Quick answer
Use Dutch for Flanders, French for Wallonia, and lean French for Brussels unless you have a specific signal otherwise. English works as a safe fallback everywhere in Belgium, since the country ranks 9th globally for English proficiency, but a single script sent to the whole country reads as "not from here" to roughly half your list, whichever one language you pick. Segment by region before you send, don't localize into one national language the way you would for France or Poland.
Belgium is not one market
I'm Hlib Storchak. I build and run B2B outbound systems, cold email and LinkedIn, for founders and sales teams, and I've booked 2000+ meetings for B2B clients doing it, several of them running campaigns into Belgium as part of a wider Benelux or Western Europe push. Belgium comes up more than its population would suggest, mostly because teams treat it like a rounding error between the Netherlands and France and then get the language wrong for half the list.
If you searched something like "cold email Belgium French or Dutch," here's the direct answer before the rest of this gets into why: Dutch for a Flanders contact, French for a Wallonia contact, French as the better default for Brussels unless you know otherwise, and English as a fallback that works everywhere but wins nowhere. A country-wide single script, in any one of those three languages, will read as correct to part of your list and slightly off to the rest.
Why high English proficiency doesn't settle it
I've written before about when native-language localization is actually worth it, using the EF English Proficiency Index as the main input: localize for Moderate-band markets like France, Spain, or Lithuania, skip it for Very High and High band markets where comprehension was never the bottleneck. Belgium sits near the top of that list. The EF English Proficiency Index 2025 put Belgium 9th globally out of 123 countries and regions, with a score of 608, up from 13th and 592 the year before, squarely in the Very High band alongside the Netherlands and Germany.
By that framework alone, Belgium would be a market where you don't bother localizing. That conclusion is correct for the comprehension question and wrong for the full decision, because Belgium isn't a single-language country whose buyers happen to read English well. It's a country built around two home languages that the English-proficiency score says nothing about, and the choice between them isn't about whether the reader will understand you.
The real split: Flanders, Wallonia, Brussels
Belgium's population sits at roughly 11.9 million as of January 2026. Split by region, the Flemish Region holds about 6.9 million people, 58% of the country, where Dutch is the official and working language. The Walloon Region holds about 3.7 million, 31%, where French is official and standard in business. The Brussels-Capital Region holds about 1.26 million, 11%, and is officially bilingual in French and Dutch, though French dominates in most private-sector business settings there (population figures from Belgium's statistical office, cited via Statbel's January 2026 regional breakdown).
That 58/31/11 split is the number most outbound teams skip past when they read "Belgium" on a target-market list and mentally file it next to the Netherlands. A Flemish company in Antwerp and a Walloon company in Liège are closer in company law and VAT rules than they are in the language their staff actually work in day to day.
| Region | Share of population | Working language |
|---|---|---|
| Flanders | ~58% | Dutch |
| Wallonia | ~31% | French |
| Brussels-Capital | ~11% | Officially bilingual, French dominant in business |
Identity beats comprehension here
The mistake I see most often when I take over a client's existing Belgium list is an assumption carried over from a market where comprehension really was the deciding factor: "their English is excellent, so language doesn't matter." That's true for whether the reader understands the email. It's not true for whether the email reads as credible. CSA Research's "Can't Read, Won't Buy" B2B survey, which polled 1,116 business buyers, found that 66% would pay up to 30% more for a product presented in a localized way (CSA Research, Can't Read, Won't Buy, B2B). That preference shows up even where the buyer could have handled the English version just fine. It's a credibility signal, not a comprehension fix, and Belgium is exactly the kind of market where that distinction matters, because the comprehension excuse for sending English by default genuinely holds up there, while the identity reason to still pick a regional language doesn't go away.
