Telegram Bot Analytics: The Metrics That Matter and How to Track Them (2026)

Why Telegram gives bot owners almost nothing by default

Telegram channels have a Statistics screen with views, shares and growth. Bots have nothing of the kind. BotFather shows you the token, the name and the description, and not a single usage number. There is no list of your subscribers anywhere on Telegram, no count of how many people pressed /start, and no way to ask "how many users do I have?" after the fact.

What Telegram does give a bot is a stream of events: a message arrived, a button was tapped, a payment went through, a delivery to this user failed. If nobody stores those events, they are gone. So every number you will ever see for your bot comes from the software running it, whether that is a bot builder or your own code. And some numbers do not exist at all: Telegram sends no read receipts for bot messages, and it does not announce when a user blocks the bot. It only tells you when you next try to message that user and fail. The question is not "where is my bot dashboard" but "which events is my platform recording, and which of them matter?"

The ten metrics that matter

Every range below is a rough guide, not a law. Bots differ enormously by purpose and traffic source, and a trend over four weeks tells you more than any single figure.

1. New users per day

Definition: people who pressed /start for the first time that day. This is the top of your funnel and it should map cleanly to what you did: a channel post, a QR code at the counter, an ad, a link in a bio. There is no healthy absolute number; the healthy pattern is a baseline you recognise plus spikes you can explain.

2. Active users (24 hours and 7 days)

Definition: users who sent a message or tapped a button in the window. Divide 7-day active users by total users and you get the number that matters most for an established bot. Rough range: 20 to 40 percent in the first month of a menu, support or ordering bot, settling to 5 to 15 percent after a few months. Community bots and bots that deliver something daily, such as an RSS digest, run well above that. Below 3 percent, most of your "users" tried the bot once and left.

3. /start to first action

Definition: the share of new users who do anything after the welcome message: tap a button, send a command, type a question. Rough range: above 60 percent is good; below 40 percent means the welcome message is the problem, not the bot. As our bot best practices guide argues, most bots are abandoned at the welcome message, not at a broken feature.

4. Command usage share

Definition: which commands and menu buttons get used, and what share of all uses each one takes. In almost every bot the top three account for more than 70 percent of usage. That is not a problem; it is a map. Put the top three first in the menu, and look hard at anything under 2 percent: it is either unnecessary, badly named, or buried.

5. Keyword reply hits

Definition: how often each keyword reply fires. Two signals hide here. A keyword that never fires means people do not phrase the question the way you expected. A rising number of plain-text messages that match nothing means your users are asking for something you have not built yet, and the messages themselves tell you what.

6. Form completion rate

Definition: submitted forms divided by started forms. Rough range: a form with three fields completes at 50 to 70 percent; at six or more fields the rate usually falls below 30 percent. Every field you remove buys you completions. Our lead generation guide goes into which fields are worth keeping.

7. Broadcast delivery and read proxies

Definition: recipients, delivered and failed for each broadcast, plus what happened in the 24 hours after it. Telegram sends no read receipts for bot messages, so "read rate" does not exist. What you can measure instead: taps on the buttons in the broadcast, replies, and the bump in active users the same day. Rough range: a tap rate of 5 to 15 percent on a broadcast with a clear button is healthy; under 2 percent means the message or the offer missed.

8. Blocked and unsubscribed users

Definition: users who blocked the bot, which you see as failed deliveries. Rough range: under 2 percent of recipients per broadcast is normal churn; 5 percent or more on a single send is a warning about frequency or content. Telegram has no spam folder, so this is the only feedback you get before people simply disappear. Read our broadcast guide for cadence and content that keeps this number low.

9. Response time when a human is involved

Definition: time from a form submission or a handoff request to the first human reply. Bots collect leads instantly, and then the lead sits. Measure the gap from timestamps if you must. Rough range: under one hour in business hours converts several times better than next-day; after 24 hours the lead has usually gone elsewhere.

10. Revenue per user, if you sell with Stars

Definition: paying users divided by total users, and Stars revenue divided by 7-day active users. Rough range: 1 to 3 percent of users paying is typical for a free bot with a paid command, a shop or a donation button. Track the two separately, because a bigger audience can raise one and lower the other.

A 15 minute weekly review

You do not need a dashboard open all day. You need the same numbers at the same time every week, compared with last week, and one decision at the end:

Pick a day and stick to it. Monday morning works: the week is complete and nothing has happened yet.
Write down five numbers: new users this week, 7-day active users, top three commands, failed deliveries on the last broadcast, and form submissions (or Stars sales). Five, not twenty.
Compare each with last week and with four weeks ago. Ignore changes smaller than your usual week-to-week wobble.
Explain every spike and dip in one sentence. "New users doubled because of the Thursday channel post." If you cannot, that is the item to investigate.
Change one thing. Reorder the menu, shorten a form, rewrite the welcome message. One change per week, so next week you know what caused the difference.

