Editorial flat-vector illustration for How to Build a Facebook Messenger Chatbot in 2026 — The Build Path, and the Four Things a Pre-2026 Tutorial Cannot Tell You

Published 26 August 202625 min read

How to Build a Facebook Messenger Chatbot in 2026 — The Build Path, and the Four Things a Pre-2026 Tutorial Cannot Tell You

Q: What happened to the Messenger message tags in 2026?

Three of them stopped working. API requests containing `CONFIRMED_EVENT_UPDATE`, `ACCOUNT_UPDATE` or `POST_PURCHASE_UPDATE` return error code 100 from 27 April 2026, per Meta's changelog. The replacement path is Utility Templates or the Marketing Messages API. `HUMAN_AGENT`, which extends the window to seven days for unresolved issues, is confirmed still alive on Meta's current policy and send-message pages.

Quick answer: Building a Messenger bot in 2026 takes a Facebook Page, a Meta app, a Page access token and a webhook. On a no-code platform you authorize the Page through the vendor's Facebook Login screen and the platform provisions all four for you; our own hands-on testing measured 12 minutes from signup to a working ten-question FAQ bot on Messenger with Manychat. The hard part is not the build. It is that Messenger changed underneath the tutorials. Meta moved the Messenger Platform documentation to a new URL tree and left the old one serving outdated content. Three message tags began returning error 100 on 27 April 2026. Recurring Notifications ended in most markets earlier in the year. The Chat Plugin is gone from both the developer docs and the Business Help Center. A guide written before April 2026 cannot account for any of the four, which is most of what is currently indexed. This one gives you the build sequence, the current limits, and the specific things not to build.

What this guide is and is not

Every Meta-side fact here was read from Meta's own current documentation on 25 August 2026 and is attributed in Sources to the page it came from. This is a build guide, not the channel case: whether Messenger deserves your effort at all is argued in our Messenger chatbot guide, and which platform to build on is ranked in best Messenger chatbot platforms. Platform facts and editorial scores are carried unchanged from our own reviews at their own verification dates. We did not build a bot from scratch for this guide; the sequence below is assembled from Meta's documentation and from what our platform reviews recorded during their own hands-on sessions, and the difference is marked wherever it matters. This guide contains no affiliate links. Some reviews it links to do; see our affiliate disclosure.

Before anything: you are probably reading the wrong docs

Meta migrated the Messenger Platform reference out of developers.facebook.com/docs/messenger-platform/ and into developers.facebook.com/documentation/business-messaging/messenger-platform/. The old tree still resolves. It also still serves outdated content: its changelog stops in November 2025 and it still names v24.0 as the latest API version, while the current Graph API version is v26.0, released 29 July 2026.

This matters more than a broken bookmark, because the old tree is what search engines have indexed and what most third-party tutorials link to. If a guide cites a /docs/messenger-platform/ URL, treat every policy claim in it as unverified until you have checked it against the current tree.

That rule applies to us. Our own Messenger chatbot guide cites five legacy /docs/ paths, not one: the platform root, the changelog, the Send API reference, the policy overview and the marketing-messages tree. We applied our own rule rather than exempting ourselves, re-read the corresponding current-tree pages on 25 August 2026, and the policy facts that guide reports are unchanged. The URLs are wrong and are queued for correction on the editorial track.

One more version detail worth knowing before you copy a code sample. The Messenger docs' own examples still call graph.facebook.com/v25.0/. The current version is v26.0. Meta has not updated its own samples, so do not treat a sample's version string as authoritative.

Step 1 — Decide whether you are building on a platform or on the API

Both are legitimate and the choice changes everything downstream.

On a no-code platform, you connect your Facebook Page through the platform's OAuth screen and it handles the app, the token, the webhook and the permission requests. The only first-bot time our corpus measured on Messenger itself is Manychat's: 12 minutes from signup to a working ten-question FAQ bot, in a hands-on session, with 89 percent intent accuracy on a 20-query English test set. Treat that as the fast end rather than the average, since Manychat grew up on this channel. The tradeoffs of the route are that you inherit the platform's interpretation of Meta policy, and you find out about a retired tag when a flow silently stops delivering.

