Native-Language Cold Email: When Localization Is Actually Worth It

Translation and localization are not the same decision

I'm Hlib Storchak. I build and run outbound systems for B2B founders and sales teams, and I've booked 2000+ meetings for B2B clients doing it, several of them selling from a single English-language base into half a dozen European markets at once. The question I get most on this is phrased as "should we translate our sequence," and it's the wrong question. Translation swaps the words for their equivalent in another language. Localization changes what surrounds the words: the formality level, the reference points, the subject line convention, sometimes the offer itself.

A cold email translated word for word from English into Polish, French, or Lithuanian usually still reads as foreign, because the directness and short-sentence style that gets replies in English doesn't map cleanly onto the local business register. That's the actual decision buried inside "should we localize": not whether to swap the language, but how much of the surrounding context needs to change with it, and whether that's worth doing at all for this specific market.

"They all speak English" is a guess, not a fact

Most cold email advice defaults to English everywhere, on the reasonable-sounding assumption that professional buyers in Europe read English well enough that a second language adds nothing. That assumption holds in some markets and doesn't in others, and the gap between them is bigger than most outbound teams assume. Treating "does my buyer read English comfortably" as a known fact rather than a market-specific question is the root of most bad localization decisions, in both directions: localizing where it wasn't needed, and skipping it where it was.

The mistake I see most often when I take over an account expanding into a new market is the opposite of what people expect: it isn't under-localizing, it's a blanket policy applied without checking, either "always English" or "always translate," carried over from whatever worked in the account's first market. The data below is what I actually check before making that call for a client, and it's public.

The proficiency data, market by market

EF's English Proficiency Index scores adult English proficiency across 123 countries and regions using standardised test results, and its 2025 edition puts these markets in noticeably different bands (EF English Proficiency Index 2025):

Two of those numbers surprised me enough to double-check them directly against EF's own country pages rather than a summary. Poland sits in the Very High band at 600, its best score yet and a 12-point improvement on the prior edition, which is why most Polish B2B outreach performs fine in plain English (EF EPI, Poland). Lithuania, despite sharing a region and a lot of business culture with Poland, sits at 543, Moderate, after a 26-point drop from its previous score (EF EPI, Lithuania). France and Spain, two of the largest B2B markets in Western Europe, are both Moderate as well. None of that means English fails in these markets. It means the safety margin most outbound teams assume, "the buyer will get it," is thinner there than it is in Amsterdam or Copenhagen, and thin margins are exactly where the copy quality of a rushed second-language send starts to matter.

A three-question framework for deciding

I use three questions with clients before committing budget to a localized track, in this order:

1. What band is the market in? Very High or High: language almost never justifies the localization cost on its own. Moderate, Low, or Very Low: localization starts to earn its keep.
2. How senior is the buyer, and how big is the deal? A junior operator with time to reread an email tolerates a second language better than a VP skimming between meetings. Raise the localization bar as seniority and deal size go up, even in a High-band market.
3. Is this a test batch or an ongoing program? A one-off list of 200 prospects rarely justifies a dedicated localized sequence and review cycle. A market you plan to run outbound into for the next year does.

If the answer to question one is Moderate or below, localize. If it's Very High or High and the answer to question two doesn't push you toward it, save the budget and spend it on tightening the copy itself instead, since a better-written English email usually beats a mediocre translated one.

What actually needs to change, beyond the words

Assuming the framework above says localize, here's what actually needs review, roughly in order of how often teams skip it:

• Formality register. English cold email tends to be direct and first-name informal. Several European business cultures, French and German among them, expect a more formal opening even in a first cold contact, and getting this wrong reads as careless rather than casual.
• Reference points. Currency, comparable company names, and industry benchmarks should match what the reader recognises, not what your home market uses.
• Subject line convention. Short, curiosity-driven subject lines that work in English don't always translate as intended; some markets read them as spammy where the English original read as concise.
• The offer, occasionally. A free trial or a specific case study logo that resonates with a US or UK buyer may mean nothing locally. Swap in a comparable one where possible instead of translating a reference the reader won't recognise.

Notice that only one of the four items above is language at all. This is why a plain AI translation of an English sequence, without a native reviewer checking these four things, usually underperforms both the English original and a properly localized version: it changes the words without changing anything that actually needed to change.

