Telegram Channel Scraper & Monitor: posts, views, reactions & channel stats

Turn public Telegram channels into clean, structured data: post text, views, reactions, media links, forwards, replies and channel stats. Run it once to export a channel's history, or put it on a schedule and get only the new posts each time, across as many channels as you like, with an optional keyword filter.

No Telegram account, API keys or login needed.

Example output

One record per post. This is a real post from the Telegram Info English channel (text and media URL shortened):

{
  "channel": "tginfoen",
  "postId": 2183,
  "url": "https://t.me/tginfoen/2183",
  "postedAt": "2026-06-14T09:46:43Z",
  "isEdited": false,
  "author": "Sominemo",
  "replyToPostId": null,
  "text": "Rich Text Formatting is Coming To Users Too\n\nOn his Twitter account, Pavel Durov announced that extended Markdown formatting … will be available not only for bots, but for regular users as well. …\n\n#bots",
  "contentIncomplete": false,
  "links": ["https://x.com/durov/status/2065899497289392440", "https://t.me/RichTextDemoBot"],
  "hashtags": ["#bots"],
  "views": 9440,
  "viewsDisplay": "9.44K",
  "reactions": [
    { "emoji": "👍", "count": 42 },
    { "emoji": "❤", "count": 19 },
    { "emoji": "🤡", "count": 5 },
    { "emoji": "🌚", "count": 2 }
  ],
  "forwardedFrom": null,
  "media": [
    { "type": "photo", "url": "https://cdn4.telesco.pe/file/…jpg", "thumbnailUrl": null, "durationSec": null }
  ],
  "channelTitle": "Telegram Info English",
  "channelSubscribers": 30200,
  "scrapedAt": "2026-10-16T09:00:02Z"
}

Every record has the same fields, with null or [] where the page has nothing, so CSV and Excel exports keep the same columns. A separate, free CHANNELS record gives each channel's title, description, subscriber count and a status for every channel you asked for.

Why use this one

• Monitoring that never skips a post. On a schedule, each run returns only posts newer than the last one. If a run can't catch up, the next one re-sends rather than leaving a gap (see Monitoring).
• Keyword filtering where you only pay for matches. Posts that don't match your keywords are never pushed or charged.
• Separate schedules don't collide. Give each schedule its own stateKey and they keep independent bookmarks, even on the same channel.
• One stable, CSV-ready record shape. Fields never appear or disappear between posts or runs.
• Complete albums. A post with several photos or videos is one record listing every item in media, in order.
• The details others leave out: edited flag, author signature on signed posts, same-channel reply links, forward source, hashtags, links, video durations and polls.
• Free channel snapshot with every run, and no start fee: you pay per post delivered, nothing else.
• Light and polite. It reads Telegram's public web preview pages directly (no browser), with retries and backoff when Telegram asks it to slow down.

Quick start

Enter one or more channels and run. All of these work: durov, @durov, t.me/durov, https://t.me/s/durov.

1. One-off export of the latest posts from a few channels:

{
  "channels": ["telegram", "durov"],
  "maxPostsPerChannel": 200
}

2. Daily monitoring (new posts only). Save this as a task and schedule it:

{
  "channels": ["telegram", "durov"],
  "monitoringMode": true,
  "stateKey": "daily"
}

The first run sets a starting point and returns the latest posts. Every later run returns only posts newer than the previous run.

3. Backfill, then monitor. Run once with a date and a generous limit to get the history:

{
  "channels": ["telegram"],
  "postedAfter": "2026-09-01",
  "maxPostsPerChannel": 5000
}

Then switch to the monitoring input above. To avoid an overlap between the backfill and the first monitoring run, dedupe on (channel, postId).

Keyword filter (combine with any of the above):

{
  "channels": ["telegram"],
  "keywords": ["update", "privacy"],
  "maxPostsPerChannel": 500
}

Input

Field	Default	What it does
channels	none (required)	Public channels to scrape, as usernames or t.me links. Duplicates are merged.
maxPostsPerChannel	100 (1,000 in monitoring mode)	How many posts to scan per channel, newest first, up to 10,000. This counts posts looked at, not posts returned: with keywords, a quiet topic may return only a handful. In monitoring mode it is the per-run budget.
postedAfter	none	Only scan posts published on or after this date. In monitoring mode it is also a floor: older posts are treated as abandoned, so a long-paused schedule doesn't pull in a stale backlog.
keywords	none	Keep only posts whose text contains any of these words (case-insensitive). Unmatched posts are not charged, but they do count towards maxPostsPerChannel.
monitoringMode	off	Return only posts newer than the last run.
stateKey	default	Names this schedule's monitoring state, so separate schedules stay separate. Letters, digits, - and _.
proxyConfiguration	Apify datacenter proxy	Usually fine as it is. Residential proxies shouldn't be needed.

