Stringers
Mobile App · Android · written 11-9
رفيق: محادثة وتعلم الإنجليزية(Rafeeq English) · Rafeeq English

"You've studied English for years and still can't speak it?": a one-person Arabic-English course app from Jordan, with most of its 9 thousand users essentially bought

What it is An AI English course app with explanations fully in Arabic and practice fully in English: AI speaking partner, instant essay grading, a graded placement test, 650 multi-part A1–C1 courses. The first 10 units are free, then four subscription tiers based on AI quota.

Why it blew up It didn't grow organically. Starting Sep 27, the developer ran Google app ads with two YouTube Shorts in Jordanian and Iraqi Arabic dialects (later adding Saudi Arabia), and daily installs went from 1 to two or three hundred.

CHow you could follow Don't copy the Arabic version as is. Learn its structure (graded test, courses, review, assessment loop, tiers based on AI quota), take it to a native-language market where a generic AI tutor doesn't land, and test with Shorts like "you keep mispronouncing this" and a small Google ads budget, watching trial-to-paid conversion.

C grade: it proves one person can build a full course and buy installs on a small budget, but not that anyone pays. Arabic-English learning is already crowded with Duolingo and a dozen paid-acquisition AI tutors. anaFluent, with the same positioning, launched half a year earlier and didn't take off, and the premium for "native-language explanations" isn't visible yet. Almost nobody knows about it yet

Est. monthly revenue (Google Play) under $1KWe estimated this from its rank on country top-grossing charts and real install counts (it isn't on the grossing chart, so confidence is lower), calibrated against real revenue; typical error is 2–3x. Read it as a tier, not an exact number.

Data (updated daily) · first launch 2026-08-30 · 11-09 installs 8,916
Android · Total installs worldwide
7.5k8k8.5k9k11-0311-098.9k
Data 2026-11-03 → 2026-11-09 (7 days, 2 observations). Google's detail page only refreshes install counts every other day, so a flat day doesn't mean nobody installed.

Rafeeq (رفيق, Arabic for "companion") is an English course app built by a solo developer in Jerash, Jordan. Explanations are fully in Arabic, practice is fully in English, an AI partner talks with you and corrects your essays, with 1 multi-part courses from A1 to C650. Subscription-priced. Launched Aug 30; by Nov 9 it had 8,916 installs on Google Play, +1,299 in the last week. This curve was drawn almost entirely by the Google app ads the developer started in Sep 27, and in the same week several AI tutors in the Arabic-English learning market each got hundreds of thousands of installs. Below: how one person built such a deep product, how users were bought, whether its selling points are worth anything, and whether you should follow.

Sep 27: a Short starts running as an ad

In 2026 Sep 27, a YouTube Short started showing up as a Google ad on phones in Iraq and Jordan, with an Arabic title in a Jordanian accent:

「بتقدر تحكي إنجليزي؟ 🤖」 — Can you speak English? 🤖

The advertiser is the developer personally, SAAHERAH MOHAMMAD IBRAHIM ETOOM, registered in Jordan; tapping it jumps straight to the Google Play install page (Google Ads Transparency Center, 11-09 check). In Oct 26 a second one was added targeting Saudi Arabia, and there are also impression records in Germany, France and Czechia (inferred to target Arabs living in Europe), with a more pointed hook:

「درست الإنجليزي سنين وما بتحكيه؟ 😳」 — You've studied English for years and still can't speak it? 😳

Both are still running. Before the ads started, AppstoreSpy estimated about 1 installs a day (9-23); the day after launch (9-28) it hit 14, and from Oct 5 it ran 170 to 440 a day. The timing lines up exactly. Its privacy policy is also blunt: "We advertise this app on Google." Firebase and Google Ads accounts are linked, and three events are passed back: first open, sign-up and purchase. Android only.