The hybrid path: English send, local-language proof

A middle option that works well for High-band markets sitting near the Moderate line, and one I use often with clients testing a new market before committing to a full localized track: send the sequence in English, but include one native-language element as social proof, a short quote from a local client, or a line naming a recognisable local company you already work with. This signals that you understand the market without the cost of translating and reviewing an entire sequence, and it's a reasonable first move before deciding whether full localization earns its budget for that specific market.

What localization actually costs, with the maths shown

Cost models here need stated assumptions, since I haven't seen a reliable published figure for "cost per localized cold email sequence" and wouldn't trust one if I had. Here's the shape of it, with every input labelled so you can swap in your own numbers:

• Assume a native-speaker reviewer, not a full translator, costs somewhere in the €150 to €400 range to review and adjust a 4 to 6 email sequence already drafted in English or machine-translated, depending on the language and how much rewriting it needs.
• Assume a second sending domain and two mailboxes for the new language track adds roughly €20 to €50 a month in infrastructure, plus the usual 2 to 4 week warmup period before it can send at volume, covered in more depth in my cold email deliverability guide.
• Assume the review cycle itself adds 3 to 5 business days to your timeline before the localized track can go live, since a rushed native review defeats the purpose.

Under those assumptions, a single localized track for one new market runs somewhere between €170 and €450 in one-off review cost plus the ongoing infrastructure line, spread across however many prospects you send to. Against a B2B deal that's worth even a few thousand euros a year, that's a small number if the market genuinely needed it, and money spent on nothing if it didn't, which is exactly why the framework above should come before the spend, not after.

Infrastructure: one domain per language, or one for all

Once you're running more than one language track at real volume, a separate sending domain per language keeps deliverability signals and reply data cleanly separated, which matters when you're trying to tell whether the Moderate-band Lithuanian version is actually outperforming the English one sent to the same segment. For a single test batch of a few hundred prospects, sending the localized version from your existing domain is fine and not worth the setup overhead.

Before you scale a language track past a test, the list itself needs to be segmented by country and language reliably, not filtered after the fact from a generic European list, the same enrichment problem I cover in my comparison of Clay vs ZoomInfo for list building. This is the stack I run for clients juggling more than one language track: Salesforge bills per workspace rather than per mailbox, so adding a second domain and mailbox set for a new language doesn't multiply the software bill the way seat-priced tools do. That's a preference based on how I run accounts, not a verdict on every alternative, several of which handle multi-domain sending perfectly well too.

Building a localized sequence, in order

1. Confirm the market's EF EPI band and run it through the three-question framework above.
2. Draft the sequence in English first, get the offer and structure right before translating anything.
3. Machine-translate as a first pass for speed, then send it to a native-speaker reviewer, not a second translator, to check formality, reference points, and subject lines.
4. Set up a dedicated domain and mailboxes if this is an ongoing program, and run the standard warmup period before sending at volume.
5. Send a small test batch, 100 to 200 prospects, and compare reply quality against an English-language control group from the same segment before rolling it out further.

Mistakes I see most often

Two mistakes account for most of the localization problems I get called in to fix. The first is sending a machine translation with no native review at all, which occasionally reads as broken enough to hurt credibility more than a plain English email would have. The second is localizing a Very High or High band market by default, out of an abundance of caution, and spending the review budget on a language barrier that didn't exist while the copy and offer, the parts that actually needed the attention, went unreviewed.

Which markets I'd localize for right now

Based on the EF EPI data above, I'd localize by default for Moderate-band and lower markets: France, Spain, Italy, and Lithuania among the ones covered here, with the caveat that Lithuania's 26-point year-over-year drop is worth re-checking against the next EF edition before treating it as settled. I would not localize by default for the Very High band markets in this list, Poland included despite the geographic instinct to bundle it with its Baltic neighbours; the copy and offer are a better use of that budget there. If you're weighing a full localized program against just hiring local expertise directly, the same due-diligence questions I lay out for vetting an outbound agency for the Nordics apply just as well to checking whether an agency actually understands a Moderate-band market's language needs, not only its compliance rules.

FAQ