TikTok Comment Scraper 2026: Buy vs Build the Data
A TikTok comment scraper you build fights signed requests, anti-bot walls, and proxy rotation. Here is when to buy a metered comments API instead, and when to build.
For comment data from TikTok videos you do not own, buy it. A do-it-yourself TikTok comment scraper spends most of its life losing to signed requests, a login wall, and proxy rotation, and it re-breaks every time the app ships an update. A metered per-call actor hands back the same comment text, author, timestamps, likes, and threaded replies with none of that upkeep on your plate. Build your own only when comments are a permanent, huge-volume core product and you will actually staff the maintenance. Everything below is the honest math behind that call.
Monid is a pay-per-call data API marketplace that lets you reach hundreds of external data endpoints from one key and one wallet, paying only when a run succeeds. Two verified TikTok comment endpoints sit behind that single wallet, so you can price the buy route in an afternoon before you commit an engineer to the build route.
TL;DR
- Buy for comment data from videos you do not own. The DIY route wins only at extreme, steady volume where marginal cost per call is the whole ballgame.
- A DIY scraper fights request signing (X-Bogus and device signing), an anti-bot login wall, and proxy rotation, and it breaks again on every TikTok app update.
- The ScrapTik actor on Apify returns top-level comments and threaded replies from just a video ID, billed a fraction of a cent per call.
- TikHub runs the same job on the same wallet, so you can switch or cross-check providers without signing a second contract.
- The official TikTok Research API is approval-gated and will not hand an arbitrary public comment section to a general pipeline.
- Managed cost magnitude and live prices are at monid.ai/tools.
What actually breaks a DIY TikTok comment scraper?
The demo is easy. You open a video, watch the comment request fire in your network tab, copy it into code, and get one page back. Then it dies, and here is the part the demo never shows you.
Request signing. TikTok's comment endpoints expect signed parameters (the X-Bogus family and device-level signing) computed by obfuscated client code. Reproducing that signing outside the app is the entire hard problem, and it is a moving target the vendor deliberately reshapes.
The anti-bot wall. Unsigned or suspicious traffic gets a login prompt, a captcha, or an empty body that looks like success but carries nothing. You end up writing detection code to tell "no comments" apart from "you got blocked," which are not the same failure.
Proxy rotation. One IP pulling thousands of comment pages gets rate-limited fast, so you buy a residential proxy pool, add rotation and retry, and now you are running proxy infrastructure to read a public comment section.
The update treadmill. Every point that works today is coupled to the current app build. A TikTok release can change the signing scheme or a field name, and your scraper returns garbage until someone notices and re-reverses it. That someone is on payroll every week, not just the week you wrote the first version.
A metered actor absorbs all four of those. The provider reverse-engineers the signing, runs the proxy pool, watches for app changes, and repairs the parser, and you send a video ID.
How do you buy comment data instead of building it?
Discovery and inspection are free on Monid, so an agent or a human can confirm the input shape and the current price before spending anything. Only a run bills.
For agents
Grab an API key at app.monid.ai, then paste this to your agent and hand it the key:
set up https://monid.ai/SKILL.md
It learns the whole discover, inspect, run workflow itself. More details in the agent quickstart.
For humans
npm install -g @monid-ai/cli
monid keys add --label main --key <your-api-key>
More details in the CLI quickstart.
The only input the comment call needs is the aweme_id, the run of digits at the end of any TikTok video URL. Inspect the endpoint for free, then run one page:
# managed: ScrapTik on Apify, flat per call (aweme_id is the number in the video URL)
monid run -p apify -e /scraptik/tiktok-comments-scraper-api \
-i '{"listComments_awemeId":"7127359192452599042","listComments_count":50}' -w
# threaded replies to one comment
monid run -p apify -e /scraptik/tiktok-comments-scraper-api \
-i '{"commentReplies_awemeId":"7127359192452599042","commentReplies_commentId":"6999860547420766982","commentReplies_count":50}' -w
# same job, a different provider on the same wallet: TikHub
monid run -p tikhub -e /api/v1/tiktok/web/fetch_post_comment \
--query '{"aweme_id":"7127359192452599042","count":50}' -w
The -w flag waits inline and prints the result. The first call returns top-level comments, each with text, author, like count, and timestamp. The second pulls threaded replies for one comment you chose, so you only pay for reply depth on the threads that earn it. The third does the identical job through TikHub. Two providers, one wallet, no second signup. That is the point managed buys you that a build never can: if one provider degrades after a TikTok update, you fail over by changing the -p flag, not by re-hiring for a reverse-engineering project.