Common misreadings

Total users is a vanity number. It only ever goes up, it includes everyone who blocked you, and it says nothing about this week. Report it if you like; decide on active users.
Not every /start is a person. Spam scripts hit public bots in waves: dozens of new users within minutes, no username, no second action, a language that does not match your audience. Exclude those runs before you celebrate, and never count them in your /start to first action rate.
Time zones move your days. A daily chart draws its midnight somewhere, and if that is not your local midnight, late-evening activity lands on the "wrong" day. Compare the same weekday against the same weekday, and treat day-to-day differences of a few users as noise.
Broadcast days are not organic growth. A send lifts active users and command usage for a day. A week with two broadcasts beating a week with none does not mean the bot got better; you messaged people more.
Small numbers lie about percentages. Going from 3 form submissions to 6 is not "100 percent growth". Below roughly 30 events, look at the count and wait another week.
Failed deliveries are not all blocks. Most are, but a deleted account fails the same way. Watch the rate over time, not the individual cases.

Getting the numbers without code in TeleMinute

TeleMinute records the events Telegram sends your bot and turns them into the numbers above, with nothing to install. Here is where each one lives.

Statistics page, per bot. Total users, active users in the last 24 hours and 7 days, a 30-day chart of new users against active users, a command usage chart, a breakdown of user languages, and a table of your users with their first interaction that you can export to a spreadsheet. The three cards at the top and the 30-day chart are your weekly review in one screen.
Form responses. Every lead form submission is stored with its answers and time. Your completion rate is submissions divided by uses of the command that opens the form.
Broadcast stats. Each broadcast, scheduled or immediate, reports recipients, delivered and failed, and the Statistics page shows the last one. Failed deliveries are your block count.
Daily bot report in Telegram. In the bot editor you switch on a daily report, choose the hour in your own time zone, and tick the metrics you want: new users, active users, total users, top commands, form responses and Stars sales. It arrives as a direct message from the official TeleMinute bot every day at that hour, with a button that opens the full Statistics page.
Sales page for Stars. If you sell paid commands, take donations or run a shop with Telegram Stars, the sales page lists every payment, so revenue per user is a division away.

Two honest limits. There is no funnel builder and no A/B testing tool, so the /start to first action rate is a comparison you make yourself between new users and command usage. And, like every platform, TeleMinute cannot show read receipts, because Telegram does not send them.

What to change based on each metric

If this is off	Change this first
New users flat	Distribution, not the bot: pin the bot link in your channel, add it to bios, print the QR code, mention it in every broadcast.
Low /start to first action	Rewrite the welcome message: one sentence on what the bot does, at most four buttons, the most useful one first.
7-day active rate falling	Give people a reason to return: a scheduled broadcast with real news, an RSS feed, a weekly offer. Then check the block rate did not rise with it.
A command under 2 percent of usage	Rename it, move it up, or remove it. Fewer buttons raise the usage of the ones that remain.
Keyword replies not firing	Read the unmatched messages, add the phrasings people actually use, and add a keyword reply for the top unanswered question.
Form completion under 40 percent	Cut fields until only the ones you will act on remain, and ask for the phone number last.
Broadcast tap rate under 2 percent	Shorter message, one button, a reason to tap now. Send at the hour your active users chart peaks.
Blocks above 5 percent on a send	Halve the frequency and stop sending anything that is not news to the reader.
Slow human response	Turn on the daily report or check submissions at fixed hours, and set an automatic reply that says when a person will follow up.
Paying users under 1 percent	Put the paid item on the main menu, show the price in Stars up front, and give a free sample of what the paid command returns.

See your bot's numbers in one minute, no token needed

TeleMinute lends you a demo bot for 15 minutes, with no card and no BotFather step. Build a menu, message your bot, and watch the statistics fill in. Paste your own token whenever you are ready and keep everything.

Try a demo bot, no token needed

Frequently asked questions

Does Telegram have built-in analytics for bots?

No. Channels get a Statistics screen, bots do not. BotFather shows no usage numbers, and the Bot API only sends your bot the events it receives: messages, button taps, form input and delivery errors. Anything you know about your bot is whatever the software running it recorded, so analytics come from your bot builder or your own code.

Can I see who read my bot broadcast?

Not directly. Telegram does not send read receipts for bot messages. You can see how many recipients the message was delivered to and how many deliveries failed, and you can use proxies for reading: button taps, replies and command usage in the 24 hours after the broadcast.

What is a good active user rate for a Telegram bot?

There is no universal benchmark, but a rough guide: for a menu or support bot, 20 to 40 percent of total users active in the last 7 days is healthy in the first month, and 5 to 15 percent is normal once the bot is a few months old. Community and daily-content bots run higher. Watch the trend week over week rather than the absolute figure.

How do I know if users blocked my bot?

Telegram tells the bot only when it tries to message that user and the delivery fails. In practice you see blocks as failed deliveries in your broadcast report. A failure rate under 2 percent per broadcast is normal; above 5 percent means the frequency or content of your messages needs to change.