Localization Arbitrage — which markets resist Western apps, and where the gaps are

📚 Superseded — this document has been merged. Everything here now lives in The Niche-Down Field Guide, together with the other research docs, deduplicated and reorganised. This page is kept so existing links keep working; it is a snapshot and will not be updated further.

---

Research doc · 2026-08-12 · Sources: AppKittie (tooling test + revenue) and Apple's public storefront charts for 23 markets: US, GB, CA, DE, FR, IT, ES, NL, SE, PL, TR, BR, MX, IN, SA, AE, JP, KR, TW, HK, SG, TH, ID, PH, VN. You were right to push on this — my earlier read was flawed, and the corrected answer is much more interesting.

---

1. Your direct question: can AppKittie do regional searches?

Partially, and not for the markets you care about. I tested every regional surface it exposes:

Three things follow:

1. get_supported_countries lists JP, KR, CN, TW, HK, SG — but the data behind them is largely absent. Do not trust that list as a coverage guarantee; I did, and it produced a wrong answer.
2. There is no CJK keyword index at all. Which means no ASO research in Japanese, Korean or Chinese — the languages where all the actual search volume in those markets lives.
3. My earlier "geography loses" conclusion was built on a broken instrument. I used search_apps's countries filter, which AppKittie defines as the app's primary language, not the market it succeeds in. And I sampled China and SE Asia while skipping Japan, Korea and Taiwan — three of the highest-ARPU app markets on earth. Both errors pushed the same way.

Even the Latin-script keyword numbers deserve suspicion: diet returns identical appsCount (210) and trafficScore (73) for both US and JP, with only popularity and difficulty varying. That smells like a partially-applied country parameter, not genuine JP data.

What to use instead

Apple publishes every storefront's charts for free. This is what produced everything below:

https://itunes.apple.com/{cc}/rss/topgrossingapplications/limit=25/genre={id}/json

cc = any storefront (jp, kr, tw, hk, sg, th, id, ph, vn…). genre = 6013 Health & Fitness · 6017 Education · 6023 Food & Drink · 6015 Finance. Also topfreeapplications / toppaidapplications. No key, no credits, every market.

It gives rank order only — no revenue. So the working method is: Apple's feed to find who wins where, then AppKittie or the app's own IAP prices to size it. That combination works.

---

2. The headline: localization resistance varies enormously by market

Share of the Top Grossing Health & Fitness top 15 held by local or regional apps:

Korea and Japan are the real opportunity, and they are closed to exactly the degree that makes them worth entering. MyFitnessPal is the #1 grossing health app in both the US and the UK and does not chart in either country. That is not a small gap; that is a whole market that solved the problem locally.

And it varies by category too

The pattern behind it: localization resistance tracks how culturally specific the underlying data is. Food databases are radically local (a cha chaan teng menu is not in the USDA database), so food apps localize. Wine is already a global culture with a global vocabulary, so wine apps don't. Workout mechanics sit in between — a squat is a squat, but coaching tone and body ideals are not.

One more finding, hiding in the Education charts: of the top 12 grossing Education apps, the number that are English-learning is 9 in Japan, 8 in Korea, 9 in Taiwan, 8 in Hong Kong. Education in Asia is overwhelmingly one behaviour. That triples the confidence on the English-learning wedge from the last doc.

---

3. The operators already running this exact play

This is the most useful thing in the whole study — a cohort of small studios building once and shipping localized across CJK. They are your direct template.

DEEP FLOW's Learna AI is the literal answer to your question — an app that charts in the US top 5 for Education and charts in Hong Kong, Japan and Taiwan. Same product, localized.

Gymwork is the best-documented. Korean company, ships English/Japanese/Korean from one binary, claims 200,000+ users, ~100k/monthbyAppKittie^′s(Western − biased, thereforeprobablylow)estimate.Noteitspriceladder : **14.99/mo and $138.99/yr at the top, down to a $22.99/yr "eligible only" tier.** Ten IAP products. Aggressive per-market price discrimination is part of the play, not an afterthought.

Solo developers are holding chart positions in these markets

Note the two baby-weaning apps. My top recommendation from the last doc — baby weaning × culture — is already being executed by solo developers in both Taiwan and Korea. That validates the thesis and warns you that the obvious version is taken.

---

4. CaLoMei — the template, and the reality check

A Hong Kong solo developer, app released ~3 March 2026, #1 top grossing Health & Fitness in Hong Kong five months later. What the localization actually consists of, from its own store listing:

• 香港地道食物資料庫 — a Hong Kong food database: cha chaan teng dishes, cart noodles, roast meat rice.
• AI 語氣評論 — 每次掃描完都會用廣東話寸你一句 — after every scan, the AI roasts you in Cantonese.

