Instagram Auto Reply
What to Actually Write in the Four Messages That Matter (2026)
Quick answer: Most Instagram auto-reply setups ship the same four pieces of copy, and the writing gets far less attention than the wiring. The four ice breakers a stranger sees before typing anything. The keyword you ask people to comment. The private reply that keyword earns you. The away message that runs when nobody is at the desk. Wiring these takes an afternoon and nobody writes about the writing, which is why so many accounts have a working automation that sounds like a parking meter. This guide is only about the words.
The mechanics live elsewhere on this site and are not repeated here. Our Instagram DM automation entry covers what the term means and what Meta's API permits, and our Instagram channel guide has the platform shortlist, the pricing structure, the ordered setup sequence, and the compliance surface. Read either of those for what to build. Read this for what it should say.
Why Instagram copy is its own problem
Generic chatbot welcome-message advice mostly transfers, and our welcome message guide is the right starting point for the principles. Three things about this channel break the defaults.
The audience arrived from content, not from a support page. Someone who opens a chat widget on your pricing page has a commercial question. Someone who comments on a Reel was scrolling. The register that reads as professional on a website reads as a form letter here, and the gap is wider than most brand guidelines account for.
The reply is competing with a personal inbox. Instagram DMs are where people talk to their friends. Your message lands between two of those, in the same bubble shape, with the same notification sound. Anything that reads as a mass send is immediately obvious in that context in a way it never is in an email client.
You get one message that counts. The comment-to-DM pattern sends a single private reply tied to that comment, and that reply opens a 24-hour window in which you are technically free to keep sending. Treat that freedom carefully. The first message is the one that decides whether anything after it is welcome, so it has to deliver the thing, prove you are real, and give the person a reason to answer.
Message one: the four ice breakers
Ice breakers are the tappable prompts that appear when someone opens your DM thread and has not written anything yet. Meta caps them at four. Many accounts leave all four blank, which is free surface area shown to people who have already opened the thread and not yet decided to type, and the cheapest thing on this list to fix whatever it turns out to be worth on your account.
The instinct is to write four questions you want to answer. Write four questions your customer wants answered instead, in their words, and make each one lead somewhere concrete.
| Weak ice breaker | Better | Why |
|---|---|---|
| "Learn about our services" | "How much does a session cost?" | Prices are the actual question, and pretending otherwise costs you the tap |
| "Contact us" | "Can I speak to someone today?" | Names the outcome instead of the mechanism |
| "FAQ" | "Do you ship to Canada?" | A category is not a question; your top-asked one is |
| "Shop now" | "What's back in stock?" | Gives a reason to tap that is not a purchase decision |
Four rules worth applying, in roughly the order that skipping them shows up in a broken ice-breaker set. Write them as first-person questions, because the customer is the one asking. Keep them under about six words, since they render as chips and long ones truncate on a phone. Make at least one of the four a route to a human, because the people who most want that are the people you least want to lose. And derive them from your actual DMs rather than from imagination: pull two weeks of messages, count the questions, and take the top three. The reading habit that produces that list is the subject of our conversational analytics guide.
Message two: the keyword, and the caption that carries it
Comment-to-DM asks people to comment a word. Choosing that word is treated as trivial and is not, because the word is doing two jobs at once: it has to be memorable enough to type from muscle memory, and distinctive enough that your automation can tell it apart from ordinary conversation.
The failure mode is collision. The generic picks — GUIDE, LINK, INFO, YES — are the ones every walkthrough reaches for, including the worked example in our own channel guide, and they are exactly the words that turn up in ordinary comment traffic. Use one and you will fire private replies at people who wrote "yes please" under someone else's comment. Every one of those is a stranger receiving an unexpected DM, and an unexpected DM is the kind of message that gets reported or blocked, which is what Meta's quality signals do measure.
What works better: a word specific to the offer rather than to the action. STRETCH for a mobility program, SOURDOUGH for a bakery's recipe, SEPTEMBER for a dated cohort. Specificity buys you three things at once — fewer false positives, a keyword you can retire cleanly when the campaign ends, and a comment section that reads like enthusiasm rather than a captcha.
Two operational notes. Rotate keywords between campaigns rather than reusing one house word forever, for the pattern-matching reason our channel guide sets out in its setup sequence, and because it keeps your matching clean. And state the keyword in the caption in the exact case and spelling your automation expects, then configure the automation to be forgiving anyway; people will type it lowercase, with a typo, and inside a sentence.
Message three: the private reply
This is the highest-stakes copy on the channel. Mechanically you are not limited to it: sending the private reply opens the 24-hour window our channel guide documents, and you can follow up inside it whether or not the person answers. Practically you should write as though you were limited to it, because a first message that lands badly turns every permitted follow-up into an annoyance.
Four jobs, in this order:
Deliver immediately. The thing you promised goes in the first line, not after a greeting and a paragraph of positioning. Someone typed a word to get a link; make the first thing they see be the link. Withholding it to "build the relationship" reads as a bait-and-switch because it is one.
