team below enterprise scale. Running three markets simultaneously splits creative attention, budget and analytical capacity three ways, and typically yields three underpowered tests rather than one conclusive one. Admiral Media’s default is to open the next market only once the previous one has cleared the offer-localization stage and is hitting its own targets. The exception is a set of markets that share a language and a creative pattern, where the marginal cost of adding the second and third market is close to zero.

How do I stop a new market from ruining my existing campaign performance?

Give the new market its own campaign, its own conversion signal and its own target from the first day of spend. Bidding systems optimize toward the objective they are given across all inventory they can reach, so a campaign containing both a high-cost, high-value market and a low-cost, low-value one will buy the cheap end and report an improving cost per install alongside a deteriorating value per user. Isolation also protects your existing market, since a young geography with a thin and poorly trained value signal should never be allowed to influence the bidding of a mature one. Report the new market separately alongside blended figures rather than inside them, so the blended number never quietly makes a decision it is not qualified to make.