International SMM

Facebook

Tools for the platformActivepiecesAdEspressoAdsPowerAgorapulseAlbatoApiX-DriveAspireAuphonicfull list ↓
RegionUSA and Canada

Platform · this deep dive is built from the official pages of the platform itself and of the tools that support it.

Overview

Facebook for audience work is first of all a Page, not a profile. Programmatic access is built so that you can publish only to a Page; publishing to a personal profile through the Pages API is not provided at all. That is the first thing to know when choosing: if you have a personal profile, none of the mechanism described below applies to it.

The site facebook.com itself is closed to robots: robots.txt forbids crawling by Claude agents.

Getting started

Working through programmatic access requires a Page access token, and it is obtained in four steps: request permissions → user consent → exchange the code for a user token → request the Page token. A single key, as on some platforms, will not do here.

Permissions are granular and requested separately: pages_show_list, pages_read_engagement, pages_manage_posts (publishing and scheduling), pages_manage_engagement (moderation), pages_manage_metadata, pages_manage_ads, pages_manage_cta. A service asking you for all seven at once asks for more than auto-posting needs: the first three suffice for that.

Publishing is possible only to a Page — published or unpublished — where you hold the CREATE_CONTENT task.

What you can publish

On AI-made content Facebook writes in detail — and the rule is shared across Facebook, Instagram and Threads.

Labelling is mandatory for photorealistic video and realistic audio. Verbatim: «We require people to disclose, using our AI-disclosure tool, whenever they post organic content with photorealistic video or realistic-sounding audio that was digitally created or altered, and we may apply penalties if they fail to do so».

There is a tool for it. The label is called «AI info» (formerly «Made with AI»), and the author uses a separate disclosure feature.

The platform also detects on its own — through industry metadata, not by eye: «we're building industry-leading tools that can identify invisible markers at scale – specifically, the "AI generated" information in the C2PA and IPTC technical standards». Both standards are named outright.

There is no direct prohibition on AI content — there is labelling and a prohibition on harm. The penalty for failing to label is described in one formulation: «…and we may apply penalties if they fail to do so». What exactly the penalty is — removal, reduced reach, loss of monetisation — the rule does not disclose: it contains no steps, timelines or thresholds.

How to grow

Facebook defines its metrics precisely, and the definitions are worth reading before comparing numbers with another platform.

Video views (page_video_views): «The number of times your Page's videos played for at least 3 seconds, or for nearly their total length if they're shorter than 3 seconds» — a three-second threshold, or nearly the whole duration for shorter clips.

Reach (page_impressions_unique): «The number of people who had any content from your Page or about your Page enter their screen» — unique people, and the metric is marked by the platform itself as estimated. The difference from impressions: impressions count every arrival on screen, reach counts people once.

Engagement (page_post_engagements): «The number of times people have engaged with your posts through reactions, comments, shares and more» — reactions, comments, shares and more, counted as occurrences, not unique people. What «and more» comprises is not spelled out, and the platform gives no engagement-rate formula.

Reach has a shelf life: the metric is marked deprecated above Graph API v25 and falls under the metric shutdown of 15 June 2026. The definition may change along with its replacement.

The platform wrote separately about showing unoriginal content: «Content that is duplicative or involves minor edits to another creator's post» is considered unoriginal and demoted; «Creators who are not adding substantial creative value will see their content deprioritized». The word «AI» does not appear in that post — that is, demotion is not for machine origin but for the absence of your own contribution.

The road to monetisation

Facebook does have payout programmes. Bearing indirectly on earnings is also the originality rule above: reduced reach hits income before any threshold does.

Tools and automation

The Pages API allows publishing and scheduling content, reading statistics, moderating comments and receiving events through webhooks. That is the official route, and its terms are stated in numbers — a rare case in this guide.

Request frequency is set by formula. The platform-level application limit: «Calls within one hour = 200 * Number of Users», where the number of users is «the number of unique daily active users an app has». The platform's own worked example: «if your app has 100 Users, your app can make 20,000 calls per hour».

What matters here is whose limit it is: the application's, not yours. An auto-posting service with a hundred clients divides twenty thousand calls an hour among them all — and your turn depends on its load, not on your plan.

There are two kinds of limit: platform limits (for app and user tokens) and Business Use Case limits (for system and Page tokens). Where they overlap, BUC applies.

