Why We Built Polaris Translation

We run Shopify stores ourselves. When we needed to translate our stores into multiple languages, we tried the existing tools — and kept running into the same problems. So we built Polaris to solve them.

The pricing problem

Most Shopify translation apps charge monthly fees that scale with your store size or number of languages. For a large catalog, you can easily pay hundreds of dollars per month — even though the actual cost of translating text with an AI model is fractions of a cent per word.

The math didn't add up. Most of your content gets translated once and rarely changes. Why pay a recurring premium for translations that are already done?

We created a BYOK (Bring Your Own Key) plan: you pay us a flat monthly fee for the infrastructure, and token costs go directly to Google Gemini with zero markup. You see exactly what the AI charges — nothing more. For stores with large catalogs, this can cut translation costs by 80% or more compared to traditional per-word pricing.

Outdated translations go unnoticed

Our own store's content changes regularly — product descriptions get updated, new collections launch, seasonal copy rotates in. With the tools we tried, there was no way to know which translations were stale. You'd update a product description in English and the old German translation would stay live, silently out of sync.

Polaris tracks source content changes and shows you exactly which translations need updating. When your English copy changes, affected translations are flagged as outdated so you can re-translate just what's changed — not your entire catalog.

Context-aware translations

Most translation tools send each field to the AI in isolation. The title gets translated separately from the description, which gets translated separately from the SEO metadata. The AI has no idea what product it's looking at.

This produces technically correct but often awkward translations. A word that means one thing in a fashion context might mean something entirely different in electronics — but without context, the AI guesses.

Polaris sends product context together: the title, description, category, and related fields are all included so the AI understands what it's translating. The result is more natural, more accurate translations that actually read like they were written by someone who understands the product.

Multi-brand stores need different voices

Our store uses different brand identities for different language markets. The tone we use for German-speaking customers isn't the same as what works for French or Japanese audiences. We needed a way to define per-language translation instructions and glossaries.

With Polaris, each language can have its own set of instructions: formal vs. informal address, specific terminology preferences, brand terms that should stay untranslated, and glossary rules that ensure consistency. The same store can speak differently to each audience — because that's how real localization works.

These aren't problems unique to us. Every Shopify merchant who takes international sales seriously runs into them. Polaris is the tool we wish existed when we started — so we built it.