Sound like a person who has read the comment. Referencing what they commented on, not just that they commented, is the cheapest credibility available. "Here's the sourdough starter guide from Tuesday's Reel" does work that "Thanks for your interest!" cannot: it proves a human read the comment. It costs one variable.
Ask one question. The window is open because your private reply opened it, and it stays useful because they reply. One easy, specific question at the end does that work: "Are you baking in a Dutch oven or on a stone?" asks for something answerable in three words, while "Let me know if you have questions" asks the customer to invent the next turn, which is work most people will not do. This is also the moment your bot learns something worth storing.
Include the exit. One short line telling people how to stop hearing from you, in plain words rather than legal ones. Some people will use it, and that is the point: an opt-out taken is a report not filed, and account signal is the thing every other automation on this list depends on. Our channel guide puts the opt-out in the first follow-up; on a private reply that may be the only message someone reads, we would put it there instead.
What to leave out: a greeting that burns the first line, more than one link, emoji used as bullet points, and any sentence that would appear verbatim in a press release. If the message would read identically for a different business, rewrite it.
Message four: the away message
The always-on responder is the one piece of copy that runs when you are not there, and the standard version does no work at all. Instagram's app has long carried a simple saved-reply and away-message feature of its own, and we have not re-checked its current state; what follows applies whether you write the copy there or in your platform's always-on responder. "Thanks for your message! We'll get back to you soon." tells the customer nothing they did not already assume and sets up an expectation nobody has defined.
Three ingredients make it work. Say when, with a real number: "We read DMs between 9 and 6, Monday to Friday, so this one lands Monday morning." Give an alternative for urgency, because the person with an actual problem needs a route that is not waiting: a phone number, an email, a link to the booking page. And answer the most likely question anyway, since the questions that arrive after hours are often the same three, opening hours and price and location, which you can check against your own inbox in ten minutes. Answering them unprompted turns a holding message into a resolution.
The tone rule here is different from the others. This message is the only one that admits a human is absent, so understatement works better than warmth. Read our greeting message entry for the general version of this, then make yours shorter than that.
The follow-up sequence, and why it is shorter than you think
Sending the private reply opens a 24-hour window, and inside it you can keep talking whether or not the person answers. The temptation is to fill that window with the sequence you would build in email. Do not, for a reason specific to this channel: the thread does not disappear the way an email does. It sits in a personal inbox alongside real conversations, and each additional message is visibly stacked on the last.
A working shape for most SMBs is two more messages, not five. One that does the job the person came for, and one that either closes or hands to a human. Our channel guide puts the hygiene ceiling at four follow-ups inside the window; two is the editorial recommendation, because beyond that you are spending goodwill you will want when the same person comes back in three months. The human handoff is the more valuable investment here than message four of a nurture: on a channel where the customer is one tap from muting you, a fast route to a person is a feature people notice.
A testing routine that takes twenty minutes
This is the copy-level version of the pilot test in our channel guide's setup sequence. Auto-replies fail silently, and on this channel the failure is invisible because the person who did not get a message does not complain to you.
- Comment your own keyword from a second account that does not follow you, and time how long the private reply takes. Not your own account, which behaves differently and may be excluded by your platform.
- Comment something that contains the keyword inside a sentence, and confirm whether it fires. Then decide whether that is what you want, because both answers are defensible.
- Open the thread as a stranger and check that all four ice breakers render, that none truncates, and that each one routes somewhere.
- Send a DM outside business hours and read the away message on a phone, in the notification preview, not in the builder.
- Ask for a human in three different phrasings and confirm all three escalate.
Run this again after any campaign that changes the copy. Our chatbot QA testing protocol is the fuller version for larger builds, and the same discipline applies at this scale in miniature.
What the platforms give you
The copy fields differ less than the trigger tooling does, and this is a channel where platform depth genuinely varies. From our platform reviews and the channel-guide comparison: Manychat is the builder whose product identity is Instagram-first, with per-post keyword rules as a first-class object; Chatfuel sits alongside it as the Meta-ecosystem specialist, with post-level keyword triggers documented by the vendor rather than measured by us; Tidio routes Instagram into a unified inbox alongside its other channels, which suits a team already answering web chat; and SendPulse treats it as one of several messenger surfaces in a broader marketing suite. Our Instagram channel guide has the nine-platform comparison with the tier at which Instagram unlocks on each, and the ranked best Instagram chatbot list is the shortlist version.
Chatbotscape earns affiliate commission on some of the platforms named above. It does not change what we publish. See our affiliate disclosure.
One capability worth checking specifically, because it is not on any feature grid and it determines whether the copy above is maintainable: whether your platform lets you set ice breakers from inside the product or sends you to Meta Business Suite to do it. Both work. One of them is inside the tool your team already opens every day, and the other is not.
Frequently asked questions
How do I set up an auto reply on Instagram?