On the API directly, you create a Meta app, request permissions, host a webhook endpoint and write the send logic. Meta is explicit that this does not require App Review if the bot serves only your own Page: "If you are developing the app to send and receive messages for your own page, you will not have to go through app review." That single sentence saves a lot of unnecessary anxiety, and it comes with a limit worth reading twice, covered in step 3.

For most small businesses the platform route is correct, and the ranked options are in best Messenger chatbot platforms. The short version from that page: SendPulse holds the highest editorial score at 86, Manychat 84 with the deepest Meta-surface work, Tidio 75 for support-inbox use, BotPenguin 70 on a free tier, Chatfuel 69 on flat pricing at volume. Those numbers come from the full reviews and this page does not recompute them.

Step 2 — Prerequisites and permissions

Four things exist before a bot can send anything: a Facebook Page, a Meta app, a Page access token, and a webhook subscription. Meta's overview is blunt about the asymmetry that governs the whole channel: the person must initiate. There is no cold open on Messenger.

The permissions, requested through Facebook Login for Business, are pages_show_list, pages_manage_metadata, pages_messaging and pages_read_engagement, plus business_management. That last one is easy to miss and worth calling out: it is a dependency of pages_messaging and pages_show_list, and Meta advises stating that dependency explicitly in an App Review submission. The person who grants the token must hold the MESSAGING and MODERATE tasks on the Page.

Business Verification is also required. This is not new, and any tutorial presenting it as a 2026 change is wrong: apps that had not verified by 19 July 2022 had pages_messaging revoked.

Step 3 — Understand Standard Access before you demo to anyone

Meta's default is Standard Access, which restricts your app's data to people who hold a Role on the app or the Page. Advanced Access, which lets the bot serve the public, requires App Review.

The practical consequence catches people out. Your bot will work perfectly for you and for your colleagues and will do nothing for a customer, and nothing in the interface announces why. If you are building for your own Page you can skip App Review for messaging itself, but fetching user profile information still requires it. Plan the review submission before the launch date rather than after the first failed test.

Step 4 — Set up entry points, and pick one greeting mechanism

Meta lists the ways a user can reach you: m.me links, Click-to-Messenger ads, the Checkbox Plugin, the Send to Messenger Plugin, the Message Us Plugin, Private Replies to comments, and the Welcome Screen.

Two configuration traps live here. Both are documented, in the Messenger Profile API reference and the changelog, which is not where anyone building a first bot starts looking.

Ice Breakers and the Get Started button are mutually incompatible. Precedence runs: Ice Breakers set through the API override the Get Started button, which overrides custom questions set in the Page Inbox UI. Setting Ice Breakers through the API also disables custom-question editing in the Page Inbox. Ice Breakers cap at four questions. Pick one mechanism deliberately, because configuring two produces a silent override rather than an error.

A Commands menu, if configured, overrides the Persistent Menu. Same class of problem: you build a menu, it never appears, nothing tells you why.

For what to actually put in that first screen, our chatbot greeting message entry covers the copy and chatbot quick reply covers the button pattern. Note one Messenger-specific constraint our Tidio review caught in the builder: the editor surfaces an inline 20-character truncation warning for Facebook Messenger quick-reply labels. Write the button text to that limit rather than discovering it in production.

Do not build the Chat Plugin. The website chat widget Meta used to offer for embedding Messenger on your own site is gone. The Business Help Center article about it now redirects to a generic business-messaging-tools page that does not list it, and the current developer Discovery page omits it while still listing the three other website plugins. Vendor write-ups date the shutdown to 9 May 2024; we could not find that date on a Meta-hosted page and are not asserting it. If you want chat on your website, that is a different product decision, covered in add a chatbot to your website.

Messenger Codes are gone too. The old scannable circular codes appear nowhere in current documentation. Generate a QR code from an m.me link instead, which is what Meta's own Discovery page suggests.

Step 5 — Build inside the 24-hour window

The standard messaging window is still 24 hours and still works the way it always did. Inside it you may send promotional content. Outside it you may not, except through the narrow paths in step 6.

