Last reviewed:
Reviewed by / developer: TrackFollows
A single Instagram export tells you one thing: who follows you at that moment. To know who left, you need an older list, and an older list only exists if you saved it at the time. This guide sets up a routine for exporting regularly and comparing two dates. TrackFollows on the web does not store the files you open, so keeping past exports and comparing them is up to you.
| Question | One export | Two exports, same settings |
|---|---|---|
| Who does not follow me back right now | Yes | Yes |
| Mutual follows | Yes | Yes |
| Followers who disappeared between two dates | No | Yes |
| New followers between two dates | No | Yes |
| Exactly when someone unfollowed | No | No. Only "somewhere between the two dates" |
| Anything before the first export | No | No |
The last two rows matter most. Comparing two files narrows an unfollow to "between October 1 and November 1", nothing finer, and nothing before your first file can ever be recovered. So the first step of the routine is not comparing; it is taking the first snapshot today.
Two files only compare cleanly if they were made the same way. Fix these four things and repeat them every time.
instagram-2026-10-28.zip. Instagram's file names carry no date, so later you cannot tell them apart.Request an export now. When the ready notice arrives, download the ZIP, add the date to its name, and open it once in TrackFollows to check that the follower and following counts are close to what the app shows. A big gap usually means the date range or the selected categories were wrong; see why the numbers differ from the app.
When the next date comes, export again with the same settings. Changing the format or date range still produces a comparison, but you can no longer tell whether a difference is a real change or a settings change.

An illustration of comparing followers lists from two dates, using sample accounts.
=COUNTIF(B:B, A2) and fill down. Column A accounts with 0 are followers who disappeared.=COUNTIF(A:A, B2), marks column B accounts with 0 as new followers.Inside the ZIP, connections/followers_and_following/followers_1.json holds one username per entry at string_list_data[0].value. Large accounts get several files (followers_2.json, …), so combine them all. In a terminal, two lines build sorted lists and show the difference:
jq -r '.[].string_list_data[0].value' followers_*.json | sort -u > followers-2026-10.txt
comm -23 followers-2026-10.txt followers-2026-11.txt # in October, gone in November
Swap in comm -13 for accounts that are new. If the structure looks different, see Inside the Instagram export ZIP.
Missing from the list does not always mean unfollowed. Open the profile and you will usually find one of four cases.
| What you see on the profile | Most likely cause | Count it as a lost follower? |
|---|---|---|
| Profile loads and the button says "Follow" | Unfollowed you | Yes |
| "User not found" | Account deleted or deactivated | No. It may come back if restored |
| Same photo and bio under a different name, listed among new followers | Username change | No. Same person |
| Profile will not open at all | They blocked you | Yes |
A username change shows up as one account gone plus one account new. If the disappeared and new counts grew by similar amounts, suspect renames first.
recently_unfollowed_profiles in the export records unfollows you made. It is not a list of people who left you, so keep it separate.following.json the same way to track what you followed and dropped. Its list sits under the top-level key relationships_following, so in Method B use the path .relationships_following[].string_list_data[0].value.I have no file from months ago. Can I still find out who left since then? No. Instagram does not provide past follower lists, and an export contains only the list at request time. Taking the first file today is the only starting point.
Can I compare HTML exports? TrackFollows can open them, but JSON is far easier to compare by hand. Keep the routine on JSON; the differences are in JSON vs HTML.
Does TrackFollows compare two files for me? The web version analyzes one file at a time and does not store it. Compare with Method A or B above.
Related guides