Working with large files

Very large files — the kind that make desktop CAT tools struggle — are handled with dedicated analysis and filtering tools rather than brute force. The platform checks file size and structure before processing and adapts how the file is loaded, split, and edited.

What happens when you upload a large file

A Large File Detected dialog shows what you're dealing with — file size, word count, estimated segments, and estimated browser memory — and recommends how to proceed:

• Load with optimized settings (recommended) — progressive loading and reduced per-segment data.
• Continue with standard settings — if you know the file is manageable.

Example: a large JSON localisation file

JSON exports are the classic case — a product's full string catalogue in one file, where only some keys need translation. Two tools keep it workable:

• Key filtering — the filter tool discovers the file's unique keys and suggests what to include or exclude. Build include/exclude lists of keys and wildcard patterns (e.g. include *.title and *.description, exclude *.id), then Preview Filter Results to see the filtered word and segment counts — and the reduction percentage — before committing. Filtering out IDs and config values routinely cuts the translatable volume by an order of magnitude.
• Splitting — break a huge JSON into smaller files that are processed in parallel and can be distributed across linguists.

For Word, Excel, and PowerPoint, the equivalent is selective extraction: restrict extraction to specific styles, columns, rows, or slide elements.

In the editor

Large files stay responsive: only visible rows are rendered, segments load progressively as you scroll, and memory is cleaned up automatically on long sessions. Turbo mode skips loading QA data for maximum speed on the largest files.