Almost always the cause is the tier. On most platforms the fast export and the revealing export are two different options on the same screen, and the fast one is the default. This page names the request you actually needed, platform by platform, and the four other causes worth checking first.
Before you re-request anything, it is worth confirming the file really is empty. The free Count reads it on your device and says plainly which lists it found and which were absent. No account, no email field, no upload.
On most platforms the fast export and the revealing export are two different requests on the same screen, and the fast one is the default.
This is not a conspiracy, and we are not going to pretend it is. Larger exports genuinely cost more to build, so the quick option is offered first and it is the one almost everybody takes. The result is the same either way: the file lands, the conclusions the platform drew about you are not in it, and most people conclude there was nothing to find.
| Platform | What you probably requested, and what to request instead reported |
|---|---|
| Spotify | You requested Account data. Inferences.json, the segments Spotify filed you under, is reported to sit in Technical log information, a separate request with its own clock of up to 30 days. Spotify's own Article 15 page names three packages you can request separately or at once. |
You took the fast file selection, which is reported to arrive in about ten minutes. Ad_Targeting.csv and Inferences_about_you.csv are reported to ship only in the larger data archive, which takes about a day. | |
| TikTok | You selected Activity, which is faster. The ad interest categories sit in the Ads and data section, which arrives only with All data. If you also chose TXT, request again as JSON. |
You narrowed the selection or the date range. Choose JSON, All time, and include the Ads information category, then look for your_topics.json. | |
| Amazon | Nothing wrong with your request, it is just slow. The Advertising category on its own is reported to come back in a day or two, where Request All Your Data is reported at one to two weeks. |
| X | The archive is one tier, so this is usually not a tier problem. personalization.js sits in the data folder, and the interests are nested under p13n.interests rather than at the top level, which is why a quick look can miss them. |
| Takeout is per product. If you did not tick My Activity and YouTube and YouTube Music, the history is genuinely not there. Google's interest and demographic guesses are a different matter, covered below. |
Every tier requirement here is labeled reported: it comes from the platforms' own help pages and consistent public reports, checked 2026-08-28, and it is tracked row by row in the inference file index. When we confirm one from an export of our own, that row will say verified and carry the date.
The wait gets the attention and the expiry is what actually costs people their export. Reported windows as of 2026-08-28: ChatGPT and Claude links about 24 hours from the delivery email, LinkedIn 72 hours, TikTok and Instagram about four days, Netflix about seven days, and Google Takeout about seven days with a limit of five downloads per archive. Miss it and the request starts from the beginning.
Request the export when you have ten minutes later that week to actually download it. Every window, per platform, is in the wait times table.
An export that looks empty in a text editor is often full. The free Count opens the file on your device, says which lists it found and which were absent, and prints the real numbers in about ten seconds. That is usually faster than waiting out a second request, and it tells you whether a second request is even needed.
The Count gives you the numbers. 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.
| Claim | Source, checked 2026-08-28 |
|---|---|
| The tier each platform hides its conclusions behind | Our own inference file index, tracked per platform and labeled reported. Free CSV under CC BY 4.0. |
| The file that should hold each thing, and the key inside it | Our own export file index, grounded in the parser at pupulcorp.com/count.js. |
| Wait windows and download expiry per platform | Our own wait times table, sourced to each platform's published help text. |
| Spotify's three separately requestable packages | Spotify's Article 15 information page, which names Account data, Extended streaming history and Technical log information. |
| Takeout expiry and the five download limit | Google's Takeout help page, which states an archive expires in about 7 days and can be downloaded 5 times. |
Almost always the tier. On most platforms the fast export and the revealing export are two different options on the same screen, and the fast one is the default. Spotify keeps its inference file in a separate Technical log information request, LinkedIn ships its inference files only in the larger archive, and TikTok's ad interests arrive only on All data.
Usually a narrow selection, a date range that is not All time, or the HTML format. Choose JSON, All time, and include the ads information category, then look for your_topics.json and advertisers_using_your_activity_or_information.json. Reported as of 2026-08-28.
Because it is reported to live in the Technical log information package, which is a separate request from the Account data package most people tick. Spotify's own Article 15 page names three packages you can request separately or at once.
Yes, you start again. Reported windows as of 2026-08-28: ChatGPT and Claude links about 24 hours, LinkedIn 72 hours, TikTok and Instagram about four days, Google Takeout about seven days and only five downloads. Request it when you have ten minutes to actually download it.
Sometimes, and no re-request fixes that. Google's interest and demographic guesses are shown on screen at My Ad Center rather than shipped in Takeout for most accounts, and no export contains a ranking model or the reason any single item reached you.
Same shape every time: the exact taps, the reported wait, the real size, what is inside, and what is missing.