How to Collect TikTok Data for Academic and Market Research

There are three realistic ways to collect TikTok data for a research project: apply for TikTok's official Research API (powerful, but gated behind an academic approval process), use a developer scraping platform (flexible, but requires code and a budget that scales with volume), or gather public post data directly in your browser with an extension like FeedRama and export it to CSV. For most theses, market studies, and content analyses, the third option gets you a working dataset the same afternoon.

Below we compare all three honestly, then walk through the no-code workflow in detail — including how to add video transcripts, which turn a metrics table into a corpus you can actually code and analyze.

What TikTok data can you actually collect?

Any public video exposes a consistent set of variables: view count, like count, comment count, share count, save count, caption text, hashtags, and upload date. Add AI transcription of the spoken audio and you also have the video's verbal content as text — which is what most qualitative and mixed-methods projects really need.

What you can't collect from the outside: private-account content, viewer demographics, watch time, or anything behind a login wall you don't own. A well-scoped study designs around public data from the start.

Option 1: TikTok's Research API

TikTok operates a Research API intended for academic researchers. If you're affiliated with a qualifying institution and your project fits its terms, it's worth applying — official access means documented endpoints and defensible methodology sections. The honest downsides: the application process takes time, approval isn't guaranteed, independent researchers and commercial market-research teams generally don't qualify, and you'll still need programming skills to query it.

Option 2: Scraping platforms and custom code

Developer platforms like Apify offer pre-built scraping actors for TikTok on a pay-per-usage model. They're genuinely powerful — good for very large datasets and scheduled recurring collection. The trade-offs are technical comfort (configuring actors, handling JSON output) and cost that grows with the amount of data you pull. If your study needs hundreds of thousands of posts, this is the right neighborhood; we compare the options in our Apify alternatives guide.

Option 3: Collect it in the browser — no code

For studies built on dozens to a few thousand posts — a niche analysis, a set of creator accounts, a hashtag sample — a browser extension removes every technical barrier. Here's the FeedRama workflow:

1. Install the extension. Add FeedRama from the Chrome Web Store. It runs on tiktok.com in desktop Chrome — no API keys or account setup.
2. Define your sample frame. Open the profiles, hashtags, or search results that match your inclusion criteria. Document these choices — future-you writing the methods section will be grateful.
3. Sort the feed. Use FeedRama to reorder the videos by views, likes, shares, or date. Sorting by date is ideal for time-bounded samples; sorting by engagement suits "top performing content" designs.
4. Select and export. Select the videos in your sample and export to CSV. Each row carries the URL, date, all public metrics, and the caption.
5. Add transcripts if your analysis needs them. Run AI transcription on the selected videos before exporting — the transcript text lands in its own CSV column.

Free-plan limits worth knowing before you plan a study: sorting is unlimited but capped to the previous 25 posts or one week per feed, transcription is 5 videos per month, and CSV export requires Pro ($10/month, or $5/month billed annually). Pro removes the range cap and transcription limit, which matters for any serious sample size.

A note on sampling strategy, because it shapes which sorting mode you use. If your research question is about typical content in a niche, sort by date and take every video in a defined window — sorting by engagement first would bias the sample toward hits. If the question is about successful content specifically (say, what features viral health-advice videos share), sorting by views is exactly right, but say so explicitly in your write-up. Mixed designs work too: pull a date-bounded sample for the baseline and a top-performers sample for contrast, and export each as its own CSV so the two never get mingled in analysis.

Turning videos into analyzable text

Metrics answer "what performed"; transcripts answer "what was said." Because FeedRama transcribes the audio track (not on-screen captions), you get the actual spoken script — usable for thematic coding, sentiment analysis, or feeding into your qualitative analysis software. Two honest limitations: accuracy is strong for clear speech but drops for music-heavy or silent videos, and on-screen text overlays aren't captured. For large samples, see how to bulk transcribe TikTok videos.

Which collection method fits your project?

Ethics, terms, and citing your method

Public-content research is well established, but treat it with the same care as any human-subjects-adjacent work. Collect only from public accounts, follow your IRB or ethics board's guidance on anonymizing creators, and don't republish videos in your outputs without permission — quote transcripts with attribution instead. In your methods section, record the collection dates, the sorting criteria you used, and the tool version; engagement numbers drift over time, so a dated snapshot is part of your study's reproducibility. For the spreadsheet mechanics after export, our guide to exporting TikTok data to a spreadsheet picks up where this one ends.

FAQ