The report always reflects the content that was available in your Archive workspace at the moment you generated it.
Our automatic detection of content is normally very fast — usually within minutes or hours of a post going live — so for past periods, the dataset should already be stable. That said, it's still possible for a later export to include posts that weren't in an earlier one, and this doesn't indicate an issue with your report.
This can happen in a couple of scenarios:
When we improve our detection mechanisms internally: Sometimes, enhancements on our side allow Archive to identify posts that existed at the time but weren't captured previously. When this happens, older posts can appear retroactively in the workspace — and therefore in newer exports.
When content is added manually at a later date: Manually imported posts are stored based on their post date, not the date they were added, so they'll also appear in newer exports for that same historical period.
Filters applied at the time of export: Differences between exports can also come from the filters used. For example, if the initial CSV was filtered to show only Instagram posts, TikTok content from the same period wouldn't appear — even though it exists in your workspace. Any variation in platform filters, date filters, or label filters will naturally change what shows up in the export.