Case Study
Polly2.com
Building an AI language learning platform for 75+ languages AND the marketing ecosystem to reach those 74% of internet users who DON'T speak English.
75+
Languages Supported
75+
SEO Target Languages
3.75B
Addressable Market
Unlimited
Localised SEO Pages
The Opportunity: A Multilingual Internet
English accounts for only around 26% of internet users worldwide, yet the vast majority of online products and services are built exclusively for English speakers. As of 2025, approximately 6 billion people are online globally — meaning roughly 4.4 billion internet users speak a language other than English as their primary language.
The growth of non-English internet usage has been staggering. Between 2001 and 2011, Arabic-language internet use grew by 2,501%, Russian by 1,826%, and Chinese by 1,277%, compared to just 281% for English. This trend has only accelerated since, with English-language website content dropping from 75% in 1998 to under 50% by 2025.
It is evident that the online language learning market — worth billions globally — was overwhelmingly non English speakers wanting to learn English. The requirements were clear: build a platform that treats every language market as a priority, not an afterthought. Having already developed a tutorial ecosystem allowing any person natively speaking any language to learn any one of the other 75+ languages (English 80% of the time), we were in a perfect place to develop a universal multilingual marketing and customer services platform.
The top 10 non-English internet languages by user count:
Chinese (888M), Spanish (364M), Arabic (237M), Indonesian (198M), Portuguese (172M), French (145M), Japanese (119M), Russian (116M), German (93M) — together representing over 2.3 billion users.
Source: InternetWorldStats / Wikipedia — "Languages used on the Internet"
English-First, Translate Everywhere
At the core of Polly2's architecture is a simple principle "English-first, translate everywhere." All content — from UI labels and button text to marketing copy and email templates — is authored once in English and then systematically translated into every active language on the platform.
This isn't a simple find-and-replace. We built a complete translation management system backed by a relational database with three key tables:
Site Languages
Each language is registered with its ISO code, English name, and native name (e.g. "es" / "Spanish" / "Español"). Languages can be activated or deactivated without touching code.
Translation Groups
Translations are organised into logical groups — "auth", "navigation", "dashboard", "welcome_email" — so content teams can work on specific areas without wading through thousands of strings.
Site Translations
The actual translated strings, keyed by language + group + key. A robust fallback chain ensures users always see content: translated text → English fallback → translation key.
The admin interface gives non-technical staff full control — adding new languages, creating translation groups, and editing individual strings — all without developer involvement. When a new language is added, the system can automatically populate translations using AWS Translate, with human review available through the admin panel.
Seamless Frontend Localisation
On the frontend, Polly2 uses Vue I18n integrated with the backend translation API. When a user selects their language, translations are loaded asynchronously and cached in the browser. The user's preference is persisted in localStorage, so returning visitors see the site in their language immediately.
Every piece of user-facing text — from the login form to achievement notifications — flows through the translation system. Developers simply use translation keys in their Vue components, and the correct language string is resolved at runtime. This means adding a new language to the entire application requires zero code changes — just database entries.
Automatically Translated Emails
One of the most impactful features we built is the multilingual email system. Every transactional email — welcome messages, email verification, streak achievements, subscription confirmations, absence reminders — is automatically sent in the user's preferred language.
Each Mailable class reads the user's pref_lang setting and pulls the appropriate
translations from the database. Subject lines, greetings, body copy, and calls-to-action are all
localised. If a translation doesn't yet exist for a particular language, the system falls back to
AWS Translate for on-the-fly translation, then caches the result for future sends.
This means a user who signs up in Bengali receives their welcome email in Bengali. A Spanish speaker gets their 7-day streak celebration in Spanish. The entire email lifecycle respects the user's language without any manual intervention.
Emails automatically localised:
Multilingual SEO at Scale
Marketing a language learning platform to non-English speakers requires more than just translating a homepage. We built a comprehensive SEO system that creates native-language landing pages targeting high-value search terms in each market.
The system pages and blog posts in every language required with
localised URLs, content, and metadata. For example, a Chinese speaker searching for English learning
resources finds pages at /zh/xue-yingyu-ai,
while a Hindi speaker lands on /hi/ai-se-angrezi-sikhen.
Each language has multiple dedicated SEO pages across multiple campaigns — Polly2.com has "Learn English with AI" (high intent) and "Learn English" (high volume) — with automated internal linking, anchor text rotation, and auto-updating sitemaps. The combined target search volume across all languages exceeds 2.8 million monthly searches.
The numbers for this particular use case can be found below:
| Language | Speakers | Monthly Searches |
|---|---|---|
| Chinese (Simplified) | 1.1B | 808,000 |
| Spanish | 500M | 505,000 |
| Hindi | 600M | 403,000 |
| Portuguese (Brazil) | 215M | 304,000 |
| Arabic | 420M | 252,000 |
| Japanese | 125M | 206,000 |
| Indonesian | 200M | 153,000 |
| Vietnamese | 95M | 102,500 |
| Bengali | 265M | 82,000 |
| Urdu | 230M | 71,500 |
| Total | 3.75B | 2,887,000 |
Automated Social Media Publishing in 75+ Languages across Multiple Locations
Write one blog post and automagically publish it across as many accounts and languages as you need. Our translation pipeline plugs directly into your social distribution workflow:
- • LinkedIn — Company pages & personal profiles
- • Facebook Pages
- • Instagram for Business
- • TikTok — via Iconosquare
- • Threads — via Typefully
- • …and more
By integrating Social Media with our translation services, a single piece of content is automatically translated and distributed across every target language and platform — multiplying your marketing productivity without multiplying your workload.
Technology Stack
Backend
- • Laravel (PHP) with Inertia.js
- • MySQL with dedicated translation tables
- • AWS Translate for automated translations
- • Queue-based translation processing
- • RESTful API for translation management
Frontend
- • Vue.js 3 with Composition API
- • Vue I18n for runtime localisation
- • Async translation loading & caching
- • Capacitor for Android/iOS apps
- • Tailwind CSS
AI & Speech
- • AI-powered conversation tutor
- • AWS Polly for text-to-speech
- • Deepgram for speech recognition
- • Real-time pronunciation feedback
Infrastructure
- • Stripe for multilingual payments
- • Firebase push notifications
- • Auto-updating multilingual sitemaps
- • Google Play Store distribution
The Result
Polly2 launched as a truly multilingual platform from day one — not an English product with translations bolted on. Users in any of the 75+ supported languages experience the full product in their native language: the interface, the learning modules, the AI tutor conversations, and every email they receive.
The admin team can add a new language to the entire platform — including all UI strings, email templates, and SEO landing pages — without writing a single line of code. The translation management system has processed over 1,300 translation keys per language, with real-time progress tracking and automated fallbacks ensuring nothing breaks during rollout.
By designing for multilingual markets from the architecture level up, Polly2 is positioned to capture a share of the massive non-English language learning market — a market that most competitors simply ignore.
Want to reach a global audience?
We build applications designed for multilingual markets from the ground up. Let's talk about your project.
Get in Touch