Blog
Translation Isn't Enough: The Multilingual Search Problem
You've translated your store. Your German customers can read product descriptions, your French customers understand checkout—but when they search for products, they get zero results. A Spanish shopper searches 'vestido de fiesta' and leaves, even though you have dozens of cocktail dresses in stock. With up to 50% of EU traffic coming from multilingual shoppers, this isn't a minor inconvenience—it's a revenue leak. Here's why translation apps solve browsing but break search, and what actually works.
You've spent thousands localizing your store to attract the 76% of shoppers who prefer buying in their native language. Your site looks native, feels native. That is, until they search and cannot find anything.
A German customer lands on your fully translated storefront. They're browsing comfortably, but when they want to find something specific, they naturally think "sommerkleid"—not "summer dress." They type it in. Zero results. You have 30 relevant dresses in stock. They leave.
This isn't rare, unexpected search behaviour. Between cross-border shoppers and foreign-born residents, customers shopping in their second language represent a massive portion of EU traffic—potentially up to 50%. And they're all hitting the same wall: Your translation app localized what customers see, not what they can search.
The Hidden Cost of Incomplete Localization
Seeing promising traffic from new markets or a desire to expand can often inspire stores to internationalize in order to drive conversion and encourage new users to make that purchase. With 76% of shoppers preferring to shop in their local language the ROI on internationalization is massive. Stores can see a 20-30% increase in revenue, and often a 3x ROI on the cost of localization.
Stores find great success using tools such as Shopify Translate and Adapt or Weglot to help them serve international customers.
Most translation apps target real-time translation or create another text layer to serve localized content. This works wonderfully for browsers - customers can see products in their language and understand what they're buying. But it doesn't help with search, which still queries the original product data.
And search matters. Fashion stores using Substanz see search customers spend 65% more per order and convert 6-7x better than browsers. When search works, it's a revenue driver. When it doesn't work for multilingual customers, you're losing your highest-value segment.
Some stores maintain expensive synonym or translation dictionaries to help their search, but this is one more thing to maintain. As new stock comes in and fashion trends change, these dictionaries quickly fall out of date. Each new market means building and maintaining another set of synonyms. It’s a never-ending game of catch-up.
Why Traditional Search Fails Multilingual Customers
Search is Language-Locked to Your Product Data
If your catalogue is in one language, traditional keyword search will also only work in that language. German speakers looking for your summer dresses could be searching for "Sommerkleid" or "sommer dress" (mixing languages and using false friends) and find nothing because you've called everything "summer dresses." They will assume you don't carry what they are looking for and will go elsewhere.
Why Bulk Translation Doesn't Solve Search
Bulk translation helps stores serve international customers. It enables customers to see products in their language and understand what they are buying. But if they can't find what they are looking for, this is wasted effort.
Most apps target real-time translation or create another text layer to display, but it doesn't help with search based on the original data. Even when the translations make enough sense to customers to help them purchase, they may not match the natural way people use language around your products. Machine translation is approximately 85% accurate for e-commerce content , and tools like Shopify's free Translate & Adapt have users reporting "manual adjustments sometimes necessary for accuracy" and "incomplete translations".
The Language Mixing Reality
Bilingual customers naturally mix languages in everyday life, and search is no different. They might search:
"Sommerkleid" when the product is indexed as "summer dress" (completely different words)
"vestido verano" (Spanish) when browsing an English site
"smoking" (German for tuxedo) vs "smoking jacket" (English)—false friends
"boyfriend jeans" when the product is tagged "relaxed fit"—same concept, different terminology
Traditional keyword search requires exact or near-exact word matches. Each failed search is a lost sale.
The Multilingual Customer Experience
Between cross-border shoppers (38% of EU customers) and foreign-born residents (14.1% EU-wide), 25-35% of EU traffic involves customers shopping in their second language. In major markets like Spain (51.6% cross-border shopping) and Germany (20.9% foreign-born), this reaches up to 50% of traffic."
Many of these second-language speakers feel perfectly comfortable reading and using websites in their adopted country's language. But when it comes to searching, they may struggle to remember the exact term or correct spelling in that foreign language. Traditional search fails them.
EU Market Makes This Critical
For stores in the EU, this search barrier is a huge problem. There are 27 countries and 24 official languages. The financial barrier of internationalisation across all these languages is inaccessible for all but the largest of players. With 38% of EU shoppers buying across borders, not having internationalized search can cut out a significant amount of potential customers.