Spend is visible. The X-App-Usage header carries call_count, total_cputime and total_time — percentages of the limit. Errors are named by code: «4: Indicates that the app whose token is being used in the request has reached its rate limit», «17» — the same for the user.

A consequence for scheduling worth knowing: retrying after a refusal is harmful — the spend keeps rising and the wait grows longer.

Limits and rules

For humans Facebook does not state limits, and says so outright. Verbatim: «We have limits in place to prevent the abuse of our features and to protect people from spam and harassment. These limits are based on different factors, like speed and quantity. We can't provide additional details on the rate limits that are enforced».

That is the most direct refusal among the platforms in this guide: the structure of the limit is explained (speed and quantity), the number is withheld deliberately.

The hourly window applies only to programmatic access; to human actions it does not — there the talk is of «speed and quantity» without numbers or windows.

Limits on friend requests, group additions and messages exist but are described by examples, not numbers: «If someone is sending out an excessive number of messages to people they aren't friends with, they may be warned or temporarily blocked from sending messages»; «If someone adds several people to a group within a short period of time…». On messages separately: «If you were warned for sending messages too quickly, you'll need to stop this activity temporarily to avoid being blocked».

Facebook publishes no separate restrictions for new accounts. In place of account age it names behavioural history: «The length of a block depends on the severity of the violation and your history on Facebook».

Who it suits

Facebook suits someone who works through a Page and is ready to automate via the official interface: there the terms are stated as formulas rather than hints, and the spend is visible in the response header.

It suits less well anyone who needs predictability in manual work: the platform deliberately does not publish limits for humans, and the only way to find the boundary is to hit it.

Anyone choosing a scheduling service should remember: the hourly limit belongs to the application, not to you, and is shared among all its clients.

Tools for this platform

The list is built from the tools' own facts: every line is a tool's statement that it supports the platform, with the source named. A tool missing from the list only means we have no such statement.

Scheduling and publishing:

Statistics and analytics:

Customer communication:

Automation:

Social listening:

Other tasks:

Verified data

The checked data this deep dive rests on.

Content

Required for photorealistic video and realistic-sounding audio. Word for word: "We require people to disclose, using our AI-disclosure tool, whenever they post organic content with photorealistic video or realistic-sounding audio that was digitally created or altered, and we may apply penalties if they fail to do so." There is no such requirement for images: the platform promises to label them itself

The rule is common to Facebook, Instagram and Threads — Meta keeps it in the "Misinformation" section of the "Community Standards". The platform's own help article ("Label AI content on Instagram", help.instagram.com/761121959519495) is unavailable to an ordinary request: the domain returns HTTP 400 without a browser, and the robots.txt of help.instagram.com contains "User-agent: ClaudeBot / Disallow: /"

source, checked 2026-08-04

The label is called "AI info" (previously "Made with AI"): "Meta began adding labels to a wider range of video, audio and image content" (since May 2024), "These are displayed as “AI Info”". The creator uses a separate tool: "we’re adding a feature for people to disclose when they share AI-generated video or audio so we can add a label to it". The platform shows the origin of the label: "shares whether the content is labeled because of industry-shared signals or because someone self-disclosed"

The wording about the disclosure tool is from the Meta blog https://about.fb.com/news/2024/02/labeling-ai-generated-images-on-facebook-instagram-and-threads/. There is no step-by-step instruction (where exactly the toggle is in the posting form) in the available sources: Meta's help pages are closed, see the note to ai_disclosure_required

source, checked 2026-08-04

Yes, through industry metadata. Word for word: "we’re building industry-leading tools that can identify invisible markers at scale – specifically, the “AI generated” information in the C2PA and IPTC technical standards"; "in the coming months we will label images that users post to Facebook, Instagram and Threads when we can detect industry standard indicators that they are AI-generated". Also: "Our “Made with AI” labels on AI-generated video, audio and images will be based on our detection of industry-shared signals of AI images or people self-disclosing that they’re uploading AI-generated content"

