پایگاه دانش GetMyBot

Broadcasts

Send a news item or reminder to a group of customers: choose content, channels and an audience, check recipients, then review the result.

در این صفحه

A broadcast sends one prepared message to many recipients. For example, invite everyone tagged Interested in the course to a class. Choose the content, audience, and channels, check who will receive it, then start sending.

GetMyBot offers two ways to do this:

  • The Broadcasts screen (/app/broadcasts): a full editor: one content document, a channel selection, a segment audience, a schedule, a pre-send check and a per-channel report. This is the main way.
  • The Broadcast button on a reaction: send a ready reaction to active subscribers. It has no draft or pre-send check, so review the reaction and audience before launching it.

Both need a plan that includes broadcasts and the bot-level Reactions right. Without the plan feature, creating a broadcast answers with a payment error; without the right, the screen shows an empty state instead.

The broadcast editor

Open Broadcasts in the left menu. Create broadcast opens an editor with four blocks.

Name and channels. Enter a name of up to 160 characters and choose one or more channels: Telegram, VK, WhatsApp, web chat, Email, browser push, mobile push, or in-app content. You can combine messengers with web chat, or browser push with mobile push. Email and in-app content must each be selected separately. These groups need different document types: Message, Push, Email, or In-app. Since one broadcast points to one document, selecting a channel from another group clears the previous channels.

Content. Choose a published document version of the type required by your channels. The broadcast stores the document ID and version number, so publishing a new version later does not change this broadcast. You can create content here or make an editable copy of the current document; after saving, the copy is published and selected. Messenger and web-chat messages support composer blocks including voice messages, GIFs, video notes, and carousels (on VK a carousel is sent as a series of messages).

For messengers, web chat, and Email, preview the content on each selected channel. Choosing a test recipient fills in their data, and Send test sends the content to that one verified contact. The test uses the same quotas and rate limits as a test send from Content studio: it does not change broadcast recipient rows, you cannot enter an arbitrary address, and too-frequent tests are rejected. Browser push, mobile push, and in-app content have a device mockup preview, but no test send from this editor.

Audience and experiment. Choose all bot subscribers, one ready segment, or bot_chats, meaning the groups, supergroups, and channels where the bot has been added. The chat list updates automatically when the bot is added to a chat or made an administrator. Select specific chats with checkboxes, or leave them unselected to send to every chat where the bot may post. This audience works only with Telegram; selecting it clears other channels. Chats where the bot has left or cannot post, and unknown IDs, are excluded with a reason (bot_left, cannot_post, chat_unknown). Subscriber substitutions ({{profile.*}}, {{event.*}}) are not filled in for chats. You can also attach a running A/B experiment.

Schedule. The broadcast timezone, how the time is interpreted, the send time and the rate.

  • Send at is optional: an empty field means "send as soon as it is confirmed".
  • Bot timezone: one absolute moment for everyone, computed in the broadcast's timezone.
  • Recipient local time: the same wall-clock time for each recipient. A person with no timezone, an unrecognised timezone, or a local time that does not exist for them that day gets the message at the broadcast's own time instead. This differs from email campaigns, where such a recipient is suppressed.
  • A send time that does not exist because of a daylight-saving change is rejected by the editor, with a message under the field.
  • Messages per minute: from 1 to 10,000. The send is spread out in batches of that size: at 60, the first 60 recipients get the start time, the next 60 a minute later, and so on.

If the bot has paid broadcasts enabled and Telegram is among the channels, a paid-broadcast toggle (Telegram Stars) appears together with a separate acknowledgement checkbox: the draft will not save without it. If the bot's paid-broadcast opt-in is turned off after the draft was created, the send does not fail: it simply goes out unpaid.

The pre-send check

Check recipients sends nothing. It resolves the audience against the current draft, channels and consent rules and answers how many people would receive the broadcast per channel, how many are suppressed and why, plus the content errors and warnings.

Suppression reasons:

  • no_identity: the subscriber has no contact on that channel;
  • unverified: the contact is not verified;
  • blocked: the subscriber is blocked;
  • channel_disabled: the channel is not connected or not enabled for the bot;
  • email_unsubscribed, email_bounced, email_complained, email_no_consent: the email consent state is not "subscribed";
  • wa_window_closed: no inbound WhatsApp message from that person in the last 24 hours, so a free-form message cannot be sent.

