The Scenario
You are a localization engineer. The Spanish translation memory in Crowdin has accumulated over three years of work and before you migrate to a new TMS, you need to export every source and target pair into a Google Sheet for quality review.
The goal is to find duplicates, flag inconsistencies, and remove segments that were translated incorrectly before the migration. Your QA lead will work through the sheet row by row.
The slow version:
- Export the TM from Crowdin as a TMX file
- Import the TMX into a converter tool to get a CSV
- Import the CSV into Google Sheets
- Realize the TMX export included metadata columns your QA lead does not need
- Spend an hour cleaning up the columns and removing the noise
- The QA lead gets the sheet four days after you planned to hand it off.
The fast version is one prompt.
The Easy Way: One Prompt in SheetXAI
SheetXAI is an AI agent inside your spreadsheet that pulls the translation memory segments directly from Crowdin and writes them in a clean format, so you skip the TMX → CSV → cleanup pipeline entirely.
Open the SheetXAI sidebar and type:
Export the Crowdin translation memory ID from cell A1 as a structured list. Write all English source text and Spanish target translation pairs into my sheet starting at row 2. Add a column C with the creation date for each segment.
SheetXAI calls the Crowdin API, retrieves the TM segments, and writes source, target, and creation date into the sheet. Your QA lead starts reviewing immediately.
What You Get
A TM export with one row per segment pair:
- Column A — English source text
- Column B — Spanish target translation
- Column C — creation date of the segment
The creation date column is what QA needs most. Segments from three years ago may predate your current style guide. Sorting by creation date descending lets the QA lead prioritize recent high-volume segments and flag old ones for a second pass.
If you need the TM in reverse direction (Spanish → English), tell SheetXAI to flip the source and target columns. Same prompt, different output.
What If the Data Is Not Quite Ready
TM exports for migration purposes need more than a raw dump of segments.
When the export should exclude low-confidence segments
Your TM includes machine-translated segments from an old experiment. They are flagged with a confidence score below 80 in Crowdin.
Export all TM segments from TM ID in cell A1 where the confidence score is 80 or higher. Write source, target, and confidence score into columns A through C. Label the column header row.
When the QA lead wants duplicate source strings flagged
The TM has been added to over the years and some English source strings have multiple Spanish translations. The QA lead needs to adjudicate which one is correct.
Export all TM segments from TM ID in cell A1. Write source and target into columns A and B. In column C, write "DUPLICATE" if the same source string appears more than once in the export, and "UNIQUE" if it appears once. Sort duplicates to the top.
When you need to import corrected segments back into Crowdin
After the QA pass, you have a sheet of reviewed segments ready to push back as a corrected TM.
Create new TM segments in Crowdin TM ID from cell A1 for each row in my sheet where column A is the English source text and column B is the corrected Spanish translation. Write "IMPORTED" into column C for each success.
When you need to export the TM, deduplicate it, and import the cleaned version back in one operation
The goal is not just a review file. You want a clean TM in Crowdin by end of day.
Export all TM segments from TM ID in cell A1 into a temporary tab. Identify duplicate source strings and keep only the most recently created translation for each. Write the deduplicated list to the Clean tab. Then create new TM segments in a new Crowdin TM named "Spanish Clean v2" using the source from column A and the target from column B of the Clean tab.
The pattern: export for review, clean in the sheet, push back the corrected version. The sheet is the workspace for the migration.
Try It
Get the 7-day free trial of SheetXAI and open a sheet with your Crowdin TM ID in cell A1, then ask it to export the source and target pairs. The Crowdin integration is included in every SheetXAI plan. For related workflows, see how to import glossary terms into Crowdin or the Crowdin in Google Sheets overview.