How the work is done, what is checked, and what you get back.
Quality is a process, not a promise.
Every supplier says their quality is high. The useful question for a vendor manager is a different one: what exactly happens to a file between the moment it arrives and the moment it comes back. This page answers that.
How a job runs
A disciplined workflow, adapted to each assignment.
The four steps below are our default. Where your own process already sets the sequence — your onboarding, your QA model, your delivery checklist — we work inside yours instead.
01
Intake
We confirm scope, deadline and reference material: family members, earlier translations, your glossaries and style guide. If something is missing we ask before starting, not halfway through.
02
Terminology
Key terms are settled against the source and the existing assets before translation begins. Open questions come back to you as questions.
03
Translation
Work is done in your CAT environment, against your translation memory and termbase, with segment status kept clean so your own checks run properly.
04
Review and delivery
A second pass against the source, automated checks on numbers and tags, and the package returned in the form it arrived.
Machine translation and AI
Used where it helps, checked where it fails.
Machine output is fluent, and fluency is the problem: a wrong term reads as smoothly as a right one. In patent and regulatory text the errors that matter are rarely grammatical, so they survive a quick read and reach the client.
We take post-editing and AI output review as work in their own right, priced and scheduled as such, and we tell you when a file is not worth post-editing and should be retranslated. That answer costs us a job and saves you a delivery.
Full post-editing to publishable quality
Light post-editing for internal use
Expert review of AI-assisted translation
Assessment of whether a file is usable at all
The Source Examination Report
When the original is wrong, you hear about it.
Source texts contain mistakes: a reference numeral that points at the wrong feature, a unit that cannot be right, a cross-reference to a section that does not exist, a sentence that contradicts the one before it. A translator who notices and translates it faithfully has done the job as instructed and passed the problem on.
Where we find something, we deliver a Source Examination Report (SER) alongside the files: what it is, where it is, and what we did about it. You decide what to forward to the client. Not every job produces one — a clean source needs no report, and we will tell you that too. It takes us a few minutes and it has stopped errors from reaching a filing.
Working with us
Built into your workflow.
Nothing for your team to convert or reformat. We take the package as you send it and return it the same way.
| English variants | US English by default, UK on request |
| CAT environment | Trados Studio, memoQ |
| File formats | SDLXLIFF, XLIFF, TMX, DOCX |
| Assets | Your TM and termbase, returned updated |
| Quality assurance | Terminology, consistency, numbers, tags |
| Reporting | Source Examination Report where the original needs one |
| Confidentiality | NDA before any sensitive file; your framework agreement and portal respected |
Run a test file.
The fastest way to judge any of this is to give us something real. Send a short sample with its reference material and compare what comes back with what you expect.