One person, a whole course

The developer registered on Play under a personal name and a gmail address, and the terms say it is "operated by the trade name Rafeeq in Jordan". This is the only app under that name. But the product doesn't look like one person's work:

  • AI speaking partner: you speak, it replies, corrects pronunciation and word choice, and explains the errors in Arabic;
  • Instant essay grading: grammar, word choice, more natural phrasing, with an Arabic explanation for each error;
  • Graded placement test: a free reading and listening test once a month; Pro users also get an IELTS-style speaking and writing test, with four skills scored on an international scale (the app notes it is not an official IELTS);
  • In reading, tap a word for its meaning (only words at your level), listening with recorded dialogues, 95 short Arabic-language grammar videos, spaced review, and 77 achievements.

The service provider list in the privacy policy shows the stack: Claude handles grading, conversation and assessment scoring; Google's TTS and Gemini handle read-aloud and speaking audio; OpenAI does speech-to-text; Firebase is the backend; RevenueCat manages subscriptions. In the 10 weeks after launch it shipped 13 versions, the latest in Nov 8.

The positioning on its website is its whole argument:

"Most English learning apps are written in English. If you can't understand the explanation, you don't know why you got it wrong — you just memorize that it was wrong, and move on."

How it makes money

The first 7 units of A1 are free, with one unlocked per day, plus one free quick placement test. After that come four subscription tiers: Plus (all courses), Pro (more AI conversation and grading quota), Pro Max (for people who practice every day), and Family Plus (four accounts). Monthly or yearly; the first three tiers come with a 11-09-day free trial that requires a card and auto-renews (website, 10). Play lists the in-app prices between $7.99 and $159.99.

Note the logic of the tiers: what mainly separates them is how many AI conversations and grading runs you get per month. LLM calls are its biggest variable cost. The more users use it, the more they pay, so cost and price are naturally tied together.

How many people pay, what the trial-to-paid conversion is, and whether ads pay back: none of these numbers is visible from outside. It has never appeared on any top-grossing chart.

How big is it really

MetricSources
Total installs7,617(11-03)→ 8,916(11-09)Google Play installs
Last 7 days+1,299, about 217 per daySame as above
ChartsSaudi Arabia Google Play Education Top Free chart 196 (11-03), 195 (11-09)Google Play charts
Rating5 reviews on tablet, 5.0 stars; not enough on phone to displayPlay store listing, 11-09
iOS0 ratings in the US, Saudi Arabia and Egypt; 1 reviews at 1 stars in Jordan, 1 reviews at 5 stars in the UAE ("تطبيق رائع جدا", very great app, 10-02)App Store,11-09

The volume is almost all on Android; iOS barely moved. One claim says it reached the top 10 of Jordan's Education Top Free chart; we couldn't verify this. Even if true, the Jordanian market is small, so the bar for the top 10 was low to begin with.

Compare it with the same Saudi Education Top Free chart (Google Play, 11-09 new installs over the past 7 days): Langla +689k, Learna +690k, Lurnia +422k, Linguza +179k, Rafeeq +1,299. The gap is two to three orders of magnitude.

Its YouTube channel: an ad creative library

The channel @Rafeeqapp is called 「رفيق: تعلم الانجليزية بالعربي」 (Rafeeq: learn English in Arabic), with 103 subscribers and 99 videos. The Shorts titles are easy to get, all classic traps for Arabic speakers learning English:

  • "stomach" isn't pronounced "stow-mach" — nobody taught you this rule 😨 ch = k
  • "pajamas" is pronounced differently than you'd expect 😵 Not بيجاما
  • "'After one hour' ❌ Don't say 'after one hour' 🙅‍♀️ Say 'in an hour'"
  • "'computer' has a p, not a b" — targeting the pain point that Arabic has no p sound

These Shorts all sit neatly between 800 and 1,100 views (YouTube channel page, 11-09), while the long videos get only a few to a few dozen views each. Numbers this uniform don't look like organic traffic; they look more like fixed exposure from ads or a recommendation pool (inference). The channel itself doesn't pull users in. Its job is to churn out material in bulk and pick the strongest clips to run as Google ads.