US Market Also Faces This Issue
The US market faces this same challenge. With 65+ million Hispanic consumers wielding $2.4 trillion in buying power, stores operating only in English are missing a massive opportunity. Many of these shoppers prefer to search in Spanish—and when your search can't understand 'vestido de fiesta' as 'cocktail dress,' you're losing sales to competitors who can.
The Compounding Problem for Multi-Market Stores
For multimarket stores this problem compounds. A German store expanding to Poland, Czechia, France, Italy, and Denmark needs to support 5+ language markets, and that's just neighboring countries.
The traditional solution? Translation tools can cost upwards of €200 per year per language. But translation alone doesn't solve the search problem, it just means your product pages are readable when customers find them.
Some stores try building synonym dictionaries for each market: "sommerkleid" = "summer dress", "turnschuhe" = "sneakers", and hundreds more fashion terms per language. This requires constant maintenance as new products arrive and trends change. Each language pair needs its own dictionary. Each update multiplies the work.
Other stores try to get around this by offering fully localized versions. A separate site per country. But this only helps native speakers of that country's official language. Have you ever been abroad and gotten stuck on the localized version of a store you know well? You know they sell something but have no idea what they call it in Spain. That's the experience multilingual customers face daily on your store—except they leave instead of figuring it out.
The Solution: Search That Actually Works Across Languages
Substanz provides multilingual search out of the box—no translation of your product data, no tagging, no synonym dictionaries to maintain, no manual configuration per language.
Your German customers can search "sommerkleid" and find your summer dresses. Your Spanish customers can search "vestido de fiesta" and find your cocktail dresses. Customers shopping in their second language can search in the language they're comfortable with instead of being forced to think in your catalog's language.
Why it works for fashion:
Substanz uses AI trained specifically on fashion data to understand concepts across languages. It knows that "boyfriend jeans" and "relaxed fit" describe the same style, understands the difference between a mermaid hem and high-low hemline, and handles regional variations automatically—whether customers search for "trainers," "sneakers," or "turnschuhe."
Fashion terminology varies by region, trends constantly create new search terms, and customers naturally use their native language—especially false friends like "smoking" (German for tuxedo). Your search works for any customer in any language from day one—no need to react when you notice Spanish traffic or prepare for new markets.
Substanz currently supports: English, German, French, Spanish, Italian, Swedish, Polish, Norwegian, Dutch, Danish, Portuguese.
Key Takeaways
Why can't international customers find products on my translated store?
Traditional keyword search only works in one language at a time. Even with translation apps, a German customer browsing your English store can't search 'sommerkleid' and find 'summer dress'—the search needs exact or near-exact word matches in the same language your products are tagged in.
How many of my customers are affected by multilingual search problems?
Up to 50% of EU ecommerce traffic could be customers shopping in their second language. This includes 38% of EU shoppers who buy cross-border and 14.1% of EU residents who were born in another country. In major markets like Germany (20.9% foreign-born) and Spain (18.2%), the multilingual customer base is even larger.
Should I just build synonym dictionaries for each language?
Synonym dictionaries require constant manual maintenance. As new products arrive and fashion trends change, these dictionaries quickly fall out of date. For a store expanding to 5 languages, you'd need to maintain hundreds of fashion terms across each language pair—a never-ending game of catch-up that costs time and money.
Why is multilingual search harder for fashion stores?
Fashion terminology varies heavily by region ("trainers" vs "sneakers" vs "turnschuhe"), trends create new search terms constantly, and customers naturally use their native language terms—especially false friends like "smoking" (German for tuxedo). Bilingual customers also commonly mix languages when searching, typing "sommer dress" instead of "summer dress."
How do I fix search for customers who speak different languages?
AI-powered vector search like Substanz understands semantic meaning across languages rather than matching exact keywords. When a customer searches "sommerkleid," the AI understands the concept (summer dress) and surfaces relevant products, even though your catalog is in English. No translation of product data required, no synonym dictionaries to maintain.
Do I need to translate my entire store to make multilingual search work?
No. Vector search works regardless of whether you've localized your storefront. Even if your store is 100% in English, German customers can search in German and French customers can search in French—the search understands cross-lingual intent automatically.
Is fixing multilingual search worth the investment?
Fashion stores using Substanz see search customers spend 65% more per order and convert 6-7x better than browsers. If half your traffic is multilingual and search is failing them, you're losing your highest-value customer segment. Stores investing in localization see 20-30% revenue increases—but only if search actually works for all your customers.
Stop Losing Multilingual Customers to Bad Search
Want to see how Substanz can help you convert customers across languages, with no extra work on your side?