Machine Translation Post-Editing for E-commerce: When, How, and How Much

Most translation conversations in e-commerce frame it as AI versus humans. The reality at scale is a third option: machine translation post-editing — running content through AI first, then having a human review and refine the output. It's faster than full human translation, more polished than raw AI, and for catalogs above a few thousand SKUs, it's often the only economically defensible path to translation that doesn't read like translation. Here's the working merchant's guide to MTPE.

What Is Machine Translation Post-Editing?

Machine translation post-editing (MTPE) is the practice of running content through a machine translation engine (GPT-4o, DeepL, Google Translate) and then having a human linguist review, correct, and refine the output. The human's job isn't to translate from scratch — it's to fix terminology, smooth awkward phrasing, and catch the predictable failure modes of AI translation. Done well, MTPE delivers near-human quality at a fraction of the time and cost of full human translation.

Light Post-Editing vs. Full Post-Editing

Light post-editing fixes only what's wrong — wrong terminology, broken grammar, mistranslated brand names. The output stays close to the machine output and is acceptable for descriptive catalog content where shoppers care about facts, not style. Full post-editing brings the output to publication quality — every sentence reads as if a native speaker wrote it from scratch. Use light for the long tail of product specs and shipping policies; use full for landing pages, marketing copy, and anything customer-facing that carries brand voice.

When Does MTPE Make Sense for E-commerce?

MTPE makes sense whenever your content volume is too large for full human translation but your brand is too established to ship raw machine output. The threshold for most stores: above 50,000 words per language and below mid-six-figure budgets, MTPE is the economically rational choice. Below 50,000 words, full human is affordable for the moments that matter. Above six figures, you're at enterprise scale where dedicated language teams enter the picture. The middle is where MTPE wins.

How Much Does MTPE Cost?

Industry rates for MTPE typically run 40–60% of full human translation rates, depending on language pair and the quality of the source machine output. Concretely: if a human translator charges $0.15 per word, the same translator doing post-editing on AI output charges $0.06–$0.09 per word. Add the cost of the AI translation itself ($0.02–$0.04 per word through a service like Lokalize) and your effective rate is $0.08–$0.13 per word — roughly half the cost of full human translation for output that approaches human quality.

Numbers above are illustrative ranges; rates vary by language pair, translator experience, and the quality of the source machine output.

How to Set Up an MTPE Workflow on Shopify

The clean workflow looks like this. First: bulk-translate the catalog with AI through Lokalize, Weglot, or another translation app. Second: build a glossary before the translation runs so brand terms and product names are correct from the start. Third: export the AI output (most apps support CSV or XLIFF). Fourth: hand the export to a human linguist with clear instructions — light or full editing, what to prioritize, what to leave alone. Fifth: re-import the edited translations back into the app. Sixth: spot-check the live storefront before publishing.

Where to Find Post-Editors

Three options. Translation agencies (Lionbridge, RWS, TransPerfect) offer enterprise MTPE workflows at predictable rates. Mid-market platforms (Smartcat, Lokalise, Crowdin) connect you to vetted freelance linguists with built-in MTPE tooling. Direct hire on ProZ or LinkedIn is the cheapest path but requires you to manage quality yourself. For most Shopify merchants, mid-market platforms are the right balance of cost and reliability — agencies are overkill below seven-figure localization budgets.

What Does a Good Post-Editor Do?

A good post-editor catches predictable AI failures: brand names translated literally, technical terms with wrong sense, awkward phrasing that sounds machine-generated, formatting and unit conversions, idioms that don't carry across. They don't rewrite acceptable AI output to flex their skills — that's wasted budget. The mark of a strong post-editor is that they leave 70% of the AI output untouched and surgically fix the 30% that needs it.

Is MTPE Better Than Full Human Translation?

For most e-commerce content, MTPE delivers comparable quality at significantly lower cost — that's the whole point of the workflow. For brand-critical copy (taglines, hero copy, campaign creative), full human translation still produces better output because the linguist works from the source language without being anchored to a machine's first draft. The right model is hybrid: MTPE for the catalog, full human for the homepage and top landing pages.

What's the Difference Between MTPE and 'AI + Glossary'?

AI plus glossary is software-only — the translation engine respects your brand and product term rules, but no human reviews the output. MTPE adds a human review pass on top. AI plus glossary is what most Lokalize merchants run for catalog content; MTPE is the upgrade path for stores that want catalog-quality polish without paying for full human translation. They're not opposed — most MTPE workflows start with AI plus glossary as the input.