No other channel shows any sign of it: the Meta Ad Library finds no ads by domain, package name, or brand name (searching 'رفيق' returns the same-named Rafeeq Kids and other course sellers); TikTok has no official account or creator videos for it; X had zero mentions in 90 days; there is no media coverage. The brand name is also taken by a pile of other 'Rafeeq' products (a Qatari food-delivery app, a Hajj assistant, a kids' app), so searching the name turns up almost nothing.

Who it's competing with

Arabic speakers learning English was already a crowded market in 2026. The crowding comes mainly from these groups:

TypeProductScale (Google Play)How Rafeeq differs
Global giantsDuolingo (has an English course for Arabic speakers)995M installs, 7-day +3.1M, #2 on the Saudi education Top Free chartBroad, gamified all-in-one
Paid-acquisition AI tutorsLearna AI (Codeway), Langla, Linguza, Lurnia, Praktika, plus a dozen others such as Fluzy and HilingoEach 1M to 50M installs; 7-day +50k to 690kThe same 'AI roleplay conversation' template, with titles swapped by country and only the interface localized into Arabic
Human tutors51Talk3.24M installs, #3 on the Saudi education Top Free chartReal teachers, expensive
Established Arabic-language English courses'تعلم اللغة الانجليزية من الصفر' (Learn English from Zero, 2018) and iStoria (2020)15.6M installs, 4.34M installsExplanations in Arabic, but no real-time AI grading
New products with the same positioninganaFluent (iOS, listed 2026-01, 'designed specifically for Arabic speakers')47 ratings in the Saudi storeNearly identical positioning, launched half a year earlier, and it hasn't taken off either

There is also a website, Taleek (طليق), which claims 1M students, 500+ lessons, six levels, and grammar explained in Arabic. Its structure is almost identical to Rafeeq's (scale unverified).

Is 'explained in your native language' worth paying for?

Rafeeq's position is clear: it offers a full graded course and Arabic explanations beyond the AI-tutor template apps, and real-time AI correction beyond the established Arabic-language courses. It's closer to 'Busuu for Arabic speakers, plus AI grading.'

But this selling point is not why it gained traction. Its volume was all bought with ads, while on the same charts a generic AI tutor that only translated the interface into Arabic was pulling in hundreds of thousands of installs a week. That suggests Arabic speakers aren't short on 'AI teaches English' and that no data yet shows they'll pay more, or stay longer, for 'explained more thoroughly in Arabic.' anaFluent, with the same positioning and a launch half a year earlier, has only a few dozen ratings so far, which is circumstantial evidence.

If it were you

C: you can learn something from it, but copying it isn't very profitable. What Rafeeq shows is that one person using LLM APIs can build a credible course, and that a small Google ads budget can reliably buy a few hundred Arabic users a day. It doesn't show that people will pay, or that payments can cover the ad spend. The Arabic English-learning space is already saturated by a dozen paid-acquisition AI tutors and Duolingo, and Rafeeq itself hasn't broken out.

Who it suits: people whose native language is one that 'most English apps don't properly explain in the native language,' who are teachers themselves or have made language content; who can work with Google app ads and calculate install cost and trial-to-paid conversion. Ideally 1–2 people: one writes the course, one runs the ads.

Who it doesn't suit: anyone who wants to make a straight Arabic copy (the category is full, and neither Rafeeq nor anaFluent has proven this path makes money); people who can code but can't write course content (the barrier for this kind of product is 650 lessons and 95 explainer videos, not code); anyone counting on organic store search ('learn English' is one of the most expensive keywords in every major language).

How to do it: copy the structure, not the language. Leveled placement test → course → spaced review → assessment, forming a closed loop; tier AI conversation and essay grading by monthly quota so heavy users pay more; use exam anxiety (it mimics an IELTS speaking test) as the reason to upgrade. Pick a language with a local dialect and local humor where generic template translation loses the flavor. Use the free units and a 7-day trial to measure trial-to-paid conversion first, and only then decide whether to add paid tiers.