That is the entire moat: a local food database plus a local sense of humour. Everything else is Cal AI. It is a 21MB app with a Vercel-hosted privacy policy and 9 Meta ads running.

Now the number that matters. By AppKittie's estimate: ~1,000 downloads and **~$6,000 in the last 30 days**, MAU 1,408, DAU 317, 288 reviews. It has also raised prices twice — 4.99→9.99 monthly, 29.99→39.99 annual.

Being the #1 grossing health app in Hong Kong is roughly a $6k/month business.

That is the single most important calibration in this document, and it cuts both ways. It is a real solo-founder income earned in five months by one person with a food database and a joke. It is also nowhere near what a mid-chart US app earns — MacroFactor sits at US #7 on $3.1M/month.

Rank is not revenue. Market size caps you regardless of how well you execute.

Rough ordering of the prize: Japan (~125M, top-3 global app market) > Korea (~52M) > Taiwan (~23M)

Hong Kong (~7.5M). Singapore is ~6M and an English market, so it offers no moat at all.

---

5. Beyond Asia — Europe, LatAm and the Gulf behave completely differently

Same method, sixteen more storefronts. Localization works outside Asia too, but the shape of the opportunity changes market by market.

Europe is not per-country localized — it is pan-European

US apps are displaced across Europe, but not by national champions. They are displaced by a handful of European companies operating across the whole continent:

So entering Europe means fighting pan-European incumbents rather than filling a per-country gap. The barrier is competitive, not linguistic — lower localization opportunity than Asia.

The exception is Poland, which behaves like a mini-Korea: three genuinely Polish apps in the top 10 — Fitatu (#1), Diet & Training by Ann (#5), Fitollo (#8). Hard language barrier, distinct national cuisine, local winners.

LatAm has real national champions

Fitia (Nutrition Technologies, Peru) at ES #6 and MX #3 · Welmi MX #5 · Symmetry ES #9, MX #7 · silBe: Fitness para Mujeres (Colombia) MX #9 · Befit BR #3.

And the one to look at hardest: Monju: Emagrecer com GLP-1 (MeuGuru, Brazil) at BR #6 — that is MeAgain's $736k/month GLP-1 play, localized into Portuguese. Direct proof of the "US winner → localize" pattern you asked about, outside Asia.

The Gulf contains the sharpest natural experiment in the study

Two adjacent countries, same language, opposite charts:

Saudi has four local apps in its top 10 and pushes MyFitnessPal to #2. The UAE chart is entirely Western. The difference is population composition: Saudi is a domestic-majority market, the UAE is an expat-majority one. Expat-heavy markets do not localize — Singapore and the UAE are the same story. That is a rule you can apply before doing any research: check who actually lives there.

And SE Asia does have local winners — my earlier read was wrong twice

Measured properly by storefront rather than by app language:

• Thailand — Kalguroo #3 (Thai), NubCal #8 (Central Content Co, Thai), Be Fit #4
• Vietnam — Nutree #3, Dietfit #5 (c0x12c), CalSnap #9 (Trinh Muoi, solo dev)
• Philippines — Bitalyze #9 (justine espinosa, solo dev)
• Indonesia — almost entirely Western

So local players do win chart positions there. But the revenue finding from the first sweep still stands: those markets pay very little. Rank is not revenue — it is the same lesson as CaLoMei, and it applies twice over in SE Asia.

The finding that survived every market

Calorie counting is localized in essentially every country on Earth. Each of these markets has a different #1-or-near-#1 calorie app:

US MyFitnessPal · GB Calorie Counter + ($1M/mo, $25/DL) · DE Yazio · FR Foodvisor · SE Lifesum · PL Fitatu · ES & MX Fitia · BR Befit/Monju · SA Kalee, Mezan, Miran · TR EatBetter · TH Kalguroo, NubCal · VN Nutree, Dietfit, CalSnap · JP あすけん, カロミル · KR 필라이즈, 칼로그 · TW WiseMeal · HK CaLoMei

Sixteen markets, sixteen different answers. A food database is the most locale-bound data structure in consumer software — and it is the single most robust finding of this whole research programme. If you want a localization play, this is the category that has already proven it in every market anyone has tested.

---

6. The tension you should decide on deliberately

Your native advantage is in the smallest market with the weakest economics. Cantonese and Hong Kong give you real, unfakeable insight — the CaLoMei kind — but HK is 7.5M people and a $6k/month #1. Japan and Korea are where the money is, and both need language depth you would have to buy or partner for.

