TikTok Profile Scraper API: Turn Any Account Into a Dataset

This is the highest-value use of the endpoint, because influencer budgets get committed on screenshots. A creator with 400,000 followers whose last twenty posts average 6,000 views is a very different proposition from one with 40,000 followers averaging 90,000.

Ask for the last 90 days, compute a median, and look at the spread. The median is the number that matters — one viral post drags an average upward and hides an otherwise flat account.

Engagement numbers tell you which posts worked. They do not tell you why. For that you need the words, and TikTok captions are rarely where the substance lives — it is in the audio.

Because each scraped video carries its URL, transcription is just the next loop. Add TikTok transcription on top of the scrape and you have the full spoken content of an entire account, in one pass, in 99+ languages.

That is the input for the questions people really want answered: what claims is this creator making about our product category, how do they open a video that performs, which talking points recur across their best posts.

And it works whether or not the creator added captions. Most do not. Ask for the transcript and you get the captions when they exist; when they do not, speech-to-text reads the audio instead, so the loop never returns an empty row. Same call shape either way — see the transcript and transcribe references.

Raw transcripts are still prose, and prose does not go into a spreadsheet. The step that makes this genuinely useful is asking for columns: define the fields you care about and get them back typed and populated for every video.

Hook, products mentioned, call to action, format. Now you can sort 200 posts by views and read down the hook column, which is a different kind of answer than reading 200 transcripts.

Competitive monitoring is where the date filter pays for itself twice — once in speed, once in cost. Point after_datetime at yesterday and each run reads only what is new, so a nightly sweep across a dozen accounts stays trivially cheap.

Store the results and you accumulate something you cannot buy: a longitudinal record of what your competitors posted and how it performed, day by day.

Scraping is the cheap part. A 600-video account is about 0.004 credits, so the metadata for a hundred accounts costs well under half a credit. Narrowing the date window lowers it further, because the scrape stops early rather than reading the whole history.

Transcription and extraction are billed separately, by video, when you chain them on — that is where real spend lives, and it is entirely under your control: filter first, analyse only the posts that matter. Full breakdown on the pricing page.

If a scrape fails, every charge for it is reversed automatically.