Errors block the launch:

  • content_not_published: the selected content version is missing or not published;
  • audience_not_ready: the segment is missing or has not been recomputed yet;
  • experiment_not_ready: the attached experiment is not running or does not assign by subscriber;
  • empty_audience: nobody would receive the broadcast;
  • per-channel content validation errors: the field names the channel, for example content:wa.

Content warnings appear next to them and do not block anything. The launch button only becomes active with zero errors and at least one recipient.

Launching and controlling a broadcast

Only a draft can be launched, and only from the check block: the button reads Send now or Schedule, depending on whether a send time is set. Launching requires an explicit confirmation and the revision currently on screen.

When you launch, the platform fixes the content version, channels, audience, experiment, and schedule together. It queues a message with a calculated send time for each eligible recipient. For each excluded recipient, it records a reason without creating a message. The report can therefore distinguish an excluded recipient from a failed send.

After launch the broadcast is immutable: the editor opens read-only and tells you to create a new broadcast if something has to change.

Statuses: Draft, Scheduled, Sending, Paused, Completed, Cancelled, Failed, Archived. Which actions are available from which status:

  • Pause: from Scheduled or Sending. Messages not yet picked up stop being dispatched; those already sending are delivered.
  • Resume: from Paused, back to Sending.
  • Cancel: from Draft, Scheduled, Sending or Paused. Everything not yet delivered is marked cancelled.
  • Archive: from Completed, Cancelled or Failed. Archived broadcasts are shown by the Show archived broadcasts toggle.

Every control action carries the expected revision, so two operators cannot silently overwrite each other: the second one gets "This broadcast was changed elsewhere".

Report and recipients

Clicking a broadcast in the list opens its results. Each channel shows Queued, Sent, Delivered, Failed and Suppressed. While the broadcast is scheduled, sending or paused, the report refreshes itself every 15 seconds.

For a broadcast to bot chats and channels, the report also has a By chat table with each chat's name, delivery status (queued, sent, failed, or cancelled), or exclusion reason. Telegram channel counts refer to chats, not subscribers. Pause, resume, cancel, and the messages-per-minute limit work the same way as for subscriber broadcasts.

The Delivered counter stays at zero at this level: the platform knows a message was handed to the channel, but delivery receipts are not collected into this report. For email, the detailed picture: opens, clicks, bounces, complaints: lives in email campaigns.

Blocked the bot after this broadcast counts blocking events within 24 hours of this broadcast's delivery. A block is attributed to the most recently delivered broadcast in that window. It counts events rather than unique subscribers. To export the list, open People → Blocked the bot → Export CSV; the file includes blocked_at.

Recipients opens a paged list showing only the channel, status (Waiting, Suppressed, Queued, Sending, Sent, Failed, Cancelled), and exclusion reason. There is no separate recipient column.

One limitation to keep in mind: if the content expands into several messages for one recipient, a broadcast sends only the first one. A broadcast keeps a single recipient row per (recipient, channel) pair.

Broadcasting a reaction from the reactions list

The older flow works exactly as before. In the Reactions list every reaction has a send-icon button with the "Send reaction to subscribers" tooltip. Clicking it queues the broadcast right away: there is no draft, no pre-send check and no per-channel report here. The dialog can turn on a paid Stars broadcast if the bot allows it. The Broadcast button on the Overview screen leads to the same reactions list.

This broadcast goes to every active subscriber of the bot. Narrowing it by labels is only possible programmatically, through the labels parameter of the API request (see REST API and tokens); there is no label picker in the interface.

Its statuses are its own and shorter: pending, running, done, cancelled, plus a paused flag. Pause, resume and cancel are available over the API, and the broadcast survives a restart: it runs page by page and continues from the stored cursor.

Scheduled broadcasts

A reaction with the "On schedule (dates)" trigger fires at the specified times and is broadcast to active subscribers automatically. This is how you build regular digests and reminders without launching anything by hand. A retry after a failed run may deliver a duplicate to some subscribers: for scheduled announcements a rare duplicate is considered better than a silently lost run.

What's next