Three resolutions, in the order I'd rank them:

A. Taiwan as the primary, Hong Kong as the test bed. Traditional Chinese covers both. Taiwan is 3× Hong Kong's population with a healthy iOS share, and TW's charts show the most solo-developer and small-studio penetration of any market tested. Validate in HK where you have native instinct and the ad costs are trivial, then ship the same Traditional Chinese build into TW. This is the option I'd take.

B. Build once in English, localize into 3–4 CJK markets. What EpopSoft, Gymwork and Munice actually do. Higher ceiling than any single market, and it makes the content investment amortise. The cost is that you are competing with those studios directly, and they have a head start.

C. Do the inverse — build for the West from Hong Kong. The earlier finding still stands: Singapore- and Türkiye-registered studios earn far more selling into the US than any SE Asian market pays. This is the highest-ceiling option and the one with no language moat protecting you.

Note that these are not exclusive: A is the cheapest way to learn whether you can build a genuinely local product at all, and it costs little if you then pivot to B.

---

7. Where the gaps actually are

Filtered to things that (a) have a proven Western original, (b) show no strong local incumbent in at least one CJK market, and (c) a solo founder can build.

Strong

1. Local-food-database calorie tracking, per market. The most reliably localized category there is — every market has a different #1 (US MyFitnessPal, JP あすけん, KR 필라이즈/칼로그, TW WiseMeal, HK CaLoMei, and in the UK, NutraTech's Calorie Counter + at GB #5 doing $1M/month on 40,000 downloads — $25 per download, one of the highest ratios in this entire study). The UK case proves this isn't only an Asian phenomenon; it is a food-database phenomenon. The open question is which market still lacks a good one.
2. Strength programming, localized. Ladder does $6.5M/month at $20.06/DL in the US and charts nowhere in Asia. Korea's chart is full of loggers (Gymwork, Flick, Bunnit) plus Planfit. A Ladder-grade programming product in Traditional Chinese has no visible incumbent in TW or HK.
3. English-learning for a specific exam or profession. 8–9 of the top 12 Education apps in every CJK market are English-learning, and the exam-specific ones are local (abceed and StudySapuri for TOEIC in JP, 刷刷庫 in TW). Duolingo and Speak own the general slot; the specific slots are still being won by small teams.

Worth noting, with caution

4. Baby weaning, Traditional Chinese. Solid Starts does $1M/month at $12.50/DL in the US, and solo devs already hold TW #7 and KR #11 with weaning diaries. Validated, but no longer empty — you would be entering against an incumbent rather than into a gap.

Also strong, outside Asia

5. Poland. The only European market with genuine national champions in this category (Fitatu #1, Diet & Training by Ann #5, Fitollo #8). Hard language barrier, distinct cuisine, 36M people, EU-level ARPU. The most under-discussed localization market I found.
6. Saudi Arabia. Four local apps in the top 10, MyFitnessPal pushed to #2, and Gulf ARPU. The constraint is that you would need Arabic and real cultural fluency — but the market is demonstrably winnable by local product.

Avoid

• Expat-majority markets — Singapore and the UAE. Both charts are ~14/15 Western apps, for the same reason: the paying population is largely foreign. Check population composition before you research anything else; it predicts localization resistance better than language does.
• Wine, and anything else already globally standardized. No local data, no local moat.
• Japan or Korea without a native-fluency partner. The moat that protects あすけん from MyFitnessPal will protect it from you too.

---

8. Caveats

• Apple's legacy RSS feed gives rank order only, with no revenue and no documented refresh cadence. Everything about relative position here is solid; everything about size comes from AppKittie estimates or app IAP prices, both weaker.
• AppKittie systematically under-indexes Asian apps — あすけん returns a 404 outright, and Gymwork only resolves because it ships an English listing. Treat any AppKittie revenue figure for an Asia-focused app as a floor, not an estimate. This blindness is also a mild advantage: if the standard Western ASO tools cannot see these markets, Western competitors are not systematically hunting in them.
• **CaLoMei's ~6, 000couldbeUSDorHKD * *—AppKittiereportstheapp^′sstorefrontcurrencyasHKDwhilelistingIAPpricesindollars.IfHKD, therealfigureisnearerUS770 and the calibration point gets sharper, not softer.
• A top-15 chart is a thin sample of a market. It shows who wins, not how large the long tail beneath them is.
• Chart position reflects recent revenue velocity, not installed base. A local app at #1 may be mid-promotion; a global app absent from a chart may still have a large dormant user base there.