How to get users: follow its playbook. Start with a batch of native-language Shorts targeting the mistakes native speakers always make ('you've been mispronouncing this word,' 'you've been saying this sentence wrong'), with titles in colloquial dialect. Run the two or three best performers directly as Google app ads, starting with cheap-install small countries and expanding to wealthier neighbors and diaspora markets. Send the 'purchase' event back so the campaign optimizes for paying users, not just installs.

Deadline: there's no hard window; this is a slow business. If you do pursue it, first use two to three months and a few thousand dollars of ad budget to measure trial-to-paid conversion, and stop if the numbers don't show up.

Signs you've misjudged it: (1) Rafeeq enters the Saudi or UAE education top-grossing chart in the next two to three months, or its cumulative installs pass 100k with ratings climbing into the hundreds and staying high. That would mean 'native-language explanation + a full course' really commands a premium, and the value of doing it in another language goes up. (2) Its Google ad creatives grow from 2 to a dozen or more, and it runs in more countries. A developer adding spend usually means the ads are paying back. (3) The reverse: if its two ads stop running after December and daily installs fall back to single digits, the bought users aren't sustaining it, and this path doesn't work in the Arabic-speaking market.

Android store page ↗

Research notes: competitor table, acquisition evidence, full research report
Competitors
NameWhereLaunchedScaleHow it differs from this one
DuolingoAndroid2013995M installs, 7-day +3.1M, #2 on the Saudi education Top Free chartBroad, gamified all-in-one with an English course for Arabic speakers; Rafeeq has real-time AI speaking and writing feedback and Arabic grammar videos
Learna AI(Codeway)Android2024-0954.9M installs, 7-day +690k, #9 on the Saudi education Top Free chartA generic multilingual AI-tutor template with only titles and interface localized into Arabic; Rafeeq is a structured CEFR course with Arabic explanations
Langla: AI Language Tutor(Aquaworld Infotech)Android2026-04-2310M installs, 7-day +689k, #1 on the UAE education Top Free chart and #5 in Saudi ArabiaPaid-acquisition AI roleplay conversation; Rafeeq's course plus assessment loop is three orders of magnitude smaller in volume
Linguza: AI English Tutor(NXL AI Solutions)Android2026-05-189.87M installs, 7-day +179k, #3 on the UAE education Top Free chartSame as above, a generic AI tutor; Rafeeq focuses specifically on native Arabic speakers
Lurnia / Fluzy / Hilingo / LearnUp / TutorMe and other AI tutorsAndroid2025–2026Each 1M–10M installs; 7-day 5–420kThe same AI-tutor template with titles changed by country to flood the market; Rafeeq has deeper content and a smaller ad spend
51TalkAndroid20213.24M installs, #3 on the Saudi education Top Free chartHuman tutors; Rafeeq is AI grading, cheaper and available for practice anytime
تعلم اللغة الانجليزية من الصفر(ZAmericanEnglish / PUMKINN)Android201815.6M installsLong-established Egyptian Arabic-language English course with no real-time LLM grading
iStoria(Oxford Academy Of Languages)Android20204.34M installsArabic-region English story-reading courses with no AI speaking partner
طليق TaleekWebNot verifiedClaims 1M+ students and 500+ lessons (unverified)Also explains grammar in Arabic with leveled courses; Rafeeq adds AI speaking and writing feedback
anaFluent - Learn English(Ingenuity Labs LLC)iOS2026-01-1947 ratings in the Saudi storeNearly identical positioning ('designed specifically for Arabic speakers'); Rafeeq has Android, runs Google ads, and has more content
Where users come from
  • Confirmed Google app ads (UAC, YouTube Shorts video creatives): the privacy policy states it advertises on Google and that Firebase sends first-open, sign-up, and purchase events back to Google Ads (Android only). The Google Ads Transparency Center lists the advertiser SAAHERAH MOHAMMAD IBRAHIM ETOOM (AR06076421367039787009, JO) with 2 video creatives running; first shown 09-27 (Iraq, Jordan) and 10-26 (Saudi Arabia, plus German/French/Czech), landing on com.rafeeqenglish.app. Daily installs took off starting 09-28.
  • Confirmed Its own YouTube channel @Rafeeqapp: 103 subscribers, 99 videos; long videos get a few to a few dozen views, while the Shorts cluster around 1K views. The ad creatives are drawn from these Shorts.
  • Confirmed Website / store search: the website download button carries utm_source=website; it ranks #195 only on the Saudi education Top Free chart, so organic search volume is limited.
  • Inferred Meta (Facebook/Instagram) ads: the recall card mentions AI female-host voiceover ads starting September; the Meta Ad Library finds no ads by domain, package name, or brand name (only the same-named Rafeeq Kids).
  • Inferred TikTok / X / media / creators: TikHub search finds no official account or creator videos; x-radar shows zero mentions in 90 days; no media coverage.

