Is Instagram DM automation allowed? What Meta's rules say and how to stay inside them

By Ranjit Bhinge, founder of Blur Studio · Updated · 7 min read

Short answer: Instagram DM automation means software that sends or answers direct messages (DMs) on an Instagram business account for you. Instagram is the photo and video app owned by Meta (the parent company of Facebook and WhatsApp), and Meta publishes an official route for this kind of software to work with professional accounts. As of November 2026, automation through that route is allowed, but with rules: you can message a person only after they message you or comment, replies must stay inside set time windows, and you must offer a way to reach a human. Meta's rules change, so check its current pages, and no tool can promise your account will never be restricted.

Ranjit Bhinge

You want to automate replies on Instagram, and you have heard it can get your account banned. Both things are true in part.

I'm Ranjit, and I run Blur Studio: 15+ years building software, 120 apps taken live, and clients who have gone on to raise $10M+. I build AI systems and processes that help businesses do a lot more with a lot less time. I also work differently from most of what else is out there, more on that further down this page. Say hello.

Yes, it is allowed when it runs through Meta's official messaging route and follows Meta's rules. Meta lets approved software read and answer messages for Instagram professional accounts. What is not allowed, or not possible, is messaging strangers first, ignoring time windows, or hiding that a bot is replying. Even a fully compliant setup can be restricted if Meta's systems think it looks like spam, and the tool vendors say they cannot reverse that.

This guide explains what DM automation is, what Meta's documents say, what is risky, a safe-use checklist and a test plan. I use plain words and name the source page for each rule. This is information, not legal advice. As of November 2026, check Meta's current rules before you build anything.

What DM automation is

A DM is a private message inside Instagram. DM automation is any setup where software does part of the work. The common types are:

  • Auto-replies. A first answer to every new message, such as "Thanks, we reply within an hour."
  • Keyword replies. A customer types a word like "price" and gets a set message.
  • Comment to DM. Someone comments on a post, and the account sends them a private message with a link or answer. Meta calls this a private reply.
  • Story reply and mention flows. A reaction to a story reply or a story mention triggers a message.
  • AI replies. A language model, a type of AI that writes text, drafts answers from your own information.

Tools such as ManyChat, a chat automation product for Instagram, WhatsApp and other channels, build these for you. You can also use Meta's own Business Suite, the dashboard Meta provides for business accounts, which has built-in automated replies. See how the two compare.

What Meta's documents allow

These rules come from Meta's developer pages for the Instagram messaging API. An API is the official door that lets software talk to Instagram.

  • Professional account only. You need an Instagram business or creator account. Personal accounts cannot use the messaging API.
  • The customer starts it. A conversation begins only when an Instagram user messages your account through your feed, posts, story mentions or other channels. Your software can send a message only after that.
  • The standard window. You have the duration of the standard messaging window to reply. For most user actions it lasts 24 hours. After a first message from someone who tapped a Click-to-Direct ad, it may last up to 7 days. A Click-to-Direct ad is an Instagram ad with a button that opens a DM. See the 24 hour messaging window explained.
  • Private replies to comments. You may send one private reply per commenter, within 7 days of the comment. For Instagram Live, only while the broadcast is running. The reply lands in the person's inbox if they follow you, or in the Requests folder if not. You may follow up only if they reply, and then only within 24 hours of their reply.
  • Human Agent window. If a person answers and needs more time, Meta describes a separate 7 day Human Agent window for manual replies.
  • Bot disclosure. Where the law requires it, automated chats must say the person is talking to an automated service: at the start, after a long gap and when a chat moves from a human to a bot. California and German users are named. Meta recommends doing it everywhere.
  • A way to a person. Meta's docs describe an escalation path for automated chats, either a custom inbox or a handover to another app.

Limits that matter

Meta's rate limiting page lists these figures per Instagram professional account. A call is one request that software sends to Instagram:

  • Send API: 300 calls per second for text, links, reactions and stickers; 10 calls per second for audio or video.
  • Private replies: 750 calls per hour for posts and reels, and 100 calls per second for Live comments.
  • If too many messages go to one thread, your app may be rate limited.
  • An account sending and receiving more than 72,000 messages can lose inbox display and sending until volume drops.

Most small businesses never come near these. The numbers that people quote online, such as "200 DMs an hour is safe", are not in Meta's documents. Meta does not publish a safe daily limit, so treat any such number as a guess. ManyChat's own help page says Meta can flag or restrict accounts, and that ManyChat has no hotline to reinstate them.

What tends to get accounts into trouble

ManyChat's help center names these triggers: too many automated messages too fast, aggressive copy-paste comment automation in public, and running several Instagram accounts with overlapping behavior. Meta's own rules add more ways to break the terms: messaging without a prior message, ignoring windows, and missing disclosure. Meta does not publish how its spam detection decides, so the safest habit is to behave like a helpful person, not a broadcast.

Plan for these failure modes:

  • A comment flow that replies to every comment, including "spam" and unrelated ones.
  • One link or message copied to everyone with no variation or choice.
  • A bot that cannot stop, so a customer who types "stop" or asks for a human is ignored.
  • A trigger that fires twice, so people get two messages.
  • A tool bug. ManyChat's status page lists an open incident since March 2026 where its "Follow to DM" trigger does not fire for some accounts because of a Meta-side issue. Comment to DM and story replies are listed as not affected. Do not build your offer on a feature with an open incident.

Worked example (made-up numbers)

A coffee roaster posts a reel with "Comment GUIDE and I will DM you our home brewing guide." About 300 people comment in the first week.