Three messaging_type values govern a send: RESPONSE and UPDATE must both be inside the window, and MESSAGE_TAG is the out-of-window path and must be non-promotional and matched to an approved use case.

What opens the window is a longer list than most builders assume, and the last three items are the ones people miss:

the user sends a message
the user taps a call to action such as Get Started
the user clicks a Click-to-Messenger ad and then messages
the user starts a thread through Send to Messenger or the Checkbox Plugin
the user clicks an m.me link into an existing thread
the user reacts to a message
the user comments on a Page post
the user publishes a visitor post

Those last three are why comment-to-DM automation is a supported pattern rather than a workaround. A reaction is a window-opening event. Build the flow so the reply lands while it is open.

Design everything you can inside those 24 hours. Our chatbot conversation flow guide covers the structure, and human handoff covers the exit, which on Messenger has its own extended window.

Step 6 — What retired, and what replaced it

This is the section that separates a 2026 tutorial from a stale one.

Retired	When	What to use instead
`CONFIRMED_EVENT_UPDATE`, `ACCOUNT_UPDATE`, `POST_PURCHASE_UPDATE` tags	Requests return error code 100 from 27 April 2026	Utility Templates, or the Marketing Messages API
Recurring Notifications	Discontinued early 2026, except in AU, EU, JP, KR and UK	Marketing Messages
Sponsored Messages	Deprecated 31 July 2024	Marketing Messages
Chat Plugin	Removed from docs and Help Center; date not verified	A non-Meta website widget
Messenger Codes	Absent from current docs	QR code generated from an m.me link

Two survivors are confirmed on Meta's current pages. HUMAN_AGENT gives you seven days, intended for cases a human cannot resolve inside 24 hours, such as a weekend closure or a multi-day investigation. NON_PROMOTIONAL_SUBSCRIPTION still exists but is restricted to Pages in the News Page Index, so an ordinary small business cannot use it.

Two limits on this section, stated rather than glossed

We could not read the complete tag list. The Send API reference page that holds the enumerated table renders an empty body, and we tried the direct URL, the markdown variant, the legacy path and a real browser. So the table above is a list of what we could verify as retired, plus the two tags we could verify as alive. It is not a complete inventory, and the plausible inference that HUMAN_AGENT and NON_PROMOTIONAL_SUBSCRIPTION are the only survivors is our reasoning rather than Meta's statement. Check the reference page before you build on any tag not named here.

Meta's own docs contradict each other. The current Policy and Send-a-Message pages still describe Sponsored Messages as a live option for out-of-window promotional messaging, and still link One-Time Notification as "(Beta)", while the Marketing Messages FAQ says Sponsored Messages was deprecated on 31 July 2024. The same FAQ dates the Recurring Notifications shutdown as 12 January 2026 in one answer and 10 February 2026 in another. We have followed the changelog for the tag date, said "early 2026" for Recurring Notifications rather than picking between two Meta answers, and we could not resolve One-Time Notification's status at all because its only documentation link points into the stale tree.

Two corrections to our own pages

Auditing the corpus for this guide turned up defects on our side, and they are recorded here rather than quietly patched because a reader following our own internal links would have been misled by us in exactly the way this guide warns about. Reviews and glossary entries are edited on a separate editorial track, so both are flagged rather than fixed in this run.

One. Our chatbot broadcast entry describes the pre-2026 world in three sentences, not one. It says Messenger broadcasts reach users within "the last 7 days if using paid-message tags (Subscription Messaging, Sponsored Messages, etc.)"; it names POST_PURCHASE_UPDATE and ACCOUNT_UPDATE as tags that "allow longer-window broadcasts"; and its FAQ repeats the model for Instagram, saying broadcasts reach users from the last 24 hours "or 7 days with paid message tags." Sponsored Messages was deprecated in July 2024, and those two tags began returning error 100 on 27 April 2026.

Two. Our best Messenger chatbot platforms page states the Recurring Notifications shutdown as 10 February 2026 and Messenger's scale as "roughly 1.1 billion monthly users," three times. Both are defensible readings, and neither is as settled as the phrasing implies. This guide declines both figures on purpose: Meta's own FAQ gives two different RN dates, and third-party MAU estimates span 1.01 to 1.3 billion. That page is flagged for the same track, and the audit sentence in our Sources counts it as a defect rather than as current.

