Original data 154 exports · 40 countries October 2026
Between July and October 2026, people in 40 countries ran 154 Google Photos Takeout exports — 1,967,805 files — through FolioSort on their own Macs and PCs. The app counts a few things about each export as it repairs it, and only those anonymous counts reach us. Here is what they show.
5,296
files in the median Takeout. A quarter of exports hold more than 13,769; the largest held 145,725.
6.6%
of files had no usable .json beside them, even in exports where pairing worked — and 4 in 5 of those exports had at least one such file.
37%
of files carried a location in Google's .json that FolioSort wrote back into the photo.
44%
of exports had faces named in Google Photos — a median of 4 names — covering 14% of all files.
1 in 6
exports contained "RAW" files (.CR2, .NEF, .ARW…) that are really JPEG or PNG inside.
18%
of the files without a usable .json still had a date in their file name — the rest have no date from Google at all.
The median export had 5,296 files. A quarter had fewer than 1,503; a quarter had more than 13,769; one in ten passed 27,019, and the largest single export in the data held 145,725 files. "Just upload your Takeout" is a different job at each end of that range — and the large ones are the ones split into several .zip parts.
Google exports one .json per photo or video with the date and place Google Photos shows (what is inside those files). In the 99 exports where pairing broadly worked, the median export had only 0.4% of files without a usable .json. But 79 of the 99 had at least one, and across all their files the share was 6.6%. For those files, the date has to come from inside the file itself — and for images that never had one (often messaging-app pictures, screenshots and scans), there is none.
About 18% of those files still had a date in their name, in a pattern such as WhatsApp's IMG-20190704-WA0012.jpg, which an organizer can read. The other 82% had no date from Google at all.
37% of all files had a location in their .json that FolioSort wrote back into the photo. The spread between exports is wide: in a quarter of them, 2% of files or fewer did; in another quarter, 59% or more. It depends on whether the photos came from a phone with location on, from a camera without GPS, or had places added in Google Photos — and those added or estimated places exist only in the .json.
44 of the 99 exports had at least one face named in Google Photos, with a median of 4 different names among them, and 14% of all files carried at least one name. Those names are not inside the photo files; delete the .json files before merging them and they are gone.
17 of the 99 exports — and 25 of all 154 — contained files with a RAW camera extension whose contents are actually JPEG or PNG: 0.37% of all files. A RAW editor will refuse or misread them. If you shot RAW and backed up to Google Photos, check what you got back before you delete anything else.
.json, how many of those have a date in their file name, files that had a location written, files with a named face, the number of distinct names, and files whose extension does not match their contents. No file names, folder paths, names of people, locations or photos are ever sent. The event is described in our privacy policy..json. The counts alone cannot tell why: the parts of a split export extracted into different folders, a folder that is not a Takeout at all, and name patterns FolioSort's matcher does not recognise yet all look the same. Those 55 are left out of every percentage about the contents; the remaining 99 exports (1,344,358 files) are what the findings describe. The file-count figures use all 154.Citing this page: FolioSort, "What Is Inside a Real Google Takeout? Numbers From 154 Exports", October 2026, https://www.foliosort.app/google-photos/takeout-export-study. The figures may be quoted freely with a link back.
| Measure | Value | Base |
|---|---|---|
| Exports / files | 154 / 1,967,805 | all exports |
| Files per export — 25th / 50th / 75th / 90th percentile | 1,503 / 5,296 / 13,769 / 27,019 | all exports |
| Largest export | 145,725 files | all exports |
| Exports where most files had no usable .json | 55 of 154 (36%) | all exports |
| Files without a usable .json | 6.6% (median export 0.4%) | 99 exports, 1,344,358 files |
| Exports with at least one such file | 79 of 99 | 99 exports |
| …of those files, date still in the file name | 18% | 99 exports |
| Files with a location written from the .json | 37% (quartiles 2% / 30% / 59%) | 99 exports |
| Exports with named faces | 44 of 99 (median 4 names) | 99 exports |
| Files with at least one named face | 14% | 99 exports |
| Exports with RAW-named JPEG/PNG files | 17 of 99 (25 of 154) | as stated |
| Files with a RAW extension but JPEG/PNG content | 0.37% | 99 exports |
.json can sit in a different .zip..json files until their dates, places, captions and names are inside the photos. For more than a third of files they hold the location, and in nearly half of exports they are the only record of who is in the pictures..json should tell you how many it could not match and why. In FolioSort's Google Photos screen that is the Orphans count, with a reason per file.FolioSort repairs a Takeout on Mac and Windows for free, with no file limit, and runs entirely on your computer. How it compares with the other Takeout tools →
FolioSort, GooglePhotosTakeoutHelper, Metadata Fixer, immich-go, PhotoMove and ExifTool on one page — including the cases where another tool is the better choice.
Every field, why the names look broken, and when you can delete them.
From requesting the export to a dated, de-duplicated library on your own drive.
The Google Photos screen, field by field: Matched, Orphans, folders and duplicates.
FolioSort pairs every photo with its .json and writes the dates, places, captions and people into the files — free, with no file limit, on Mac and Windows.