Frequently Asked Questions

Researchers, journalists, creators, and the people whose feeds it is built from. The Hub runs on data that everyday platform users have donated. Working with that data by hand is slow: every platform exports a different format, useful metadata is missing, and someone has to sit and code thousands of videos. The Hub does that part for you, so you can get to the analysis.

The For You Research Project is the research. It's an academic project based at QUT's Digital Media Research Centre and the University of Sydney, funded by the Australian Research Council, and it studies how short-video recommendation shapes what people watch and who gets seen.

The For You Data Hub is the software we built to do that work: a browser-based workbench that turns donated feeds into datasets. It's open source, and other teams are welcome to use it. More about the project · More about the Hub.

Short-video activity donated by study participants: TikTok feed activity, Instagram activity and YouTube watch history, taken from the data export each platform lets a user download of themselves. We then add public metadata about the videos that appeared in those feeds (captions, creators, popularity) and AI-generated content annotations (categories, themes, style and more).

None of it is scraped from private accounts or pulled from a platform API. It is the feed a real person was actually served, shared by that person. What is data donation?

TikTok is the project's focus, and the Hub also ingests donated Instagram activity and YouTube watch history, so you can compare recommendation dynamics across platforms. Adding another platform is deliberately small work in the code, so the list can grow.

An account is free, and you need one for two things: sharing your own data, and looking around the analysis tools. Register on the sign-up page with an email address and a password. Depending on how this instance is set up, an administrator may review new accounts first, in which case you'll see "pending approval" for a short while.

A new account opens on a default study built from donated data, with the analysis tabs ready to use. Ingesting your own participants' data and running the full pipeline is a separate conversation, so get in touch when you're at that point.

Yes, please do. We're actively looking for participants who use TikTok, Instagram Reels or YouTube Shorts. You ask the platform for a copy of your data, review it in your browser when it arrives, and share what you're comfortable with. How to participate.

A view of your own feed that no platform gives you. Once your data is in, the Hub builds your short-video persona: what you binge, when you scroll, your rewatch habits and viewing rhythms, and how that compares with everyone else who has shared a feed. You also get to see what personal data the platforms have been storing about you.

No. Your export is opened and read inside your own browser, so it never goes anywhere to be inspected. Direct messages are never uploaded at all. They are dropped before you even see the review screen, along with your profile and settings details. What's left is the activity the research needs, such as which videos were served to you and when.

You then go through it section by section and delete anything you'd rather not share. The file that leaves your device is rebuilt from what you kept, so the rest simply isn't in it. Nothing is shared until you consent, and you can read the full statement you agree to on the consent and research ethics page.

Your data is anonymised, and we don't link it to your real-world identity. It's stored securely under QUT's data management policies, only approved accounts can reach the Hub, and each account only sees what its role allows. The analysis views are built around aggregate patterns rather than individuals, so findings are about how feeds behave, never about you.

All of this runs under approval from the QUT Human Research Ethics Committee (approval number LR 2024-8002-21442). The ethics page has the details.

Requesting your data is a couple of taps. The platform then takes somewhere between a few minutes and a couple of days to prepare the file. That's the slow part, and nobody can speed it up. Reviewing and sharing takes about two minutes. After that, the Hub fetches and analyses the videos you watched in batches over the following days, and your My Collections page shows how far along it is.

No. There's no app and no extension to install. You share a one-off copy of the data the platform already holds about you, through this website. We don't watch your account afterwards, and nothing changes about how you use the platform.

Yes, at any time and without giving a reason. You can withdraw a collection you've shared from My stuff → My Collections in the Hub, or by emailing us at info@foryouresearch.net. Withdrawal removes the collection from the corpus and from any study built on it.

Both. One of our three research questions is specifically about creators: how do creators adapt their strategies to stay visible and relevant? Alongside the feed analysis there's a research stream that works directly with creators of all sizes, through interviews, about how you navigate visibility. More about the research.

Not as a formula, and be sceptical of anyone who offers you one. What the project can show is what recommendation looks like from the audience side, at scale: which kinds of content get served to whom, how feeds change over time, and which patterns hold up across many people's feeds rather than in one viral story.

If a participant was recommended one of your videos, that video becomes part of their feed data. What the Hub stores about it is public information: the caption, the creator handle and engagement counts, plus AI-generated annotations describing the content. Analysis and publications report aggregate patterns, not profiles of individual creators. If you have questions about how your content is handled, email info@foryouresearch.net.

Two ways, and you can do either or both. You can share your own platform data like any other participant, which for a creator covers your posting activity as well as what you watched: see how to participate. Or you can talk to us, because the creator research stream is interested in creators at every scale. Email info@foryouresearch.net, or find us on TikTok and LinkedIn.

