How Third-Party Telegram Analytics Sites Get Their Numbers

Third-party Telegram analytics sites get their numbers mainly by reading what Telegram already makes public about public channels: subscriber counts, posts, view counters and forwards. They collect this through Telegram's public web preview pages and through Telegram's API using ordinary accounts, then store regular snapshots and calculate figures like averages, engagement and growth. Anything Telegram does not publish, such as who the subscribers are, can only be estimated or inferred, and sites differ widely in how honestly they label those estimates.

This guide explains each data source, what can and cannot be measured from outside, how the common metrics are built, why two services rarely agree, and how to read a stats page critically, whether you are an advertiser, a buyer or a channel owner.

The starting point: what Telegram makes public

Every outside service is limited by what Telegram exposes. Understanding that limit explains most of what you see on any stats site.

Visible to anyone for a public channel

• The subscriber count. Telegram shows it on the channel's profile.
• Every post. Text, media, links and the date it was published.
• A view counter on each post. Telegram counts views and shows the total under every channel post.
• Reactions. The count of each reaction shown under a post, if the channel allows them.
• Comment counts. If the channel has a linked discussion group.
• Forward headers. When a post was forwarded from another channel, the "Forwarded from" label names the source.
• Mentions and links. Usernames and t.me links inside posts.

Through Telegram's API, a client can also read some details apps do not display prominently, such as how many times a channel post has been forwarded and whether it was edited.

Visible only to the channel's admins

Telegram gives admins of channels above a certain size a statistics page. It covers things like follower growth, how many people joined and left, where new followers came from, views by hour, shares, and the languages of the audience. These figures come from Telegram's own records, not from public counters.

Outside services cannot see these numbers unless the owner chooses to share them, for example by connecting their account to a service. Our guide to Telegram channel statistics explains what that admin page shows.

Not available to anyone outside Telegram

• The list of subscribers of a channel, for outsiders.
• Who viewed a post.
• What happens to a post after it is forwarded into private chats.
• Demographics like age or gender.
• Anything inside private channels, groups or chats the service is not a member of.

Any outside figure that claims to describe these things is an estimate or a guess.

Data source 1: the public web preview

Every public channel has a web preview page at t.me/s/ followed by its username. It shows recent posts in a browser without logging in, along with their view counts, and the channel's name, photo and description.

How services use it

A service requests the preview page for a channel, reads the posts and their view counters, and stores them with the time it read them. By reading the same page again later, it can see how view counts grew and record new posts. Older posts can be loaded page by page, so a service can read back through a channel's history.

Strengths

• No Telegram account is needed.
• It reads exactly what any visitor sees, so the figures are public facts.
• It is light on resources for a moderate number of channels.

Limits

• Public channels only. Private channels have no preview page.
• Some channels cannot be previewed. Channels restricted for sensitive content, and some others, may show only a message that the channel cannot be displayed. Our guide to why a channel cannot be displayed explains common reasons.
• Rounded numbers. Large counters may be shown in rounded form, such as a figure followed by K. That loses precision for big channels.
• Format changes. If Telegram changes the page layout, the service's reader breaks until someone fixes it.

TGDirectory reads channel figures from this public preview. Our guide on how TGDirectory measures channel statistics explains exactly which figures we take from it.

Data source 2: Telegram's API through ordinary accounts

Telegram's main API, called MTProto, is the same interface Telegram's own apps use. Developers can write programs that log in as an ordinary Telegram account and read public channels the way an app does. Our guide comparing the Bot API, MTProto and userbots explains the difference between these interfaces.

What it gives a service

• Exact subscriber counts, not rounded ones.
• Full post history with exact view counts, forward counts, reactions and edit dates.
• Detail about forwards and replies.
• The ability to read channels faster and in larger numbers than the web preview allows.

Why services use many accounts

Telegram limits how quickly one account can make requests. When an account asks too much too quickly, Telegram makes it wait. A service tracking a very large number of channels therefore spreads the work over many accounts and schedules its reading carefully.

This also means a service's coverage depends on how much capacity it has. Smaller or newer channels may be checked less often, or not at all.

Joining channels

Reading a public channel through the API does not always require joining it. For private channels, a service's account must join through an invite link, which means the service shows up as a subscriber. Some services ask owners for an invite link to add private channels.