A safe setup looks like this:

  1. The tool sends one private reply to each commenter, within the 7 day limit, with a short message that says it is an automated assistant, gives the guide link and offers two buttons: "Show me beans" and "Talk to a person".
  2. It does not send a second message unless the person replies.
  3. If the person replies, the 24 hour window opens and the bot may continue. If they ask a human question, it pauses and alerts the owner.
  4. The owner replies manually inside the window, or within the 7 day Human Agent period, if the account is using it.

A risky setup replies to all 300 comments with one long sales message and a discount link, then sends a follow-up the next day to people who did not answer. The second message is outside the rules for private replies, because follow-ups are allowed only after the person replies. These numbers are invented to show the pattern, not to show what results to expect.

Safe-use checklist

  1. Use a professional account and connect through an official partner tool or Meta's own tools.
  2. Reply only to people who messaged or commented, and only inside the window.
  3. Add a bot disclosure line to the first message.
  4. Give a clear way to reach a person, and name who answers and when.
  5. Keep replies short and varied. Let the person choose what they get.
  6. Limit comment triggers to a keyword, not every comment.
  7. Honor "stop" and "agent" requests at once.
  8. Do not run several accounts with the same flows at the same time.
  9. Start small. Switch on one flow, watch it for a week, then add another.
  10. Keep a record of what is live and who changed it, so you can undo fast.

For agency work with several client accounts, see safe DM automation for agencies.

Test plan with pass and fail criteria

  1. Disclosure test. Send a message from a second account. Pass: the first reply states it is automated. Fail: it reads like a person.
  2. Window test. Message the account, wait past the window, and see whether any automation still sends free text. Pass: it does not. Fail: it sends, so fix the flow.
  3. Comment test. Comment twice on the same post from one test account. Pass: only one private reply. Fail: two messages.
  4. Human handoff test. Type "talk to a person". Pass: the bot stops and you get an alert. Fail: it keeps talking.
  5. Week-one review. Read the first 20 real conversations. Pass: no complaints, no odd replies. Fail: any complaint about spam. Pause and rewrite.

Who should skip DM automation

  • You want to message people who never contacted you. Instagram does not allow your software to start those conversations.
  • You use a personal account. The official route needs a professional one.
  • You get only a few DMs a week. Replying yourself is faster and safer.
  • You cannot watch the flows. An unwatched flow can annoy people for weeks.
  • You need a guarantee against bans. Nobody can give one, including the tool vendors.

If the main goal is to get customers to book, the topic in booking from Instagram DMs shows a gentler setup. For the specific comment trigger, see comment to DM automation. If you want to understand how ManyChat handles the account side, ManyChat Instagram automation covers it.

How I'd approach this

I would treat DM automation as a helper for people who already raised their hand. They commented, asked a question or replied to a story. That is exactly the group Meta's rules protect and where automation helps most. I would use it only for people who already reached out.

Start with one flow: either an instant first reply or a comment keyword. Add the disclosure line and a human button before launch. Review real chats for a week. Then decide on a second flow. If you want a version that answers from your own business knowledge and passes leads into your CRM (customer record system) and calendar, or you need a half-built setup finished, Blur Studio builds and maintains that, and you can read what this includes. Either way, re-read Meta's pages for any new rule before each change.

Questions people ask

Is Instagram DM automation against Instagram's rules?

Not by itself. Meta provides an official messaging route for professional accounts. The rules are about how you use it: message only after the person contacts you, stay inside the time windows, disclose bots where required and offer a way to a human. Check Meta's current rules, because they change.

Can automation message people who do not follow me?

Only if they messaged or commented first. Meta says a private reply to a comment goes to the inbox if the person follows you, or to the Requests folder if not. Cold messages to people who never contacted you are not something the official route lets software start.

Will I get banned for using DM automation?

No tool can promise you will not be restricted. ManyChat's help center says Meta can flag or ban accounts even when its tool is used properly, and that it cannot reverse bans. Following the rules and starting small lowers the risk. See the safe automation guide.

How many DMs can I send per hour?

Meta does not publish a safe hourly figure for normal DMs. It does publish technical limits, such as 750 private replies per hour for posts and reels. Treat any other number you read online as unverified.

Do I have to say it is a bot?

Meta says automated chats must disclose this where the law requires it, and recommends it as good practice everywhere. California and German users are named. This is not legal advice, so check local law.

Can I use free text after 24 hours?

Usually not. The standard window is 24 hours for most actions, and up to 7 days after a Click-to-Direct ad message. A person can use the separate 7 day Human Agent window for manual replies. See the 24 hour window guide.

Is Meta Business Suite automation different from a chat tool?

Yes. Meta's own tool offers templates such as instant reply, away message and keyword replies. Its custom keywords are limited to 5 and are case-sensitive, and keyword replies are sent after 15 minutes unless a person replies first. A chat tool offers more flows. See the comparison.

More in this topic: ManyChat: rules, basics and problems (5 guides) →

Sources

Ready for the next step?

A 30 minute call. If you do not need me, I will tell you.

Ranjit Bhinge You talk to me, Ranjit. I reply within a few hours.

Not sure what you need? Tell me what is going on. I will point you the right way, even if it is not me.

Blur Studio · 30 minute call

Book a 30 minute call

You talk to me, Ranjit. If you do not need me, I will tell you.

  • We talk about your business and what is going on.
  • You leave with a clear next step, even if it is not me.
  • I reply to every message within a few hours.
Loading the calendar…