Step 7 — Know what Messenger actually costs

The most common cost error in 2026 is assuming Messenger now bills like WhatsApp. It does not, and the distinction is worth getting right before you build a business case.

Ordinary bot messaging is free. Meta states that the Messenger Platform is free to use, and that businesses are not charged for their replies inside the 24-hour window.

Marketing Messages are paid, billed per message actually delivered to the device, on the same bill as Meta ads, requiring an ad account with a payment method and a daily or lifetime budget. Meta publishes no rate card; partners use an Estimation API that returns expected delivery volume for a given budget, with pricing that varies by recipient country. Any specific per-message figure you find elsewhere is unsourced.

Two constraints decide whether Marketing Messages is even available to you. It runs in 20 markets, and because the list is the section's whole job, here it is in full: the US, Brazil, Mexico, Chile, Colombia, Peru, India, Indonesia, the Philippines, Malaysia, Singapore, Thailand, Vietnam, Taiwan, Hong Kong, Australia, New Zealand, Israel, Saudi Arabia and the UAE. It excludes the EU, the UK and Canada. And it is capped: one marketing message per unique user per day, a 12-hour cooldown per subscription token, and frequency caps enforced since 10 April 2026 that block sending when the ratio of marketing threads to organic threads exceeds 20 percent over a trailing seven days or 40 percent over one day.

If your audience is European, the practical reading is specific and cuts both ways. Your Messenger bot is free, Marketing Messages is not available to you, and your re-engagement options are the 24-hour window, HUMAN_AGENT, and Recurring Notifications, which Meta retained in the EU, UK, Australia, Japan and Korea when it retired the product everywhere else. That last one carries a caveat that decides whether it is useful to you: new integrations were blocked from 1 September 2025, so it is available if you already had it and not if you are starting now. Plan the flow around what you can actually send. Our chatbot broadcast entry covers the general pattern, with the correction noted above.

Step 8 — The policy limits that actually restrict accounts

Three of these are rarely covered and all three have teeth.

The 30-second responsiveness rule. Automated bots must respond to any and all user input within 30 seconds: freeform text, quick replies, CTA buttons, persistent-menu taps. Get Started is exempt. A bot with a disabled composer must still respond to everything else. Violations produce a notification in the Page Support Inbox, and if it is not fixed within seven days, Meta can limit the Page's messaging. This is the rule that punishes half-built flows, because a dead-end branch is a non-response.

Webhooks must return 200 OK within 20 seconds. If you are on the API route, make the handler acknowledge first and process afterwards.

Block rate is the quality signal Meta surfaces. Meta maintains a dedicated article on troubleshooting Messenger block rate and exposes the metric through the Insights API. Its guidance is worth following literally: make calls to action say explicitly that they open a Messenger conversation, give people a way out, test on a sample before scaling, and target people who messaged you recently. Note that block rate and opt-out rate are different signals. Opt-out applies to marketing messages and Meta says Page operations are not typically impacted by it; block rate is the one that reflects on the Page.

Published rate limits, for anyone building at volume: overall calls in 24 hours are 200 multiplied by the number of engaged users; the Send API allows 300 calls per second per Page for text, links, reactions and stickers but only 10 per second for audio and video; the Conversations API allows 2 per second; Private Replies 750 per hour; the Messenger Profile API 10 calls per 10 minutes. Above 40 messages per second the Page Inbox stops displaying and sending new messages until volume drops, which is a failure mode that looks like a platform bug and is not.

Step 9 — Test, then pilot small

This step is the one part of the guide that is procedure rather than documentation, and we are marking it as such because the rest of the page is sourced line by line and this is not. What follows is the test routine our own reviewers use, adapted to Messenger's constraints. It is judgment, not a Meta requirement.

Test against Standard Access reality: use an account with a Role on the Page, then a second account without one, so you see both behaviors before customers do. Walk every branch, including the ones you expect nobody to take, because the 30-second rule makes a dead end a policy problem rather than a UX problem. Send a message from a phone rather than only from a desktop preview.