Data source 3: bots added by the owner

The Bot API, used by ordinary bots, cannot read a channel unless the bot is an admin of it. Some services ask owners to add their bot to the channel. Once there, the bot receives updates about the channel, which can include member changes and new posts depending on the permissions and settings.

This gives more accurate data for that one channel, at the cost of the owner granting access. Give such bots only the permissions they need. A statistics bot generally has no reason to post, delete messages or add admins. Our guide to adding a bot to a channel explains the permission choices.

TGDirectory uses its bot in a narrow way: to confirm ownership when someone claims a listing, by checking with Telegram that the bot is an admin of the channel. Our guide to claiming your channel on TGDirectory explains the process.

Data source 4: owners connecting their own statistics

Some services let a channel owner connect their Telegram account so the service can read the admin statistics page. That gives access to figures outsiders cannot see, such as where followers came from and audience languages.

Before connecting anything, understand what you are giving away. A service that logs in as your account can, in principle, do anything your account can. Never share a Telegram login code or two-step verification password with a third party, whatever it promises. Our guide on never sharing your Telegram login code explains why. If you want to share your statistics with an advertiser, screenshots or an export are far safer than handing over account access.

Data source 5: discovering channels in the first place

A service can only measure channels it knows exist. Telegram has no public list of every channel, so services discover them by:

• following mentions and links in channels they already read,
• reading "Forwarded from" headers,
• accepting submissions from owners,
• searching Telegram for keywords,
• reading public directories and websites that link to channels.

This is why small or isolated channels are often missing from a service. A channel that nobody mentions and nobody submits may simply never be found.

What happens when a channel changes its username

Services that track channels by username can lose a channel that renames itself, or treat it as a brand-new channel with no history. Services that store Telegram's permanent numeric channel ID keep the history intact. If your figures vanish from a service after a username change, that is the likely cause.

How the common figures are built

Once a service has raw counters and timestamps, it calculates the figures it displays. These calculations are where services differ most.

Subscriber count and growth

The subscriber count is read directly. Growth comes from comparing snapshots taken at different times. A service that records one count per day can draw a daily growth line. One that checks weekly can only draw a weekly one.

The quality of a growth chart depends on when the service started watching. Some services fill earlier periods with estimates to make a longer chart. TGDirectory does not: our growth chart uses one real subscriber count per day recorded since we started watching the channel, and appears only once enough real points exist.

Average views per post

The service adds up the view counts of a set of posts and divides by the number of posts. Three choices change the result:

• Which posts. The last ten, the last thirty, or everything in a time window.
• How old each post is. Views keep growing for days after posting. A post read an hour after publishing has far fewer views than the same post read a week later.
• Whether outliers are removed. One viral post can lift an average a long way.

Some services also show a median, which is less affected by one unusual post. TGDirectory shows both average and median views per post for this reason.

Reach over a period

Some services show a figure meant to represent how many views a typical post gets within a set time after publishing, such as the first day. To calculate it, a service needs to read each post at that exact age, or estimate from readings at other times. Ask which one a service does, because an estimate built from a curve is not the same as a measurement.

Engagement rate

The common form divides average views per post by subscribers. Other services include reactions, comments or forwards, or divide reach over a day by subscribers. Two services can both call their figure "engagement" and mean different things.

Posting frequency

Counted from post dates over a period: posts per day or per week. This is a direct measurement and is usually consistent between services, apart from whether deleted posts were ever counted.

Mentions and citation scores

By recording which channels mention or forward which, a service builds a map of references. It then counts mentions or calculates a weighted score, sometimes called a Citation Index.

Advertising detection

Some services try to label posts as ads, for example by spotting phrases, links to other channels or repeated post patterns across many channels. This is an inference. A recommendation the author was not paid for may be flagged as an ad, and a well-disguised paid post may be missed.

How often services refresh their data

Reading every channel all the time is impossible, so every service decides how often to check each one. That schedule shapes the numbers more than most readers realise.

Common approaches

• Priority by size. Large channels are checked often, small ones rarely. A small channel's figures can be days or weeks old.
• Priority by activity. Channels that post often are checked more, because there is more to record.
• Priority by demand. Channels people look up on the service, or owners who pay, are refreshed sooner.
• Fixed schedule. Every tracked channel is checked once in a set period, such as daily. This is fairest for growth charts, because every channel gets the same number of points.

