What happens if a memory has no date or location?

Then we leave it empty. The file is included in the archive, just without the details that were not in the export.

That may sound like a limitation, but it is a deliberate choice and the most important thing we do. A made-up date is worse than no date at all, because it quietly ruins exactly the chronology you paid to get back. And you only notice years later, when you wonder why a picture is in the wrong place.

The same goes for locations. Some exports contain no GPS whatsoever. Then the map stays empty, and that is correct: the memory was taken somewhere, but we do not know where, and guessing at a city would be inventing a memory for you.

There is a second reason that is less obvious, and it applies even to a completely complete export. The times and locations sit in a separate file, memories_history.json, and it does not say which photo or video each entry belongs to. We measured it on a real archive: in all 401 entries, the field for the download link was empty. What remains is date, media type and position, but no key back to the file.

Filenames only help part of the way, since they contain the day and then a random identifier. That means we can join the two sources per day and type, but not always within a day. If you have twelve snaps on a Saturday and the history lists twelve times that Saturday, there is no information about which time belongs to which picture. We tried matching them in order anyway: on an archive of 1,182 files it was right in only 39 per cent of cases, which is another way of saying it is wrong.

So we set only what we can substantiate. If there is a single memory of that type on a given day, the link is unambiguous and you get both the time and the position. If all the memories that day share the same position, they get that position, since it is correct however they pair up. If the positions differ, you get the day and nothing more. The alternative would be to guess, and a photo stamped with the wrong city is worse than one with no stamp at all.

The email you get when the archive is ready states exactly how many of your memories got a time and how many got a position. The number varies between exports, so we give it per archive instead of putting an average on the website.

The other common reason is that the JSON files were not ticked when the export was ordered. Without them we can only read the date from the filename, which gives the right day but no time and no position at all. If you notice afterwards, you can order a new export for free and run it again at no extra cost with us.

Files with no date are sorted last in the archive, in a folder of their own, with clear names. So they are not mixed in among the dated ones and do not disturb the order.

In the summary you get afterwards you can see exactly how many memories got their date and their location back. You do not have to take our word for it. You can count yourself.