رفيق: محادثة وتعلم الإنجليزية (Rafeeq English, com.rafeeqenglish.app) research report

Research date 2026-11-09. Data sources: our googleplay / metaads / googleads databases (agent-im server), the Google Ads Transparency Center API (queried directly through the Mac mini exit), the Meta Ad Library web interface (one-time query), Play / App Store store pages, the official site rafeeqenglish.com (homepage, terms, privacy policy), the YouTube channel page, TikHub (TikTok), x-radar (X), and Jina / Exa / Brave search.

One line

An AI English course app built by an individual developer in Jerash, Jordan, where the explanations are all in Arabic and practice is in English: AI voice practice, essay grading, a CEFR placement test, 650+ lessons across A1–C1, subscription-based. It hasn't broken out yet: 10 weeks after launch it has 8,916 installs in total, with two to three hundred new per day, and nearly the whole curve comes from Google app ads (UAC, YouTube Shorts video creatives, targeting Jordan, Iraq, and Saudi Arabia) that started in Sep 27. Its differentiation is 'a course designed for Arabic speakers from scratch + Arabic explanations,' but the same Arabic-region education charts are crowded with Duolingo, Learna, Langla, Linguza, and Praktika, AI tutors that already have localized Arabic and millions to tens of millions of installs. Its differentiation hasn't yet translated into volume that can compete with them.

1. Who they are

  • Developer: the Play 'About the developer' section lists SAAHERAH MOHAMMAD IBRAHIM ETOOM, address Alnabaa, Jerash 11118, Jordan, contact email [email protected]; the iOS seller name is SAAHERAH ETOOM [Play detail page; iTunes lookup id 6794187358]. The terms state it is 'operated under the trade name Rafeeq in Jordan, governed by Jordanian law' [rafeeqenglish.com/terms]. It is an individual developer, not a company; our database shows this developer has only this one app [googleplay developer].
  • Listed: Play 2026-08-30, iOS 2026-08-31 [googleplay app table; iTunes lookup]. Versions have reached 1.1.13 (11-08 updates); 10 versions in 13 weeks, shipping very fast [googleplay app_snap; privacy policy mentions changes to 1.1.4, 1.1.10, 1.1.11].
  • Product (Play listing, website):
    • AI voice practice partner: listens to you speaks back, corrects pronunciation and word choice, explains mistakes in Arabic;
    • Instant writing review: grammar, word choice, more natural phrasing, each error explained in Arabic;
    • Level test: one free reading/listening test per month; Pro users get an IELTS-style Part 1-3 speaking test and writing, with all four skills scored on international levels (disclaimer: not an official IELTS product);
    • Reading with tap-to-define dictionary (only words at your level), recorded dialogue listening, 95 short Arabic grammar explainer videos, spaced review ladder, 77 achievements, A1–C1 with 650+ lessons.
    • Tech stack (privacy policy 'service providers' table): Anthropic Claude for writing/speaking feedback, AI conversation, and test scoring; Google Cloud TTS + Gemini for read-aloud and speaking-test audio; OpenAI for speech-to-text; Firebase for backend; RevenueCat for subscriptions. A typical solo-dev + LLM API assembly.
  • Pricing: first 10 A1 units free (one per day) + one free quick level test; subscriptions Plus / Pro / Pro Max / Family Plus (4 accounts), monthly or yearly; Plus/Pro/Pro Max have a 7-day free trial, auto-billed at expiry [website]. Play in-app purchase range $7.99–$159.99 [googleplay app_snap 11-09].