What this means for you

A stats page should show when it was last updated. If it does not, assume it may be stale. A channel that grew quickly last week can still show last month's subscriber count on a service that checks it rarely, and an advertiser looking at that page will see an older, smaller channel.

Groups and bots: what outsiders can measure

Most analytics sites focus on channels because channels publish the clearest public figures. Groups and bots expose much less.

Groups

A public group shows its member count, and anyone can read its messages, so an outside service that joins or reads a public group can count:

• members over time,
• messages per day,
• how many different accounts posted in a period,
• busy hours.

There are no view counters on group messages, so there is no equivalent of views per post. A group's member count also says little about how many members are active. Private groups cannot be measured at all without joining.

Bots

A bot's conversations are private between the bot and each user. Outsiders cannot see how many people use a bot or what they do with it. Some bot profiles in Telegram's apps show a user count that Telegram itself publishes, and a service may read that. Beyond it, any outside figure about a bot's audience is an estimate or comes from the bot's owner.

Estimates and inferences: where to be careful

Some figures on stats pages are not measured at all. They are calculated guesses. Good services label them; others present them with the same confidence as a counted figure.

Audience demographics

Outsiders cannot see who subscribes to a channel. A service that shows a breakdown by country or age is inferring it, perhaps from the channel's language, its topics or data from its own users. Treat such breakdowns as rough at best.

"Real" versus "fake" subscribers

Services may score a channel's audience quality by looking for patterns common in bought audiences:

• subscriber jumps that do not match any visible promotion,
• view counts that are almost identical on every post,
• views that are very low compared with subscribers,
• views that appear in sudden bursts rather than growing gradually.

These patterns are useful warnings, but they are not proof. A channel can have a sudden jump from a legitimate mention by a large channel, or low views because it posts late at night for a sleeping audience.

Estimated earnings or ad prices

Some services suggest what a channel could charge for an ad or what it earns. These are formulas built from views and assumptions about the market. They are not figures anyone measured.

Back-filled history

As mentioned above, a growth chart that stretches back before the service started tracking a channel has been reconstructed or estimated. Ask how. If the service cannot say, assume it is modelled.

A quick test for any figure

When you are unsure whether a number was measured, ask three questions. Could someone outside Telegram have seen this directly, for example on the channel's public posts? Does the service say when it read it? Could two people check it and get the same answer? If all three are yes, it is a measurement. If any is no, treat it as an estimate until the service explains otherwise, and give it less weight in any decision that involves money.

Why two services show different numbers

It is normal for two services to show different figures for the same channel. The common reasons, roughly in order of how often they matter:

1. Timing. View counters keep rising. A snapshot taken a few hours later shows higher views and a different average.
2. Which posts are counted. The last ten posts and the last thirty days can give very different averages.
3. Definitions. Engagement, reach and growth can be calculated in different ways under the same name.
4. Precision. A service reading the rounded web preview and one reading exact API figures will differ slightly.
5. Coverage gaps. If a service missed some days, its growth chart and averages are built on fewer points.
6. Deleted and edited posts. A post removed after one service read it but before another did changes both averages.
7. Estimates. Any modelled figure depends on the model.
8. Errors. Readers break when Telegram changes something, and bad data can sit unnoticed for a while.

None of these mean one service is lying. They mean every figure needs a definition and a time before it can be compared.

Worked example: same channel, different averages

This example uses made-up numbers to show how timing alone changes an average.

A channel publishes one post a day. Each post typically gains most of its views in the first day and keeps gaining slowly after that.

• Service A reads the channel every morning and averages the last seven posts. The newest one is only a few hours old, so it has few views. That drags the average down.
• Service B ignores posts younger than two days and averages the seven before that. Each has had time to collect most of its views.

Service B's average is noticeably higher, though neither service made a mistake. An advertiser comparing the two without knowing their methods might think the channel had lost readers between the two sites.

The lesson: when you compare channels, compare figures from the same service, measured the same way, ideally at the same time.

How to read a third-party stats page critically

Use this checklist whenever a number on a stats page will influence money or a decision.

1. Find the definition. What exactly does this figure measure, and over which posts or which period?
2. Check the date. When was the data last updated?
3. Separate measured from estimated. Is this counted from public data, or calculated from a model?
4. Look at raw posts. Open the channel itself and look at recent view counts. Do they match the picture the stats page paints?
5. Look at the trend, not a single value. A steady pattern over weeks says more than one number.
6. Compare like with like. Do not mix figures from different services.
7. Look for gaps. A chart with missing days or a sudden jump may reflect the service, not the channel.
8. Ask the owner. For an ad purchase or a channel sale, ask the owner for screenshots of Telegram's own admin statistics. Compare them with the outside figures.

