Most language companies have a commercial incentive to tell you that everything you publish needs their most premium service. We think that's backwards, and it's cost us work to say so. But telling a client the truth about what a job actually needs is the fastest way to earn the right to the work that really matters.

So here's the honest guide we'd give you, free of charge: how to tell when plain translation is enough, and when spending more is not just worth it but necessary.

Start with one question: what does it cost if this is wrong?

Almost every sensible routing decision comes back to this. Not whether this is important content (all of it feels important), but specifically: if a sentence here lands slightly wrong, what actually happens? The answer sorts your content into three honest buckets.

Bucket 1: Translation is genuinely enough

For high-volume, low-stakes content where the job is to be accurate and clear, plain translation, increasingly machine translation with a human check, is the right call. Paying for creative adaptation here is spending money for no return.

When plain translation is the right answer

Internal knowledge-base articles. Product spec tables. Support documentation. High-volume catalogue data. Patch notes. If it needs to be understood rather than felt, and a small slip is a minor inconvenience rather than a risk, don't overspend on it.

Bucket 2: Localisation earns its cost

When content is customer-facing and the experience matters, you need more than accuracy. It has to feel native: the tone, the format, the local references, the way people there actually phrase things. This is where cutting corners quietly costs you conversion.

When to pay for localisation

Your website. Your app. Onboarding flows. Product pages that have to convert. Anything a customer judges you by. Here the difference between understood and trusted is worth paying for, because it shows up in the numbers.

Bucket 3: Transcreation, or don't bother

For creative work, a literal translation often destroys the very thing that made it work. A tagline, a campaign, a brand line: these have to recreate the effect, not the words. If you're not willing to invest in that, you're often better running the campaign in the source language than shipping a flat translation of it.

When only transcreation will do

Taglines. Flagship campaigns. Brand narrative. Anything built to persuade or move people. A perfectly accurate translation of a great tagline is usually a bad tagline.

A partner who only ever recommends the expensive option isn't a partner. They're a supplier protecting their margin.

The mistake that costs most: treating it all the same

The single most expensive habit we see is applying one standard to everything: either running premium creative adaptation on content that only needed a machine pass, or, far worse, running cheap automated translation on the regulated or brand-critical content that could genuinely hurt you. The money isn't in doing everything to the highest standard. It's in matching the level of care to the stakes, piece by piece.

What this means for choosing a partner

When you're evaluating a language partner, ask them where they would not recommend their premium service. If they can't answer, or won't, that tells you something. The good ones will happily point you to a cheaper route for the content that doesn't need them, because they're confident about the work that does.