Through a chatbot platform connected to your Instagram professional account, not through the Instagram app alone. The app offers simple saved replies and an away message; automated replies triggered by comments, keywords, or story replies need the messaging API, which is what the platforms in our channel guide provide. The ordered setup steps are in that guide; the copy for each message is on this page.
What should an Instagram auto reply say?
Deliver whatever was promised in the first line, reference the specific post or question rather than thanking someone generically, ask one easy question so the conversation stays open, and include a short opt-out. Skip the greeting, keep it to one link, and cut any sentence that would read identically for a different business.
What are Instagram ice breakers and how many can I have?
They are tappable prompts shown when someone opens your DM thread before typing, and Meta caps them at four. Write them as short first-person questions in the customer's words, derived from your actual DM history, with at least one routing to a human. They are frequently left blank, and configuring them costs nothing but the writing time.
Why is my comment-to-DM automation not firing?
Three causes are worth ruling out first. The keyword collides with ordinary comment language, or the person typed a variant your matching does not catch. The account or the post is not in the state the API requires, which our glossary entry covers, including the public-account condition for comment notifications. Or the connection itself is stale: our Manychat review records operators reporting exactly this, a lost connection to an Instagram or Facebook profile that causes automations to fail silently rather than error. Test from a second account that does not follow you before assuming the copy is the problem.
Should the away message and the private reply say the same thing?
No, and running one piece of copy in both places is a common shortcut that costs you both jobs. The private reply answers a specific ask from someone who just typed a keyword, so it delivers the thing and asks a question. The away message answers an unpredictable question from someone who does not know you are closed, so it gives a time, an urgent alternative, and the answer to whatever is most often asked. If they read identically, the away message is almost certainly the one doing no work. For the boundary on who you are allowed to message at all, see Instagram DM automation.
How many follow-up messages should I send?
Two, for most small businesses, counted from the private reply rather than from the customer's response. Our channel guide puts the hard hygiene ceiling at four follow-ups inside the 24-hour window; two is the editorial recommendation, because the thread sits in a personal inbox next to conversations with actual friends and each extra message is more visible than the equivalent email.
Related guides
- Instagram DM automation (glossary) — what the term means and what Meta's API permits
- Instagram chatbots — complete guide — platforms, pricing, setup sequence, compliance
- Chatbot welcome message design — the general principles this page specializes
- Conversational analytics guide — how to derive your ice breakers from real DMs
- Chatbot QA testing protocol — the fuller version of the twenty-minute routine
- Best Instagram chatbot platforms 2026 — the ranked shortlist
Sources
- Chatbotscape. Instagram chatbots — complete guide 2026. /channels/instagram-chatbot-guide — messaging-window mechanics, private-reply send windows, published rate limits, the nine-platform comparison and tier requirements, and the compliance surface, all deliberately not re-derived on this page.
- Chatbotscape. Instagram DM automation. /glossary/instagram-dm-automation — the term's three senses, the account and access prerequisites read from Meta's Instagram Platform documentation on 5 August 2026, and the API-side data-retention limits.
- Chatbotscape platform reviews. Vendor-page verification 26 May 2026 across all four; authenticated hands-on sessions 26-30 May 2026 (SendPulse 26-27 May, Manychat 27 May, Tidio 28-29 May, Chatfuel 30 May). Instagram channel connection was exercised first-hand on Chatfuel only; Instagram OAuth was not run on Manychat or SendPulse per those reviews' evidence boxes, and on Tidio the Instagram surface was observed in the inbox and flow-trigger picker rather than connected, so the Instagram characterizations here rest on vendor-page verification plus our Instagram channel guide comparison table rather than on first-party Instagram deployment: Manychat, Chatfuel, Tidio, SendPulse.
- Chatbotscape evaluation methodology. /methodology (continuously updated).
About this guide
Chatbotscape launched in 2026 as an independent review site for chatbot platforms. This guide is part of our SMB chatbot Academy. It covers the copy layer of Instagram automation only; the build, the platform choice, and the policy surface are handled by the two pages linked at the top. To share what has worked on your own account, write to editorial@chatbotscape.com.
Methodology
The copy patterns, the four-message frame, and the twenty-minute test routine are editorial working recommendations derived from Instagram's documented messaging constraints and from conversation-design principles we apply across channels; they are not the output of a controlled test, and we publish no conversion figures of our own on this channel because we have not run one. Platform characterizations carry over from Chatbotscape's vendor-page verification of 26 May 2026 and from authenticated sessions run between 26 and 30 May 2026, with the caveat stated in the Sources block: Instagram channel connection was exercised on Chatfuel only, and two of the four reviews record Instagram OAuth as not run. The Instagram API constraints referenced here were read from Meta's Instagram Platform documentation on 5 August 2026 and are cited in full in the companion glossary entry. See our methodology for how platform facts are verified.
Last updated
7 August 2026 — Initial publication aligned to methodology v3.12.1. Next scheduled refresh: 7 November 2026.