Then pilot on a small share of traffic and watch block rate rather than volume. Our chatbot QA testing protocol covers the general checklist and chatbot escalation playbook covers the handoff path you will need on day one.

Realistic timelines

Route	To a working bot	Gating factor
No-code platform, own Page, no App Review	Hours	Flow design, not setup
No-code platform, public launch	Days to weeks	App Review, if the platform's app does not already cover it
Direct API, own Page	Days	Webhook hosting and the send logic
Direct API, public launch	Weeks	App Review plus Business Verification if not already done

These are ranges rather than measurements. We did not build a Messenger bot end to end for this guide, and the honest basis is Meta's documented requirements plus first-bot times our reviews measured on comparable channels. Treat the App Review row as the one most likely to surprise you.

FAQ

Do I need to know how to code to build a Messenger chatbot?

No. Every platform in our Messenger ranking connects a Facebook Page through an OAuth screen and handles the app, token, webhook and permissions for you. Code becomes necessary when you want behavior the platform does not expose, or when you want to own the integration rather than rent it. If you do go direct, Meta confirms App Review is not required for a bot serving only your own Page.

Does a Facebook Messenger chatbot cost money?

The bot itself does not. Meta states that the Messenger Platform is free to use and that businesses are not charged for replies inside the 24-hour window. You pay your chatbot platform, and you pay Meta only if you use Marketing Messages, which is billed per delivered message on your ads bill. Messenger does not work like WhatsApp here, and assuming it does will distort your business case.

What happened to the Messenger message tags in 2026?

Three of them stopped working. API requests containing CONFIRMED_EVENT_UPDATE, ACCOUNT_UPDATE or POST_PURCHASE_UPDATE return error code 100 from 27 April 2026, per Meta's changelog. The replacement path is Utility Templates or the Marketing Messages API. HUMAN_AGENT, which extends the window to seven days for unresolved issues, is confirmed still alive on Meta's current policy and send-message pages.

Can I still send Recurring Notifications on Messenger?

Not in most markets. Meta blocked new integrations from 1 September 2025 and discontinued the product in early 2026 everywhere except Australia, the EU, Japan, Korea and the UK. Meta's own FAQ gives two different dates for the shutdown, 12 January and 10 February 2026, which is why we say early 2026 rather than choosing between them. The successor is Marketing Messages, which is paid and is not available in the EU or the UK.

How long can my bot message someone?

24 hours from the person's last qualifying action, and that action list is broader than most people assume: sending a message, tapping a CTA, clicking a Click-to-Messenger ad and then messaging, using the Send to Messenger or Checkbox plugin, following an m.me link into an existing thread, reacting to a message, commenting on a Page post, or publishing a visitor post. Outside the window you need an approved tag or a template. HUMAN_AGENT extends it to seven days for issues a human cannot resolve in time.

Why does my Messenger bot work for me but not for customers?

Almost certainly Standard Access. Meta's default restricts your app's data to people who hold a Role on the app or the Page, so the bot behaves perfectly for you and does nothing for anyone else, with no error explaining it. Advanced Access requires App Review. This is the single most common launch-day surprise on the direct-API route.

Can I put Messenger chat on my website?

Not with the Chat Plugin, which is gone. Meta's Help Center article about it now redirects to a page that does not list it, and the current developer documentation omits it while still listing the Checkbox, Message Us and Send to Messenger plugins. For chat on your own site, use a website widget from your chatbot platform instead; our add a chatbot to your website guide covers the options.

Is Messenger still worth building on in 2026?

That is the channel question rather than the build question, and our Messenger chatbot guide argues it properly. The short version: Meta's own framing puts Messenger above a billion monthly users, and the 2026 changes tightened the policy surface without removing the channel. The honest caution is that the free-broadcast era ended, so build for conversations you can answer rather than for an audience you can push to.