2. Growth data

DateTotal installsDaily addsSource
09-21 ~ 09-23—~1/dayAppstoreSpy estimate
09-28—14 per dayAppstoreSpy estimate
10-02—30 per dayAppstoreSpy estimate
10-05 ~ 10-16—170–440/dayAppstoreSpy estimate
10-21—1,199/day (single-point peak; don't take the averaged value seriously)AppstoreSpy estimate
10-25 ~ 11-05—110–400/dayAppstoreSpy estimate
11-037,617—googleplay app_day
11-098,9167 days +1,299 (about 217/day)googleplay app_day
  • Charts: Across the 30 storefronts we track, it only appeared on the Saudi Education Top Free chart 196 (11-03) and 195 (11-09). It hasn't charted anywhere else [googleplay chart_rank]. Jordan, Iraq and Egypt aren't among the 30 storefronts we collect. The recall card says "Similarweb shows Jordan Play Education Top Free at #9–10", but on our side Similarweb and AppBrain both show 403, so unverified. Jordan is a small market, and the bar for the top 10 of the Education free chart is low to begin with, so even if true, it doesn't mean volume.
  • Ratings: on the Play tablet view it shows 5 reviews and a 5.0 rating; phone view doesn't show a rating yet. iOS: US/Saudi/Egypt 0 reviews, Jordan 1 reviews (1 stars), UAE 1 reviews (5 stars) [Play listing; iTunes lookup 2026-11-09]. iOS has essentially no volume; the volume is all on Android.
  • Relative scale: on the same Saudi education Top Free chart, Langla +689k in 7 days, Learna +690k, Lurnia +422k, Linguza +179k, Rafeeq +1,299 [googleplay v_chart 11-09]. It is two to three orders of magnitude smaller than the leaders.

3. Competitors

Products that "teach English to Arabic speakers" fall into three groups:

A. Global AI tutors with Arabic localization (largest volume, also the most crowded), Saudi/UAE education Top Free 11-09 [googleplay v_chart]:

ProductLaunchedTotal installs7 new in the last daySaudi Education Top Free
Duolingo (has an "Arabic speakers learn English" course)2013995M3.1M2
Langla: AI Language Tutor(Aquaworld Infotech)2026-04-2310M689k5 (UAE 1)
Learna AI (Codeway; Saudi title "تعليم الانجليزية")2024-0954.9M690k9
Lurnia AI Language Tutor(Astronex)2026-06-173.54M422k13
Linguza: AI English Tutor(NXL AI Solutions)2026-05-189.87M179k19 (UAE 3)
Praktika202221.3M163k33
Fluzy / Hilingo / LearnUp / TutorMe / DailySpeak / Elyra / EngLearn, etc.2025–20261M–10M5–220k37–90

Nearly all of these are utility publishers from Turkey, Vietnam, India, and China, using the same "AI tutor + role-play dialogue" template, swapping titles per country, and scaling via paid acquisition.

B. Long-established Arabic-region courses that teach English in Arabic:

  • 51Talk (live human tutors; Saudi title "معلمك للغة الإنجليزية"): #3 on Saudi education Top Free, 3.24M cumulative [googleplay].
  • تعلم اللغة الانجليزية من الصفر (ZAmericanEnglish / PUMKINN): 2018 years, 15.6M installs, an Egyptian course brand; "دار إشراقة" is selling courses for it on Meta [googleplay; Meta ad library lookup].
  • iStoria (Oxford Academy of Languages): 2020 years, 4.34M installs [googleplay].
  • Taleek (طليق): the website claims "1M+ students, 500+ interactive lessons, 6 levels"; its structure is nearly identical to Rafeeq's (A1–C2 levels, grammar explained in Arabic) [taleek.com; scale unverified].