Three scripts, compared
Here's how a single national script actually performs against a properly segmented approach, based on what I've seen running both for clients.
| Approach | Where it goes right | Where it breaks |
|---|---|---|
| One English script, nationwide | Fast to build, defensible comprehension-wise everywhere | Reads as generic to both language groups, no regional credibility signal |
| One French script, nationwide | Covers Wallonia and most of Brussels well | Reads as tone-deaf to the 58% of the country that works in Dutch |
| One Dutch script, nationwide | Covers Flanders well, Belgium's largest region | Misses Wallonia and most of Brussels entirely |
| Segmented: Dutch / French / French-leaning Brussels | Matches the actual language map, best reply quality per segment | More setup: three copy variants and a region-tagged list instead of one |
The segmented approach isn't more work than most teams assume, because the offer, the proof points, and the structure stay identical across all three versions. What changes is the opening line, the formality register, and the sign-off, not the whole sequence.
Tip. Don't translate one script into three languages and call it segmented. Write the English version first to lock the offer and structure, then have each regional version reviewed by someone who actually works in that market, the same way you would for any other localized sequence.
Segmenting the list by region, not by guess
The practical blocker isn't deciding on three scripts, it's tagging a list by region reliably before you send. Company head office city is a weak signal on its own, since plenty of Walloon companies have a Brussels-listed registered address. The faster, more reliable signal is the postal code: Belgian postal codes run roughly 1000 to 1299 for Brussels-Capital, 1300 to 1499 and 4000 to 7999 for Wallonia, and 1500 to 3999 plus 8000 to 9999 for Flanders. It won't be perfect at every border town between regions, but it's accurate enough to drive which script a contact gets, and far better than defaulting the whole list to one language because tagging felt like extra work.
If you're building the list from scratch rather than segmenting an existing one, this is the same enrichment problem I cover in my comparison of Clay and Apollo for B2B list building: pull the postal code or region field at build time, not after the fact, because retrofitting region tags onto a list that's already been scored and sequenced is the slower way to do the same job.
The Brussels exception
Brussels is the one region where "just use the local language" doesn't resolve to a single answer, because it's officially bilingual and genuinely mixed in daily use. In practice, French is the stronger default for most private-sector business correspondence in Brussels, with Dutch more common in specific public-sector and some corporate contexts. Absent a stronger signal, French with an English fallback line is the safer call for a Brussels contact than Dutch, though this is the one region in the article where I'd adjust fastest the moment a client's own reply data said otherwise.
What actually changes beyond the opening line
Once you've split the list, the changes that matter go beyond swapping the greeting:
- Formality register. Flemish business culture tends to be more direct and informal than Walloon French, which leans closer to the more formal register common in France. A Dutch version can open more casually than its French counterpart to the same job title.
- Company legal form. A company registered in Flanders uses "BV," the same legal form a Walloon company calls "SRL." Likewise "NV" in Dutch and "SA" in French refer to the same entity type. Using the wrong one in a reference to the recipient's own company is a small detail that reads as a sender who didn't check.
- Proof points. A client logo or case study well known in Flanders may mean nothing in Wallonia, and the reverse. Swap in a regionally recognizable reference where you have one, rather than keeping one national proof point across both versions.
The opt-in question, briefly
This article is about language, not compliance, so I'll keep this short and point you elsewhere for the detail. Belgium's Code of Economic Law treats email advertising as opt-in by default, with a long-standing exception for addresses of legal persons rather than named individuals. A Brussels Market Court ruling from December 2025 added real uncertainty here, finding that business contact data pulled from Belgium's Crossroads Bank for Enterprises or bought lists doesn't automatically carry a "legitimate interest" basis for cold outreach, and referring questions to the EU's top court rather than settling it outright. If you're weighing Belgium against other EU markets on legal risk rather than just language, I've covered the broader picture in what's actually legal for B2B cold email across the EU, and the country-by-country contrast in Germany's much stricter opt-in rule is a useful comparison point for how much this varies next door.
What running two language tracks actually costs
I haven't seen a reliable published figure for the specific cost of a two-language Belgium program, so here's the shape of it with every assumption labelled, the same way I'd build it for a client:
- Assume a native-speaker reviewer for each regional version, Dutch and French, costs roughly €150 to €350 to review and adjust a 4 to 6 email sequence already drafted in English.
- Assume you don't need two separate sending domains for this specific case, since both versions can run from the same domain and mailboxes with the list tagged by region, unlike a program spanning entirely different countries.
- Assume the review cycle for both versions adds 3 to 5 business days before either can go live, since a rushed native review on either side defeats the point of segmenting in the first place.