Facebook Messenger chatbot guide — the channel case, the platform-support matrix, and the policy story in full
Best Messenger chatbot platforms 2026 — the ranked five, with editorial scores from the underlying reviews
How to build a WhatsApp chatbot — the same exercise on a channel where messaging is metered and templates are approved in advance
Instagram auto-reply setup — the sibling Meta surface, usually run from the same platform account
Telegram chatbot setup guide — the contrast case, since Telegram has no equivalent messaging window
How to build a chatbot — the channel-agnostic version of this sequence
Choosing chatbot channels — deciding whether Messenger belongs in your mix at all
Chatbot conversation flow — designing what happens inside the 24 hours
Chatbot QA testing protocol — the pre-launch checklist referenced in step 9
Chatbot escalation playbook — the handoff path, which on Messenger has a seven-day window of its own
Chatbot broadcast (glossary) — the general pattern, and the entry corrected by this guide
Human handoff (glossary) — the HUMAN_AGENT case in general terms
Instagram DM automation (glossary) — the neighboring Meta surface and its own window rules

Sources

Meta, Messenger Platform overview, current documentation tree, page stamp "Updated: Mar 23, 2026", read 25 August 2026. Source of: the four prerequisites; the user-must-initiate rule; the permission list including the business_management dependency and the MESSAGING and MODERATE Page tasks; the Standard versus Advanced Access distinction and the App Review quotation; and the statement that the Messenger Platform is free to use with no charge for in-window replies. developers.facebook.com
Meta, Messenger Platform changelog, page stamp "Updated: Jun 30, 2026", read 25 August 2026. Source of the 27 April 2026 error-code-100 date for the three retired tags (changelog entry dated 27 March 2026), the 19 July 2022 Business Verification enforcement, the Ice Breakers and Get Started precedence note, the 10 April 2026 Marketing Messages frequency caps, and the Instagram Handover Protocol migration. developers.facebook.com
Meta, Messenger Platform send-messages documentation, read 25 August 2026. Source of the three messaging_type values, the full list of window-opening actions, the inline warning repeating the 27 April 2026 tag retirement, and the HUMAN_AGENT seven-day window. Also the origin of the v25.0 sample-code observation. developers.facebook.com
Meta, Messenger Platform policy documentation, read 25 August 2026. Source of the 30-second responsiveness requirement and its seven-day remediation window, the HUMAN_AGENT corroboration, and the News Page Index restriction on NON_PROMOTIONAL_SUBSCRIPTION. This page also still describes Sponsored Messages as live and links One-Time Notification as "(Beta)", contradicting the Marketing Messages FAQ; the contradiction is disclosed in the body rather than resolved silently. developers.facebook.com
Meta, Marketing Messages on Messenger FAQ, page stamp "Updated: Mar 31, 2026", read 25 August 2026. Source of: the Recurring Notifications shutdown including the five-region exception and the internal date contradiction between 12 January and 10 February 2026; the 31 July 2024 Sponsored Messages deprecation; the paid-per-delivered-message model and its ads-billing mechanics; the absence of a public rate card and the Estimation API; the 20-market availability list and the EU, UK and Canada exclusions; and the per-user daily cap, the 12-hour cooldown and the frequency-cap ratios. developers.facebook.com
Meta, Messenger Platform rate-limiting documentation and Messenger Profile API reference, read 25 August 2026. Source of every figure in the rate-limit paragraph: the 200-times-engaged-users daily calculation, 300 and 10 calls per second on the Send API, 2 per second on the Conversations API, 750 per hour on Private Replies, 10 per 10 minutes on the Messenger Profile API, and the 40-messages-per-second Page Inbox cutoff. The Ice Breakers four-question cap and the Commands-menu override come from the Messenger Profile API and Discovery pages. developers.facebook.com
Meta, Messenger Platform Discovery documentation, read 25 August 2026, for the current entry-point list and for the m.me-plus-QR-code recommendation. Used as negative evidence for two claims: the Chat Plugin and Messenger Codes are absent from the list of supported website plugins and discovery mechanisms respectively, which is how their removal is established here. developers.facebook.com
Meta Business Help Center, "About the Chat Plugin" (/business/help/1661027437357021), observed on 25 August 2026 to 301-redirect to "About Business Messaging Tools", which does not enumerate the Chat Plugin. This redirect plus the Discovery omission is the whole evidentiary basis for the removal. The commonly cited 9 May 2024 shutdown date is vendor-sourced only; we could not find it on a Meta-hosted page and do not assert it. facebook.com/business/help
Meta Business Help Center, "Troubleshoot Your Messenger Block Rate", read 25 August 2026, for the block-rate guidance and the Insights API exposure. facebook.com/business/help
Meta for Developers, "Introducing Graph API v26.0 and Marketing API v26.0", published 29 July 2026, read 25 August 2026, for the current API version. developers.facebook.com/blog
Documentation-tree observation, stated as our own finding rather than as a Meta statement. On 25 August 2026 the legacy developers.facebook.com/docs/messenger-platform/ tree still resolved and still served a changelog ending in November 2025 naming v24.0 as current, while the /documentation/business-messaging/ tree carried entries through June 2026. That comparison is ours. It is the basis for the opening section and for the correction flagged against our own channel guide's citation path.
Chatbotscape review corpus, searched 25 August 2026. Editorial scores in step 1 are carried unchanged from best Messenger chatbot platforms, which in turn carries them from each platform's full review: SendPulse 86, Manychat 84, Tidio 75, BotPenguin 70, Chatfuel 69. The 20-character Messenger quick-reply truncation warning is recorded in our Tidio review from a hands-on session on 28 May 2026. The correction to /glossary/chatbot-broadcast was found by grep -rlniE "CONFIRMED_EVENT_UPDATE|POST_PURCHASE_UPDATE|ACCOUNT_UPDATE|recurring notification|sponsored message|chat plugin" --include=*.md academy glossary sample-reviews best, run case-insensitively per our own rule, which returned five files; each was read in full, three were found current and two were found defective, and both defects are documented in the body. An earlier pass of this audit cleared best/best-messenger-chatbot.md and was wrong to: that page asserts two figures this guide argues cannot be asserted from the available sources, which is a standard we have to apply to ourselves before we apply it to anyone else. The five legacy /docs/ paths in sample-reviews/channels/messenger-chatbot-guide.md were found by a separate search and are counted alongside them.
Messenger scale. Meta's own most recent public framing puts Messenger above a billion monthly users. Third-party trackers give figures between roughly 1.01 and 1.3 billion for 2026, a spread too wide to quote as a number, which is why this page uses Meta's framing rather than a precise figure. Secondary and not relied on for anything operational.
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 and is written for the owner or operations lead who has decided to build on Messenger and needs the current sequence rather than the 2019 one. It covers the build path only: prerequisites, entry points, the messaging window, what retired in 2026, cost, and the policy limits that restrict accounts. It contains no affiliate links; some linked reviews do, and our affiliate disclosure explains the arrangement. What we do with your data is in our privacy policy.