C. Small new products with the same positioning:

  • anaFluent - Learn English (Ingenuity Labs LLC): on iOS for 2026-01-19, "designed specifically for Arabic speakers," 47 ratings in the Saudi store [iTunes lookup]. Its positioning is identical to Rafeeq's, it launched earlier, and its volume is very small.
  • Name collisions: the name "Rafeeq" is taken by a Qatari food-delivery super-app, a Hajj assistant, an Arabic reading app, Rafeeq Kids (Makeity LLC, Arabic and English for children, advertised on Meta by November), and many others [Jina search; Meta ad library]. Searching the brand name almost never surfaces it, so it can't capture brand-search traffic.

4. Growth tactics (where users come from)

  1. Google App Ads (UAC) is the main channel, confirmed.
    • Privacy policy text: "We advertise the app on Google. Google Analytics for Firebase is linked to our Google Ads account... first open, sign-up, purchase events are sent back... This applies to the Android app only" [rafeeqenglish.com/privacy].
    • Google Ads Transparency Center: the advertiser is SAAHERAH MOHAMMAD IBRAHIM ETOOM (AR06076421367039787009, registered in JO), with 2 video creatives, all currently running:
      • CR14513915268918411265: first shown 2026-09-27, most recently 11-09; shown in Iraq and Jordan; lands on com.rafeeqenglish.app; creative is a Shorts video from the channel @Rafeeqapp: "بتقدر تحكي إنجليزي؟ 🤖" (Can you speak English?).
      • CR17830514303657050113: first shown 2026-10-26, most recently 11-08; shown in Saudi Arabia, plus Germany, France, and the Czech Republic (EU records, presumably targeting Arab migrants in Europe); creative is a Shorts video: "درست الإنجليزي سنين وما بتحكيه؟ 😳" (Studied English for years and still can't speak it?) [adstransparency.google.com SearchCreatives / GetCreativeById; 2026-11-09 lookup].
    • Timing lines up precisely: AppstoreSpy estimates daily installs at 09-23 then 1/day; 09-28 (the day after ads started) 14/day; from 10-05 onward 200+/day. Our own googleads database doesn't track it (it isn't on the winners list).
    • The copy uses Levantine dialect (بتقدر تحكي، رح، ماحد علمك), which fits a Jordanian developer; the core pain point is "studied for years and still can't open your mouth."
  2. YouTube content channel: @Rafeeqapp "رفيق: تعلم الانجليزية بالعربي", 103 subscribers, 99 videos (grammar explainers, listening dialogues, Shorts pronunciation gags). Long videos get a few to a few dozen views each; Shorts are uniformly around 1,000 views, which is very consistent [YouTube channel page 11-09]. Such uniform 1K numbers look more like fixed exposure from paid promotion or a recommendation pool than organic virality. This channel mainly supplies ad creatives and does some SEO; it is not an acquisition channel in itself.
  3. Website + store search: the website download button carries utm_source=website. The Arabic keyword "تعلم الانجليزية" is extremely competitive; in the storefronts we crawl it only ranks #195 on the Saudi education chart, so store search yields very little organic volume.
  4. What I couldn't find: searching Meta's ad library for rafeeqenglish.com, the package name, "رفيق محادثة وتعلم الإنجليزية", and "Rafeeq English" turned up no ads from it (the "رفيق" ads that do appear belong to Rafeeq Kids and other course sellers). The recall card's claim of a "Facebook page starting September with an AI female-host voiceover" could not be verified; it may have confused Rafeeq Kids with this app. TikTok shows no official account or creator videos (@rafeeqapp has 0 followers and looks like an empty placeholder, ownership unclear). On X (x-radar, 90 days) there are zero mentions, and there's no press coverage.

