Data

Multilingual reviews on your website: which ones to show whom

20 September 2026 · JustReview

If your site sells in more than one country, your reviews are already multilingual whether you planned for it or not. We counted the language of every review in our database on 2026-09-20: 1,392,773 reviews across 3,756 accounts, written in 156 languages. Among accounts with at least 25 reviews, 63.6% hold two or more languages, 37.8% hold three or more, and one account holds 69. The practical question is therefore not whether to handle several languages on a website, but what a Dutch visitor should see when 80% of your proof is in English.

The answer depends on a number most widget setups never check: how many reviews you actually have in the second language. In our data the median is 14. Filter a page down to its own language without looking at that number and you replace a busy widget with an almost empty one.

How this was measured

Every review we import carries a language code, set when the review is pulled from its source. We read summary counts only, grouped by account, language, source and rating. No review text was read and no account is named.

The base is every review in production with a language code on 2026-09-20: 1,392,773 of 1,392,832 rows, so 59 reviews are untagged. Where the figure concerns an account rather than a review, the set is the 2,259 accounts holding at least 25 reviews, because a ten-review account flips between “one language” and “three languages” on a single import. Our own demo account is excluded from the per-account figures. Reviews hidden in a widget are still counted, since they exist on the source platform.

One caveat on what this sample is. These are businesses that connected a review widget, and they skew toward English-speaking and European markets. Read the table below as the language mix of review sets on European small business sites, not as a map of world languages.

What language your reviews are actually in

Those nine languages cover 97.9% of everything we hold. The remaining 147 languages together account for roughly 32,000 reviews, which is fewer than a single mid-sized marketplace account. That matters for how you build: a language switch with nine options covers almost every visitor, and the long tail is better served by a fallback than by nine more empty widgets.

Notice that the ranking of languages and the ranking of markets are not the same thing. Polish is second by volume but only fifth by number of accounts, because Polish accounts tend to connect Allegro, and every one of the 209,658 Allegro reviews we hold is in Polish.

Most review sets are multilingual, including yours

Among the 2,259 accounts with at least 25 reviews, 36.4% are single-language, 25.7% have two languages, 15.2% have three, and 16.7% have five or more. Raise the bar to 100 reviews and 72.7% are multilingual with a median of three languages. The more proof you collect, the less likely it is to be in one language.

The mix is very uneven by market. Sorted by how mixed the review sets are, the market with the most language spread comes first.

Czech sits out of order because 12 accounts is a small sample; treat it as a direction, not a benchmark. The two ends of the table are the useful part. A Spanish, Portuguese, German, Dutch or Czech business almost never has a single-language review set, and typically a quarter of its proof is in some other language. An English-first business is the opposite: half of those accounts are purely English and the median one has nothing else at all.

So the assumption built into most widget installs, one site, one language, one wall of reviews, is roughly correct in English and wrong nearly everywhere else in Europe. Across all multilingual accounts the median main language holds 80.3% of the reviews, the middle half between 63.9% and 96.1%. Put differently, 40.3% of accounts with real review volume have more than one review in ten sitting in a language their main page does not speak.

The second language is almost always English

For every non-English market we measured, the runner-up language is English, and it is not close: 116 of 121 multilingual Polish accounts, 130 of 135 German, 90 of 93 Spanish, 65 of 67 Dutch, 58 of 60 Italian and 58 of 62 French have English as their second language. Among the 714 multilingual English-first accounts, the second language is Spanish for 152, German for 93, French for 76 and Italian for 47.

This has a direct consequence for setup. If you run a German site with an English version, the two versions cover 130 out of every 135 German accounts’ languages between them. You do not need a per-language widget for each of nine markets; you need the page language plus English, which is exactly the two-setting combination described in our guide to the filter by page language function.

What is left on screen after you filter

Here is the number that decides whether filtering helps or hurts. On multilingual accounts, the second language holds a median of 14 reviews, with the middle half between 4 and 39. In 40.4% of multilingual accounts the second language has fewer than 10 reviews, and in 28.3% fewer than 5. The third language, where there is one, has a median of 2 reviews, and 72.6% of third languages sit below 5.