Output

Posts (the dataset)

Field	Type	Notes
channel	string	Channel username, lowercase.
postId	integer	Post number within the channel. (channel, postId) identifies a post.
url	string	Link to the post.
postedAt	string	Publication time, ISO 8601, UTC.
isEdited	boolean	Telegram marks the post as edited.
author	string or null	Signature on signed channel posts. A display name, not a verified identity.
replyToPostId	integer or null	The parent post, when the post replies to another post in the same channel.
text	string or null	Plain text with line breaks. For a poll without its own text, the poll question.
contentIncomplete	boolean	true when the web preview doesn't show the whole post (see Limitations).
links	string[]	Web links in the text and its link preview.
hashtags	string[]	Hashtags, including the #.
views	integer or null	Approximate view count (Telegram rounds it).
viewsDisplay	string or null	The view count exactly as shown, for example "1.2M".
reactions	array	{ "emoji", "count" }. emoji is the emoji, ⭐ for paid Stars reactions, or custom:<emoji-id> for a channel's custom emoji. Counts are rounded like views.
forwardedFrom	object or null	{ "name", "url" } of the original source. Both are null when the original sender is hidden.
media	array	{ "type", "url", "thumbnailUrl", "durationSec" } for every item, albums included. type is one of photo, video, gif, document, audio, voice, sticker, poll.
channelTitle	string or null	The channel's name at scrape time.
channelSubscribers	integer or null	Approximate subscriber count.
scrapedAt	string	When this run scraped the post.

In the Console, the Output tab offers three views of the same data: Overview, Engagement and Content. Exports (JSON, CSV, Excel and so on) always contain every field.

Channels (the CHANNELS record)

Find it in the run's Output tab under Channels, or in the default key-value store. One entry per input channel, in input order:

{ "channel": "telegram", "status": "ok", "title": "Telegram News",
  "description": "The official Telegram on Telegram. Much recursion. Very Telegram. Wow.",
  "subscribers": 9310000, "postsReturned": 30, "newestPostId": 460,
  "truncated": false, "scrapedAt": "2026-10-16T13:06:50Z" }

status	Meaning
ok	Scraped normally.
not_found	There's no public channel with that name.
preview_disabled	The channel exists but has turned off its web preview, so its posts can't be read without Telegram.
error	Something else went wrong, such as an invite link, a group or user account instead of a channel, or Telegram not responding. The run log explains.

A bad channel never fails the run and is never charged. truncated is explained below.

Monitoring explained

With monitoringMode on, the Actor remembers the newest post it has seen in each channel (its bookmark), separately for each stateKey. Each run scans newest-first and stops when it reaches the bookmark, so you get only what's new. The bookmarks live in a key-value store in your account named telegram-scraper-state.

When a channel posts faster than your budget. If a run uses up maxPostsPerChannel before reaching the bookmark, there's a gap between what it returned and the last run. The Actor does not move the bookmark in that case. It marks the channel truncated: true in CHANNELS and logs a warning. The next run starts from the newest post again and reaches further back, so no post is ever skipped. The price of that guarantee is duplicates: posts delivered in the truncated run are delivered, and charged, again. So:

• Dedupe on (channel, postId) downstream, always.
• If a channel keeps showing truncated: true, raise maxPostsPerChannel or run more often.
• postedAfter is the escape hatch. Once the gap is older than that date, the run stops there and the bookmark moves on.

The same applies when a run stops because it reached your maximum charge: the bookmark stays put, and the next run picks up the rest.

Outside monitoring mode, truncated: true simply means there were more posts than maxPostsPerChannel allowed (and postedAfter, if set, wasn't reached).

Keywords don't hold the bookmark back: it moves past every new post, matched or not.

Pricing

$0.002 per post delivered, which is $2 per 1,000 posts. There's no start fee, the CHANNELS snapshot is free, and posts filtered out by keywords cost nothing.

Example	Posts delivered	Cost
One-off: 10 channels × 100 posts	1,000	$2.00
Keyword search: scan 500 posts, 12 match	12	$0.024
Daily monitoring: 5 channels × ~10 new posts a day	~50 a day	~$0.10 a day, ~$3 a month
A second monitoring run with nothing new	0	$0.00

Set a maximum charge on the run and it stops cleanly when it gets there, charging only for posts actually delivered. Re-sent posts after a truncated monitoring run are charged again (see above).

Limitations

• Public channels only. It reads Telegram's public web preview (t.me/s/<channel>). Private channels, groups, user accounts, and channels with the preview turned off can't be scraped; they're reported in CHANNELS instead of failing the run.
• Counts are approximate. Telegram's web preview rounds view, reaction and subscriber counts (for example "1.2M"). views is the numeric reading and viewsDisplay is exactly what the page shows.
• Some posts aren't fully visible. For some posts (typically long-form article posts) the web preview shows only "Please open Telegram to view this post". Those records have contentIncomplete: true and text holds whatever was visible, possibly nothing.
• No comments or reply counts. The web preview doesn't show discussion threads.
• Media links, not files. You get URLs to Telegram's CDN, not downloads. Download anything you need soon after the run, as these links may not work forever. Videos too large for the preview have url: null but keep their thumbnail and duration.
• Polls: you get the question (as text when the post has no other text), not the options or vote counts.
• Custom emoji in reactions appear as custom:<emoji-id>, because the web preview doesn't say which emoji they are.

Responsible use

This Actor reads only the public web previews of public Telegram channels, the same pages anyone can open in a browser. It doesn't log in, use the Telegram API, bypass access controls or touch private content. Public posts can still contain personal data, and you're responsible for how you use what you collect, including Telegram's terms of service and the data-protection laws that apply to you (such as GDPR). This Actor isn't affiliated with or endorsed by Telegram.

Feedback

Found a channel that doesn't parse properly, or want a field added? Open an issue on the Actor's page and include the channel name so it can be reproduced.