Signs a stats service deserves your trust

The checklist above is about individual figures. These signs are about the service as a whole, and they help you decide which one to rely on in the first place.

• It publishes its methods. A page explaining how each figure is calculated, which posts are included and how often channels are read.
• It shows update times. Every figure, or at least every page, says when it was last measured.
• It labels estimates. Modelled figures are marked as such, ideally with a short explanation of the model.
• It shows gaps honestly. A chart with missing days shows the gap rather than drawing a smooth line through it.
• It lets owners correct and claim. There is a way to report a wrong figure, and owners can prove control of their channel without handing over their account.
• It never asks for your login code. A legitimate service asks for, at most, a bot with limited permissions or a screenshot.
• Its figures match the channel. When you open a few channels yourself, recent view counts roughly agree with what the service shows for the same time.

A service that meets most of these is useful even if its numbers differ from another's, because you know what its numbers mean. One that meets few of them may still be right, but you have no way to tell.

What channel owners can do

Check how your channel appears

Look your channel up on the services advertisers in your niche use. Note what they show. If a figure looks badly wrong, see whether the service has a way to report errors or claim your listing.

Keep your channel readable from outside

If you want outside services and directories to measure your channel accurately, keep it public and avoid settings that hide it from the web preview. Content restrictions that block forwarding do not stop view counts being read, but they do stop forwards, which some services use to measure reach and mentions.

Use Telegram's own statistics as your source of truth

Your admin statistics come from Telegram itself. Use them for your own decisions, and treat outside figures as how others see you.

Protect your account

Only grant access you understand. Adding a bot with limited admin rights is very different from logging a service in as your account. Never share login codes or passwords.

Grow the numbers honestly

Bought subscribers and views tend to show up in exactly the patterns these services look for, and they damage the trust that makes your figures worth anything.

Common mistakes

• Treating any outside figure as official. Only Telegram's own admin statistics come from Telegram. Everything else is a reading or a calculation.
• Comparing figures across services. Definitions and timing differ, so the numbers are not comparable.
• Taking estimates as measurements. Demographics, earnings and back-filled history are modelled.
• Ignoring post age. Young posts have fewer views. Averages that include them look lower.
• Reading an audience-quality score as proof. Patterns suggest; they do not prove.
• Handing over account access for a stats report. Use screenshots or limited bots instead.
• Assuming a missing channel does not exist. Services only measure channels they have discovered.

FAQ

Can analytics sites see who subscribes to my channel?

No. Outside services can see the subscriber count of a public channel but not the list of subscribers. Any breakdown of your audience by country, age or interest on such a site is an estimate.

Can they see my private channel?

Only if one of their accounts has joined it, for example through an invite link, or you connect it yourself. A private channel has no public web preview.

Why do views on a stats site differ from what I see in Telegram?

Most often because the site read the count at a different time, or uses a different set of posts for its average. View counters keep rising for days.

Do analytics sites count views from private forwards?

When a channel post is viewed after being forwarded, Telegram may count that view on the original post's counter, so it can be included in the public view count. But outside services cannot see where the forward went or who read it.

Are third-party statistics accurate?

Measured figures read from public counters are usually accurate for the moment they were read. Derived figures depend on the method. Estimated figures can be far off. The best services say which is which.

Where can I see my channel's figures on TGDirectory?

Once your channel is listed, its page shows subscribers, average and median views, engagement rate, posts per week and a growth chart built from real daily counts. If it is not listed yet, you can submit your channel for free.

Summary

Third-party Telegram analytics sites work from what Telegram makes public: subscriber counts, posts, view counters, forwards and mentions in public channels. They collect it through the public web preview, through Telegram's API with ordinary accounts, through bots owners add, and sometimes through owners connecting their own statistics. From those readings they calculate averages, engagement, growth and citation scores, and some add estimates for things no outsider can see, such as demographics or earnings. Numbers differ between services mainly because of timing, which posts are counted and how each figure is defined. Read every figure with its definition and date, separate measured from estimated, and check the channel itself before you trust a number with your money.