Telegram Channel Scraper — verified results or an explicit error

Scrape public Telegram channels without an API key, a bot token or a phone number: posts, views, reactions, media URLs, forwards, hashtags and mentions, with date and keyword filtering.

The difference from every other Telegram scraper on this Store is one rule:

An empty result is never used to report a problem, and a run that returns nothing costs you nothing.

Why that matters

Telegram answers HTTP 200 for channel handles that do not exist. The response is a normal-looking page with no channel content in it. A scraper that only counts posts on that page sees zero, calls the run a success, and bills you for it. One typo in a channel name and you get a clean, empty, paid-for dataset.

The same silent failure happens when a channel exists but its web preview is not served, when Telegram restricts the channel, when the handle is a group or a user profile instead of a channel, and when a page fails halfway through pagination.

This Actor classifies the channel before it trusts any post count:

State	What it means	What you get
PUBLIC_PREVIEWABLE	Real channel, readable	Posts
EXISTS_NO_PREVIEW	Real channel, Telegram serves no web preview	Explicit error
PRIVATE	Invite-only	Explicit error
RESTRICTED	Telegram serves no public page for the handle	Explicit error
NOT_A_CHANNEL	Handle is a user, bot or group	Explicit error
NOT_FOUND	Nothing at this handle (usually a typo)	Explicit error
UNREACHABLE	Telegram could not be reached	Explicit error

Every one of those states except PRIVATE is checked against live Telegram by scripts/smoke.py, which fails the build when a handle stops classifying the way it did when the signature was captured.

On top of that, every channel gets a verification report:

• verdict — OK, EMPTY_VERIFIED, PARTIAL or FAILED
• firstPostId / lastPostId / idGaps / coverageRatio — how complete the history actually is (post ids are not contiguous; deleted and service messages leave holes, and a scraper that paginates by arithmetic walks past real posts)
• reachedEndOfHistory, stoppedBy, pageFailures

EMPTY_VERIFIED is only ever emitted when the channel header rendered, the history container rendered, and the channel's own metadata does not contradict the empty result.

What this Actor deliberately does NOT do

No members, no user profiles, no phone numbers, no personal data of any kind. Channel member lists are personal data under GDPR and their extraction conflicts with Telegram's Terms of Service. This Actor reads public broadcast content only. If you need member lists, this is the wrong tool — on purpose.

Input

{
  "channels": ["durov", "https://t.me/telegram"],
  "maxPostsPerChannel": 100,
  "minDate": "2026-01-01",
  "searchTerms": ["release", "update"],
  "includeMediaUrls": true,
  "maxConcurrency": 5,
  "failOnUnreadableChannel": true
}

Channels are fetched in parallel (maxConcurrency, 1-10). Telegram round-trips dominate the wall clock, and compute time is billed, so waiting serially would be your money spent on latency.

failOnUnreadableChannel (default on) ends the run as FAILED with a reason when any requested channel could not be fully read. Turn it off to accept partial results — the reports still tell you exactly what was missed.

Output

One dataset item per post:

{
  "channel": "durov",
  "postId": 385,
  "url": "https://t.me/durov/385",
  "datetime": "2026-08-01T10:00:00+00:00",
  "text": "…",
  "views": 1200000,
  "author": null,
  "isForwarded": false,
  "forwardedFrom": null,
  "isReply": false,
  "isEdited": false,
  "hashtags": ["telegram"],
  "mentions": [],
  "links": ["https://example.com"],
  "media": [{ "type": "photo", "url": "https://cdn.telegram.org/…" }],
  "linkPreview": { "title": "…", "url": "…" },
  "reactions": [{ "emoji": "👍", "count": 12 }],
  "reactionsTotal": 15
}

The dataset holds only posts — one clean row each, and the only thing you are charged for. The verification reports (per-channel state, verdict, id coverage, page failures) are written to the key-value store as RUN_SUMMARY on every run, including runs that fail.

Pricing

Free while this Actor is in early access. You only pay Apify platform usage (compute), and that bill is deliberately small: this Actor talks plain HTTP to Telegram's server-rendered pages instead of driving a headless browser, so it runs in a fraction of the memory a browser-based scraper needs.

Pay-per-result pricing will be introduced later. When it is, the rule stays the same as the promise above: no start fee, and you are charged only for posts that passed verification. A run that returns nothing will never cost you anything.

Use cases

News and OSINT monitoring, brand and competitor tracking, crypto signal channel archiving, market and disinformation research, dataset building.

Need a scraper for something else?

I build custom Apify Actors and browser automation the same way this one is built: a verified result or an explicit failure, never an empty dataset sold as a success.

• My other Actors: https://apify.com/marekhartmann
• Code and test suites: https://github.com/marekhartmann-creator

Tell me the site and what you need out of it.