5. Why it's climbing (at its current scale)

  • One person plus LLM APIs can build a "full course"-grade product: Claude for grading and conversation, Gemini/TTS for read-aloud, OpenAI for transcription, Firebase, RevenueCat. 10 weeks, 13 versions, 650 lessons, 95 explainer videos, 4 subscription tiers. The content and feature density look like a small team's work.
  • Starting volume with a small Google UAC budget: ads began 9-27, daily installs went from 1 to 200–400, with two creatives, targeting low-CPI Arabic markets like Jordan, Iraq, and Saudi Arabia. Nearly all 8,916 installs were bought.
  • Clear positioning: "Most English apps are written in English, so if you can't understand the explanation you don't know why you got it wrong." This speaks directly to Arabic beginners and A1–A2 learners; the ad hook is "studied for years and still can't open your mouth."
  • Designed to pay back subscriptions: a 7-day trial that auto-converts to paid (card required), a Family tier, and an IELTS-style speaking test reserved for Pro, turning exam anxiety into a reason to upgrade. Firebase sends purchase events back to Google Ads, which suggests it is optimizing campaigns for paying users. Conversion rates and payback have no public data at all.

6. Differentiation

  • Versus group A (Learna / Langla / Linguza / Praktika): those are "AI character chat + scenario dialogue" with a generic multilingual template, localizing only the interface. Rafeeq is a structured CEFR course system + grammar explained in Arabic + Arabic-language videos, with AI conversation as just one piece. It's closer to "an Arabic Busuu/Duolingo with AI grading."
  • Versus group B (ZAmericanEnglish, iStoria, Taleek, 51Talk): these have Arabic explanations or live human teachers, but no LLM-powered instant speaking/writing feedback. Rafeeq puts "Arabic explanations" and "real-time AI correction" together.
  • Versus group C (anaFluent): nearly identical positioning; the differences are that Rafeeq has Android, runs Google ads, and has much more content.
  • Is that why it's growing? Not the main reason. The growth came directly from Google ads; the differentiation mainly affects whether bought users stay and pay. On the same chart, generic AI tutors without that differentiation pull in hundreds of thousands of installs a week, which suggests Arabic-speaking users aren't short on "AI English teaching"; the premium for "Arabic explanations" hasn't been proven by the data.

7. Visibility and copycats

  • Awareness: almost nobody knows about it yet. No public revenue, no press/X/TikTok mentions, 103 YouTube subscribers, and the brand name is buried under same-named products.
  • Copycats: none seen. If anything, it is following others. The Arabic-region AI English category has been filled for 2026 years by dozens of paid-acquisition AI tutors, and anaFluent launched the same positioning half a year before it.

8. Worth borrowing

  • A solo dev using LLM APIs to build a "whole course" rather than "a chatbot": a closed loop of level test → lessons → review → assessment. Subscription tiers are differentiated by monthly "AI quotas" (conversation and grading), so cost and price are naturally linked.
  • Build a vertical "native-language explanation" version for a small-language market, use local dialect in ad hooks, buy in low-CPI regions (Jordan, Iraq), then expand to the Gulf (Saudi Arabia) and overseas diaspora (Germany, France).
  • Use your own YouTube Shorts as an ad creative library: make a batch of content and send the best ones straight to UAC.
  • The privacy policy is very detailed (no ad ID collection, only three events sent back), which helps with trust in Middle East markets and with store review.

9. Still unclear

  • No data on conversion, trial-to-paid rate, retention, ad spend or payback. The highest IAP of $159.99 only shows there are annual/family tiers, not that anyone buys them.
  • Top 10 in the Jordan Education Top Free chart (recall card cites Similarweb) is unverified; our library doesn't cover Jordan.
  • 10-21's 1,199/day is AppstoreSpy averaging across two observations. It could be one budget boost or one featuring; the cause is unknown.
  • Whether it ran on Meta (the recall card says it ran AI female-host promo videos): no hits by domain, package name or brand name. It leans toward not running or having stopped, though it can't be fully ruled out that it used another Page name.
  • Unclear whether the developer behind it is a small team (personal name registration, gmail contact address).

This article was written by AI from public data and the data we collect daily. Numbers are sourced, and inferences are labeled.