Under those assumptions, splitting one Belgium campaign into Dutch and French versions runs somewhere between €300 and €700 in one-off review cost, against a buyer base that's otherwise easy to reach given how connected and well-documented the Belgian B2B market is. Swap in your own numbers for sequence length and reviewer rate before treating this as a quote.
The cadence and channel mix don't change
One thing that doesn't change once you've split the list by language: the follow-up cadence and whether to bring in LinkedIn alongside email. I've written separately about the follow-up rule that beats sending more, and when to mix in LinkedIn and calls, and that logic applies identically to a Dutch-language Flanders sequence and a French-language Wallonia one. Segmenting by language affects what the message says, not how many times you say it or which channel you add it to.
Mistakes I see most often
The most common one isn't picking the wrong language, it's picking no language at all, defaulting to English for the whole country because the proficiency data makes that defensible, and then wondering why reply rates sit noticeably below a comparable campaign into the Netherlands, a genuinely single-language Very High band market. The second is treating Belgium as "basically France" and running a French script nationwide, which works for Wallonia and loses Flanders, the larger of the two regions, entirely. The fix in both cases is the same: tag the list by region before you write a single line of copy, not after the first campaign underperforms.
Key takeaways
- Belgium splits roughly 58% Dutch-speaking Flanders, 31% French-speaking Wallonia, and 11% officially bilingual, French-leaning-in-business Brussels.
- Belgium's EF EPI 2025 score, 608 and 9th globally, means comprehension isn't the issue the way it is in France or Lithuania, but identity and credibility signalling still favor the regional language.
- Segment the list by postal code before writing copy: roughly 1000 to 1299 Brussels, 1300 to 1499 and 4000 to 7999 Wallonia, 1500 to 3999 and 8000 to 9999 Flanders.
- What changes between regional versions is formality register, legal-form naming (BV vs SRL, NV vs SA), and proof points, not the offer or structure.
- A December 2025 Brussels Market Court ruling added real uncertainty to relying on legitimate interest for scraped or bought Belgian contact data, get current legal advice before leaning on the old exception.
- Follow-up cadence and channel mix don't change by language segment, only the message itself does.
FAQ
Should I send cold email in Belgium in French, Dutch, or English?
It depends on where the contact sits. Dutch for Flanders, French for Wallonia, and lean French for Brussels unless you know the person's actual working language. English is a defensible fallback everywhere in Belgium given the country's very high EF English Proficiency Index score, but it's the safe middle option, not the best-performing one, because it skips the regional-identity signal a native-language line sends.
Does Belgium's high English proficiency mean I don't need to localize at all?
Not entirely. High proficiency removes the comprehension problem, which is the main reason to localize in a market like France or Lithuania. It doesn't remove the identity problem, which is specific to a country split roughly 58% Dutch-speaking and 31% French-speaking. A Flemish reader getting a French opening line reads it as a sender who didn't check, even if their English and French are both fluent.
What language should I use for Brussels specifically?
Brussels is officially bilingual, but French is the dominant working language in most private-sector business settings there, with Dutch stronger in specific public-sector and some corporate contexts. Without a specific signal otherwise, French is the better default for a Brussels business contact, with English as the safe fallback for anyone in an international or EU-adjacent role.
How do I segment a Belgian prospect list by language region without guessing?
The postal code is the fastest signal: Belgian postal codes run roughly 1000 to 1299 for Brussels-Capital, 1300 to 1499 and 4000 to 7999 for Wallonia, and 1500 to 3999 plus 8000 to 9999 for Flanders. It isn't perfect at every border town, but it's accurate enough to split a list before you send, which beats guessing from a company's head office city alone.
Is cold email even legal in Belgium?
Belgium's Code of Economic Law sets email advertising as opt-in by default, with a long-standing exception for addresses of legal persons rather than named individuals. A Brussels Market Court ruling from December 2025 added real uncertainty around relying on legitimate interest for contacts sourced from the Crossroads Bank for Enterprises or bought lists, with questions now referred to the EU's top court. Get current legal advice before you rely on the old exception, and treat this as a moving target, not a settled rule.