The export file index. Published 2026-08-28. Last checked 2026-08-28

Which file in your data export holds what

Checked 2026-08-28: your watch history, your searches and the conclusions each platform drew about you sit in specifically named files inside the export zip. This index gives the literal file name, the literal key or column inside it, and the export tier that ships it, for ten platforms.

Both are free under CC BY 4.0, and both carry a permanent id on every file, so you can cite one filename instead of the whole page. Cite the index as: Pupul, The data export file index, pupulcorp.com/data/files, checked 2026-08-28. Cite one file by its id, for example pupulcorp.com/data/files#file-spotify-inferences.

If the export is already in your downloads folder, the free Count opens the files named in this index, in this browser. Nothing is uploaded, and there is no account or email field.

Why this index exists

A data export is not one file. It is a zip of many, and on every platform the part you actually came for lives in one or two of them, under a name nobody tells you.

Guides tell you how to request an export. Almost none tell you what to open once it lands. So the zip sits in the downloads folder, and the moment passes. The fix is small and specific: the name of the file, and the name of the key inside it.

The index

PlatformFileWhat it holdsKeys or columns insideTier you must requestStatus
TikTok #user_data_tiktok.json
older: user_data.json
Watch history, searches, likes and saved videos, and off-TikTok activity other companies reportedVideoList; SearchList with Date and SearchTerm; ItemFavoriteList; OffTikTokActivityDataList with SourceAll data, JSON formatin our parser
Instagram #your_topics.jsonThe interest topics Meta uses to pick your adstopics_your_topicsAds information, or All informationin our parser
Instagram #advertisers_using_your_activity_or_information.jsonAdvertisers who uploaded a customer list your account matchedig_custom_audiences_all_types, custom_audiences_all_types, has_data_file_custom_audienceAds information, or All informationin our parser
Instagram #ad_preferences.json
other_categories_used_to_reach_you.json #
Ad preference values and further categories advertisers used to reach youlabel_values, each holding a vec or dictAds information, or All informationin our parser
Instagram #liked_posts.json
likes_and_reactions_1.json
Your likes and reactionslikes_media_likes, likes_comment_likes, page_likesYour activityin our parser
YouTube #watch-history.json
or watch-history.html
Every video in your watch history, with the channeltime; titleUrl; subtitles[0].name for the channelYouTube and YouTube Music historyin our parser
Google #search-history.json
or Search/MyActivity.json
Everything you typed into Google searchtimeMy Activity, Search selectedin our parser
Spotify #StreamingHistory0.json
Streaming_History_Audio_2024_1.json
Every play, its length, and whether the file marks it skippedts or endTime; ms_played; master_metadata_album_artist_name; master_metadata_track_name; skippedAccount data, or Extended streaming history for the whole account lifein our parser
Spotify #Inferences.jsonThe marketing segments Spotify has filed you underA list of segment strings, written in Spotify's own wordsTechnical log information, a separate requestin our parser
X #data/personalization.jsInterests X assigned you, interests supplied by advertising partners, the advertiser audiences you were added to by company name, television shows, and inferred languagesp13n.interests.interests[].name; partnerInterests; audienceAndAdvertisers; shows; p13n.demographics.languagesFull archivein our parser
X #data/like.js
data/ad-impressions.js #
Posts you liked, and the ads you were shown with the advertiser behind eachLike records carry no timestamp. Impression records carry an advertiser nameFull archivein our parser
ChatGPT #conversations.jsonEvery conversation, with per-message timestampsA mapping object per conversation; create_time; titleFull exportin our parser
ChatGPT #chat.html, user.json, message_feedback.json, model_comparisons.jsonA browsable copy of your chats, basic account details, your thumbs up and down, and A/B comparisons you took part inNot read by our CountFull exportreported
Claude #conversations.jsonEvery conversation with timestamps. Same file name as ChatGPT, different shape insideA chat_messages array; created_at; name for the titleFull exportin our parser
Grok #prod-grok-backend.jsonYour grok.com conversations. Grok used inside the X app lands in the X archive insteadA conversations arrayFull export from accounts.x.aiin our parser
LinkedIn #Ad_Targeting.csvThe targeting attributes advertisers can select you byOne header row of attribute names, one row of your values, semicolon separated inside a cellThe larger archive, not the fast file selectionin our parser
LinkedIn #Inferences_about_you.csvConclusions LinkedIn drew about you, and whether each one was inferredAn inference column and a description columnThe larger archive, not the fast file selectionin our parser
Amazon #Advertising.AdvertiserAudiences.csv
Advertising.AmazonAudiences.csv #
Advertisers who added you to an audience by name, and the segments Amazon built itselfOne advertiser or audience per rowThe Advertising category, or Request All Your Datain our parser
Amazon #Retail.OrderHistory.1.csvYour order history, with dates and product titlesAn order date column and a product title columnYour Ordersin our parser

Each platform name links to the export guide with the exact taps for the tier in that row. The # beside each platform links to that one row, and the second file in a paired row has a # of its own. Last checked 2026-08-28.