Methodology

Every Meta-side claim on this page was read from Meta's current documentation tree on 25 August 2026 and is attributed in Sources to the specific page. Where Meta's own pages disagree with each other, the disagreement is printed rather than resolved. Where a page would not render, that is stated and the affected claim is scoped down accordingly.

The editorial judgment on this page, listed here rather than flagged line by line:

Opening on the documentation migration rather than on prerequisites. That is a deliberate inversion of the standard tutorial order, taken because a reader following stale docs will get the rest wrong regardless of how good the steps are.
Scoping the page to the build path and deferring the channel case and the platform ranking to two existing pages. This makes the guide shorter and less self-sufficient than a single all-in-one tutorial, which we think is the right trade against duplicating our own coverage.
Publishing an incomplete message-tag table with the incompleteness marked, rather than filling the gap by inference from secondary sources. A reader is better served by a short verified list plus a warning than by a long list that might be wrong.
Recording the correction to our own glossary entry in the body rather than patching it quietly. Reviews and glossary entries are edited on a separate track, so a silent fix would have hidden a defect a reader following our internal link had already been exposed to.
Presenting timelines as ranges with the absence of a hands-on build stated plainly. The alternative was to publish specific figures we did not measure, which our integrity rules forbid.
Writing for a smaller search cluster than we normally would. We took it because the pages currently ranking for these terms predate the April 2026 changes, so correctness is worth more here than volume is.

We did not build a Messenger bot end to end for this guide. See our methodology for how platform facts are verified.

Last updated

26 August 2026.