HELP
FAQ
Short answers to the questions people ask most — about EasyQLab in general, and about using it day to day.
About EasyQLab
What is EasyQLab?
A quality-control tool for translated, laid-out documents. You upload the original and the translations; it pairs every paragraph, cell and box, checks text, format and layout, marks each finding on the page and gives a verdict per language and per project.
Do I need the PDF?
No. Word, PowerPoint and Excel are checked on their own. With a PDF, eleven more layout checks run (margins, collisions, hidden text, text boxes…) because they need the drawn page. If you have both the PDF and the Word, upload both: pairing is then instant and exact.
Which file types work?
PDF, .docx, .pptx and .xlsx. The original and the translations must be the same type. Scanned PDFs and PDFs with outlined text are compared by images and colour only.
Which languages?
69 languages, with English as the source. Most have full support; right-to-left scripts, Chinese, Japanese, Korean, Thai and complex scripts have partial support: everything is checked, but the checks that depend on line breaks are indicative. The full list is in the docs.
How long does it take?
About 40 seconds per page, in parallel across languages: two pages in eight languages take about two minutes; a 60-page document in one language, about seven. The checks that don’t use AI take under a second.
How accurate is it?
In our benchmark against a full manual review (163 errors over 109 pages of two real projects), EasyQLab found 100 % of the serious errors and 9 in 10 overall. It doesn’t judge style or tone: that stays with your reviewer, who now knows where to look.
Does it use AI?
Yes, where it helps: reading the pages, reviewing the translation, pairing paragraphs and answering your questions. Rules handle the rest. We can change the AI model at any time without changing what you see.
What if a finding is not an error?
Mark it “Not an error”. It leaves the count and the PDF report (the CSV keeps it, marked as ignored), and EasyQLab lowers its confidence in that kind of finding for your next projects. “Real error” does the opposite.
Who can see my files?
Your organization, and nobody else. Every project, file and setting is tied to your organization.
How much does it cost?
It depends on volume and on how many languages you ship. Request a demo and we’ll quote per project.
Can I check the corrected file?
Yes. “Second round” compares the corrected translation with the previous one: resolved, still there, new. Only that language is billed.
Using EasyQLab
What does “partial support” mean?
Some languages break lines differently from Latin-script languages: right-to-left scripts, languages without spaces between words, and complex scripts. They are marked “partial support ⓘ” on the screen.
They are still analysed, and their findings count in the result. Only the checks that depend on where a line breaks — orphan words, text volume, margins — should be taken as a hint. See Languages.
Why does a language show “Incomplete”?
One or more languages could not be analysed, and the reason is shown for each one. The verdict only covers what was analysed, so the project can’t be approved: it shows Needs review with “Incomplete: one language was not analysed”.
Fix what the reason says and run those languages again to complete the project. See The traffic light.
Why did the number of findings change after I moved the sensitivity slider?
It shouldn’t, for a saved analysis. A saved analysis keeps the sensitivity it ran with, and says so: “Analysed at 60% sensitivity…”. A new setting applies to new analyses only, so an approved project doesn’t change verdict because someone moved the slider.
What does change the numbers is your review: marking Real error on a doubtful finding, Not an error, ignore, or the Show translation review (N) switch. The sensitivity can only be changed on some plans.
Can two people review the same project?
Yes, if they are in the same organization. Everybody in an organization sees all its projects. Decisions, ignored findings and comments are saved with the project, and Review complete shows who reviewed each language and when.
Can I re-run a project?
Yes. Open the project and start a New analysis. Each run is numbered (“Run number 2 of this project”), and each one is recorded in Billing.
A second run of the same files gives the same result: the AI reading of an identical page is reused. A project marked as finished can’t run new analyses until you click Reopen project.
Why is my file left out?
The most common reasons:
| Reason | What to do |
|---|---|
| The language was not recognised from the file name | Pick its language from the list on the screen, or rename the file with the language in it (for example _es-US). |
| It is not the same type as the original | Upload the original and the translations as the same type: both PDF, both Word, both PowerPoint or both Excel. |
| The PDF is password protected | Ask for a copy without a password. |
| The file is too big | The limit is 50 MB per file. |
| Two translations have the same language | Each language goes once per analysis: remove one, or run the other separately. |
What happens if the page says “N pages not read”?
The AI reading of a project has 15 minutes, and reads up to 60 pages per document. Pages it didn’t reach are reported as “N pages not read” instead of making you wait.
The rest of the checks ran on every page; on those pages the AI review of the translation and the AI visual review are missing. The analysis is final. To have them read, run the analysis again, or check those pages by eye.
Does the Quick analysis save anything?
No. A Quick analysis doesn’t store your files, and the result is lost when you leave the screen: it can’t be reopened. It is billed all the same, by characters processed. If you need to come back to it, create a project.
Results and review
Where do I see why a project needs review?
Next to the verdict, at the top of the results. The reason names the languages that decide it, for example “Needs review: 2 languages need review (French, Korean)”. Each language card and each row of the Summary tab has its own verdict; hover over a verdict in the Summary to read its reason. See The results screen and How the project gets its verdict.
What is the difference between “ignore” and “Not an error”?
Both take the finding out of the count and out of the annotated PDF, and both keep it in the CSV marked “ignored”. Not an error also says the finding was wrong, so EasyQLab learns from it; ignore says it was right but doesn’t matter this time. See Ignore or Not an error?
Why are some findings hidden as “doubtful”?
EasyQLab isn’t sure enough about them: they are below the confidence cut-off (60 %, or 50 % for linguistic findings). They don’t count, but the list says how many there are and you can show them with show ignored. Click Real error on any that is real and it counts. See Confidence.
Can I see only the layout findings, without the translation review?
Yes. Turn off Show translation review (N) in the results header: the findings of the AI review of the translation leave the cards, the traffic light and the summary. Turn it on to bring them back. The choice is saved with the project. See The translation review switch.
What is in the findings CSV?
One row per finding for every language of the analysis, with the language, file, pages, check, category, severity, title, explanation, source and target text, suggestion, state, the reviewer’s note and who wrote it, and whether a person wrote or edited it. Doubtful findings are left out. See The columns, one by one.
How do I compare EasyQLab with my reviewer’s marks?
Open the language and click Upload the reviewed PDF with the PDF your reviewer returned, comments and all. You get what both found, what EasyQLab missed and what only EasyQLab flagged, mark by mark. See Measure against the human review.