Cite one file, not the whole page

The useful unit here is one filename. Every file has a permanent id, and the id is the link. The claim that Spotify's marketing segments sit in Inferences.json is pupulcorp.com/data/files#file-spotify-inferences, and it will still be that tomorrow.

What the status labels mean here

In our parser, reported, and the difference

In our parser means this file name and these keys are what our free Count matches and reads. You do not have to take our word for that. The parser is plain JavaScript served at pupulcorp.com/count.js, the file matching lives in its detector table near the end, and you can read exactly what it opens before you drop anything on the page. It is the strongest label we can honestly apply without publishing an export of our own.

Reported means it comes from the platform's published documentation or from consistent public reports, checked on the date shown, and we have not read it with our own code. Platforms rename files without telling anyone, so a row can go stale between checks. If one has, tell us and it gets fixed rather than quietly rotting.

Three places this trips everyone up

Meta renames the folders

Instagram and Facebook exports move their folders around more than any other platform here. The file names above have held, but the paths they sit under have not. Unzip the export and search it for the word topics, then for advertisers, rather than trusting any fixed path, including one printed here.

Two companies, one file name

ChatGPT and Claude both name their export conversations.json. Inside, ChatGPT writes a mapping object per conversation and Claude writes a chat_messages array. A tool built for one shape reads zero from the other and usually says nothing about it. Our Count checks for both and tells you which it found.

The .js files in an X archive are not code you need to run

X wraps its JSON in a JavaScript assignment so the archive's own viewer can load it. personalization.js is the single most revealing file in this index: it names the advertiser audiences you were added to, by company. You do not need to run anything to read it, and the free Count opens it directly.

Read your own files, free

Knowing the file name is step one. The Count is step two: drop the export on the page and it prints the real numbers out of exactly these files, in about ten seconds. It runs entirely in your browser, nothing is uploaded, and there is no account, no email field and no card.

Open the free CountHow long each export takes

The Count gives you the numbers out of one file. The Reading is the next question: what the pattern across your files says the algorithms decided you are, written out in plain language you can argue with and mark wrong.

The full Reading is 6.99 a week or 69.99 a year. No free trial. There is a Refund button in your dashboard for 7 days, one tap, no reason required. The Count stays free forever either way.

What this index does not cover

Honest gaps

Questions about export file names

Which file in my TikTok export holds my search history?

user_data_tiktok.json, in the SearchList array. Each record carries a Date and a SearchTerm, and repeats are not collapsed, so the same term typed thirty times is thirty rows.

Which file in an Instagram export holds my ad interests?

your_topics.json holds the interest topics, under the key topics_your_topics. advertisers_using_your_activity_or_information.json holds the advertisers who uploaded a customer list your account matched. Meta renames these folders often, so search the unzipped export for the word topics rather than trusting a fixed path.

Why do ChatGPT and Claude both call their export conversations.json?

They independently chose the same name. The shapes differ: ChatGPT writes a mapping object per conversation, Claude writes a chat_messages array. A tool that assumes one shape silently reads zero from the other, which is why our Count checks for both.

Which file names does the free Count actually look for?

The ones in this index. The matching lives in the detector table of our parser, which is plain JavaScript served at pupulcorp.com/count.js, so you can read what it opens before you drop anything on the page.

Can I use this index in my own article or tool?

Yes. The page, the CSV at /data/export-files.csv and the JSON at /data/export-files.json are licensed CC BY 4.0. Cite the whole index as: Pupul, The data export file index, pupulcorp.com/data/files, checked 2026-08-28. To cite one filename, use that file's permanent id, for example pupulcorp.com/data/files#file-x-personalization.

The rest of the index

The inference file indexwhere each app writes what it decided you areEvery export wait, one tableall 11 platforms, reported windowsMy export is missing the good partthe tier trap, platform by platformWhat TikTok knows about youanswered from the file, not the policyWhat ChatGPT knows about youand which file actually holds itThe free Count, all platformsone page per export, none ask for emailThe TikTok Countwatch history, searches, ad interestsThe Instagram Countad topics, off platform activity, searchesThe Spotify Countevery play, every skip, the inference fileThe X Countassigned interests, advertiser audiencesThe Google Takeout Countsearch, YouTube, activity by productThe ChatGPT Countconversations, messages, hours