A widget showing 4 reviews reads as a business nobody talks about, even when the same business has 600 reviews one page across. That is the trap of a strict language filter, and it is also the single most common reason a widget “goes empty” after the setting is turned on. Three ways out, in order of how often they apply:

1. Keep English alongside the page language. The Include English setting exists for exactly the 14-review case, and given that English is the second language nearly everywhere, it fills the gap for most sites.
2. Check what language your pages actually declare. The filter reads the lang attribute of the html tag. A CMS that stamps lang="en" on every page, including the German one, will serve English reviews everywhere and look broken. A page declaring lang="de-AT" is matched as German.
3. Pin the language manually with an ISO code when your markup is unreliable, or on one page that differs from the rest of the site.

If your counts still look wrong after this, the mismatch may not be about language at all; our breakdown of why review counts do not match between a platform and a widget covers the other causes.

No, French customers are not harsher

A tempting second use for language data is to compare markets by rating. Across all sources, the average rating of the reviews we hold runs from 4.27 in French to 4.95 in Polish, with German at 4.56, Italian at 4.48, Spanish at 4.48, English at 4.67, Portuguese at 4.74 and Dutch at 4.77. Only 67.4% of French-language reviews are five stars, against 95.8% of Polish ones. It looks like a national-character story, and it is not.

Restrict the comparison to a single source, Google, and the spread between those same eight languages collapses from 0.67 stars to 0.13: German 4.61, Dutch 4.64, French 4.64, Spanish 4.65, Polish 4.71, English 4.72, Italian 4.72, Portuguese 4.74. The differences in the first paragraph were a source effect. French-language reviews on amazon.fr average 4.09, while French-language reviews on Google average 4.64, a gap of 0.56 stars within one language.

The practical takeaway: when your German page shows a lower average than your English page, look at which sources feed each language before you conclude anything about the market. The same reasoning applies to overall averages, which we went through in our state of online reviews for 2026.

Your national marketplace is not a national language

The last assumption worth killing is that a country-specific source produces country-language reviews. It does not, because marketplaces show reviews of a product across their European storefronts.

Sorted by how national each marketplace actually is, amazon.es comes last by a distance: fewer than one review in five on the Spanish storefront is in Spanish, and there are more English reviews there than Spanish ones. Connect it expecting a Spanish wall of proof and the language filter on your Spanish page will discard 82.5% of what you imported. The same mechanism is why our guide to using Amazon reviews on your own site recommends checking the mix before wiring the widget to a single storefront.

Sources differ enormously in how many languages they carry. Google reviews in our database span 139 languages and eBay 98, while Booksy carries 2. A local services business built on Booksy has a language problem it can ignore; a shop built on marketplaces does not.

Setting this up in JustReview

Three reasons this is worth doing in our widget rather than by hand. First, the filter runs on the original text, so nothing is machine-translated and the customer’s words stay the customer’s words. Second, the page language is read automatically from the html tag, so one snippet works across every language version of the site instead of one widget per market. Third, the Include English fallback is built in, which given a median second language of 14 reviews is the setting that keeps the widget from looking abandoned.

You can combine it with a pinned ISO code when your markup is unreliable, and it works the same way whether the reviews come from a Google profile, a marketplace or a booking platform, which matters because the language mix problem is at its worst when several sources meet in one widget. Two limits to be clear about: the language filter is available on the Pro plan and above, on the Testimonials and Page widget types, and like every widget setting it is stored in the snippet, so you have to regenerate the code and replace it on your site or nothing changes. Plans and what each includes are on the pricing page.

If you run more than one language version, the setup is a five-minute job: create an account, connect your sources, check the language split in All reviews, then decide per version whether the page language alone is enough or whether English should ride along. Businesses running several locations as well as several languages will want to read our notes on review widgets for multiple locations before choosing a plan, since the two limits interact.