Yes, and we'd rather talk to you first than have the numbers travel on their own. Email info@foryouresearch.net with what you're working on and your deadline. The About page lists the investigators and what each of them works on, so you can see who to ask about platform policy, creator labour, content moderation or the data analysis itself.

Donated feeds are good evidence for lived experience: what real people were actually served, how a feed behaves from day to day, and how one participant's feed differs from another's. Neither platform transparency libraries nor APIs can show you that.

What they can't support are population-level claims. Participants volunteered, so they aren't a representative sample, and "X% of TikTok users see Y" is not a sentence this data can support. The longer version is here.

The participant data isn't a public download. People shared it with a named research team under a specific consent statement and ethics approval, and that consent doesn't cover passing it on. What we can do is walk you through the method, share aggregate figures, and explain how a particular result was produced. Ask us at info@foryouresearch.net.

The software itself is a different matter. It's open source, and you're welcome to create an account and look around the analysis tools yourself.

Check with us first if the image comes from a study of participant data, and never publish anything that could identify a participant. When you do use material from the Hub, cite the software and acknowledge the project. The citation formats are in the researchers' section below, and they work in a methods box just as well as in a reference list.

Two routes: use our hosted version of the Data Hub, or install the Hub on your own machine or cloud project. For a real study you probably want your own instance, so your participants' data sits on infrastructure you control and under your own ethics approval. We're really interested in working with other research teams, so get in touch and we can talk about how we might collaborate: info@foryouresearch.net.

Yes to both. The Hub is released under the MIT licence and developed in the open at github.com/pwikstrom/foryou-research. Everything you see here (ingestion, enrichment, annotation and the dashboard) is in that repository, and you are free to run it, fork it and adapt it for your own project.

Start with the installation guide. A setup wizard writes your configuration, and the Hub runs locally on a laptop for small studies and on Google Cloud Run for larger ones. Enrichment and AI annotation call external services, so your instance brings its own API keys. The guide lists which ones and what they cost.

Not as a dataset download. Participants consented to a named research team under QUT ethics approval LR 2024-8002-21442, and that consent doesn't cover passing the data on. Collaboration is a different question, and one we're open to. Arrangements between institutions, with the appropriate ethics in place, are how this normally works. Write to info@foryouresearch.net with what you have in mind.

Yes, and it's one of the better reasons to run your own instance. Annotation prompts are declarative contracts you edit in the browser rather than in code, so a project can define its own codebook and variables, version them, and evaluate the results against human coders. Role-based access lets you bring in collaborators and students at the level of access each of them should have.

A large multimodal AI model generates the video annotations against a structured, versioned codebook, and we continuously evaluate annotation quality, including comparisons with human coders. Some outputs will still be imperfect or unexpected, so treat individual annotations with appropriate care and look at aggregate patterns where you can.

Two separate things, in this order. Cite the software in your reference list, the same way you would cite any other research software you used. Copy whichever form your publisher wants:

@software{wikstrom2026foryoudatahub,
  author  = {Wikstrom, Patrik},
  title   = {The For You Data Hub: a research data toolbox for studying user experiences of algorithmically curated short-video platforms},
  year    = {2026},
  version = {0.2.0},
  doi     = {10.5281/zenodo.21994399},
  url     = {https://doi.org/10.5281/zenodo.21994399}
}

Then acknowledge the project in your funding or acknowledgements section. The Hub and the data in it exist because of that funding:

The For You Research Project is led from QUT's Digital Media Research Centre in Brisbane with the University of Sydney, and funded by the Australian Research Council (grant DP240102939). It is also supported by the Australian Internet Observatory. The About page introduces the investigators and the institutions.

No. TikTok is a registered trademark of Bytedance Ltd. This website and the research project are not affiliated with or endorsed by TikTok or Bytedance Ltd. The same goes for Instagram and YouTube.

The Hub is under active development, so bugs happen. Bug reports and feature requests belong in the public issue tracker: browse the open issues first in case yours is already known, then open a new issue saying what you were doing, what you expected and what happened instead. Screenshots help a lot.

Two things don't belong in an issue, because issues are public: participant data (including screenshots that identify a participant) and suspected security problems. Report the latter privately, following the security policy. For anything else you'd rather not post publicly, email info@foryouresearch.net.

Please do. The contributing guide covers the branch workflow, the test suite and the conventions the codebase relies on. Adding support for another platform is deliberately small (one ingestion class and one scraper class), and bug fixes, documentation and test coverage are just as welcome. If you're planning something substantial, open an issue first so we can talk it through before you write the code.