Buy vs build, compared fairly
The DIY route is not a strawman. It genuinely wins on marginal cost at extreme steady volume, because your only ongoing spend is proxies and servers. Here is the whole picture, rival by rival.
| Axis | DIY scraper | ScrapTik on Apify via Monid | TikHub via Monid |
|---|---|---|---|
| What breaks it | Signing changes, anti-bot wall, IP bans, app updates | Provider handles it; you inspect fields | Provider handles it; you inspect fields |
| Setup | Reverse-engineer signing, build proxy pool, parser | One key, one endpoint, a video ID | One key, one endpoint, a video ID |
| Returns replies? | Only if you build a second signed path | Yes, commentReplies_* call | Depends on route; inspect first |
| Proxy + signing burden | All yours, forever | None | None |
| Maintenance | Staffed, weekly, unpredictable | None on your side | None on your side |
| Cost shape | Fixed infra plus salary, low marginal per call | Flat per call, a fraction of a cent | Flat per call, a fraction of a cent |
| Best fit | Comments as permanent huge-volume core product | Bursty research, comment mining, per-video sweeps | Same, plus a cross-check or failover provider |
What does a comment sweep actually cost?
Managed billing here is per call, at a fraction of a cent, and each call returns a full page of comments. So the bill tracks pages, not comments. Reason it out per step:
- One video, a few hundred comments, 50 per page. That is a handful of calls, which lands in single-digit cents total. Add reply pulls on ten hot threads and you are still in cents.
- A research batch of a hundred videos. Still a few calls each, so the whole batch is single-digit dollars for thousands of comments.
- A DIY comparison. Your per-call marginal cost can be lower, but you carry fixed proxy spend plus the salary of whoever keeps the signing alive. Divide that fixed cost across your monthly call volume. Below a very high, very steady volume, the managed per-call price is cheaper all-in, because you are not paying an engineer to fight
X-Boguson a slow week. Above it, DIY marginal cost eventually wins, which is the honest case for building.
The per-call price shows before every run and is listed at monid.ai/tools, with no subscription and no per-vendor minimum underneath it.
One honest caveat
Public scraping only sees public comments. Deleted comments, hidden comments, and comments on private accounts return nothing, and that is correct behavior, not a bug to code around. Coverage and field names also shift as TikTok changes and as each provider updates, so do not trust this post over your own run. Sample one video before you scale a batch: pull a page, eyeball the text and like counts against what the app shows, and only then wire in the rest of your list.
When should you actually build it yourself?
Build when all three are true: comment ingestion is a permanent core product (not a research project with an end date), the volume is huge and steady enough that managed per-call pricing dominates your bill, and you will fund the ongoing reverse-engineering headcount to keep signing alive through every TikTok update. Miss any one of those, especially the third, and a home-grown scraper becomes a slowly rotting liability that a metered actor would have replaced for cents.
For everyone else, and that is most comment work (competitor research, sentiment on a launch, FAQ mining, moderation snapshots), buying is the correct default. You get comment text, authors, timestamps, likes, and threaded replies from a video ID, and you spend your engineering time on the analysis instead of the plumbing.
FAQ
Is a TikTok comment scraper legal to run? This is about public comments only, the ones any logged-out visitor can read. Deleted, hidden, and private-account comments return nothing by design. Follow the platform terms and applicable law for your use, and keep to public data.
Can I get threaded replies, not just top-level comments?
Yes. Replies are a separate call keyed on the parent comment ID plus the video's aweme_id, so you pull reply depth only on the threads you choose and never pay for it on low-signal comments.
Why not just use the official TikTok API? The TikTok Research API is approval-gated and aimed at qualifying institutions, and there is no supported route that hands a general pipeline the public comment section of an arbitrary video. That access gap is exactly what the metered actors fill.
How much does the managed route cost? Per call, a fraction of a cent, and each call returns a full page, so a whole video's comments is usually single-digit cents and a research batch is single-digit dollars for thousands of comments. The live price shows before every run at monid.ai/tools.