Both standards are named directly — C2PA and IPTC. The second quotation is from https://about.fb.com/news/2024/04/metas-approach-to-labeling-ai-generated-content-and-manipulated-media/. Meta admits the count is incomplete: "labeling methodology is still evolving" and the data "may not capture labels on some content that is edited with AI"

source, checked 2026-08-04

There is no direct ban on AI content — there is a label and a ban on harm. Word for word: "We may place an informative label on the face of content – or reject content submitted as an advertisement – when the content is a photorealistic image or video, or realistic sounding audio, that was digitally created or altered and creates a particularly high risk of materially deceiving the public on a matter of public importance." Removed regardless of the way it was produced: "We remove misinformation or unverifiable rumors that expert partners have determined are likely to directly contribute to a risk of imminent violence or physical harm to people"; "We consult with leading health organizations to identify health misinformation likely to directly contribute to imminent harm to public health and safety"; on elections and the census — "We remove misinformation that is likely to directly contribute to a risk of interference with people’s ability to participate in those processes"

Meta has no list of "you may not make voices of real people / faces / elections / medical claims" separately for AI: the prohibitions are worded by harm, not by the way of production. Hence — partial

source, checked 2026-08-04

"…and we may apply penalties if they fail to do so" — the only wording about punishment for a missing label. What exactly the punishment is (removal, limited distribution, loss of monetization) the rule does not disclose

No terms, steps or thresholds are named in the rule — only "may apply penalties". No more detailed page about the consequences of a missing label was found on the available Meta resources

source, checked 2026-08-04

AI is not addressed directly; the rule is worded about unoriginal content: "Content that is duplicative or involves minor edits to another creator’s post" is considered "unoriginal and deprioritized"; "Creators who are not adding substantial creative value will see their content deprioritized, and they may see their account deemed non-recommendable and demonetized, if they continue to post primarily unoriginal content"

The word "AI" is not in the text of this Meta blog entry about the originality of content — that is, the platform does not punish AI as such, it punishes unoriginality. The "Content Monetization Policies" themselves could not be checked: they are at www.facebook.com/business/help, a domain closed in robots.txt ("User-agent: * / Disallow: /")

source, checked 2026-08-04

Limits and restrictions

Named by formulas. The platform limit of an app: "Calls within one hour = 200 * Number of Users", where "The Number of Users is based on the number of unique daily active users an app has"; the explanation: "if your app has 100 Users, your app can make 20,000 calls per hour". For a user key the numbers are hidden: "Due to privacy concerns, we do not reveal actual call count values for users". The business limit for pages: "Calls within 24 hours = 4800 * Number of Engaged Users", where "The Number of Engaged Users is the number of Users who engaged with the Page per 24 hours". Messenger: "Calls within 24 hours = 200 * Number of Engaged Users"; "Conversations API: Your app can make 2 calls per second per Facebook Page"

Meta has two different mechanisms: Platform Rate Limits (an hour, counted from the app's active users) and Business Use Case (24 hours, counted from the page's engaged users). The rule for resolving a conflict is named directly: "If both Platform and Business Use Case rate limits can be applied to a request, BUC rate limits will be applied"

source, checked 2026-08-04

For programmatic access they are described in detail: the expenditure is shown by the header "X-App-Usage" with the fields call_count, total_cputime and total_time (percentages of the limit); the errors — "4: Indicates that the app whose token is being used in the request has reached its rate limit", "17: Indicates that the User whose token is being used in the request has reached their rate limit", "32: Indicates that the User or app whose token is being used in the Pages API request has reached its rate limit", "613: Indicates that a custom rate limit has been reached". The rule of behaviour: "When the limit has been reached, stop making API calls. Continuing to make calls will continue to increase your call count, which will increase the time before calls will be successful again." For a human the punishment is described in words: a warning, then a temporary block on the feature

An important consequence for planning: continuing to make attempts after a refusal is harmful — the expenditure keeps growing and the waiting time lengthens. This is a statement by the platform itself, not our conclusion

source, checked 2026-08-04

It is not named by a number, and the platform directly refuses to name it. Word for word: "We have limits in place to prevent the abuse of our features and to protect people from spam and harassment. These limits are based on different factors, like speed and quantity. We can't provide additional details on the rate limits that are enforced"

This is the most direct refusal of the thirteen platforms: Facebook explains the arrangement of the limit (speed and quantity) and immediately reports that it will not give details. The page was read in the lightweight version of the help centre without scripts; the robots.txt of Facebook does name the ClaudeBot agent, but forbids it only individual sections, and the help centre is not among them

source, checked 2026-08-04

They are not named by a number. The limits on friend invitations and on adding to groups are described only by examples: "If someone is sending out an excessive number of messages to people they aren't friends with, they may be warned or temporarily blocked from sending messages"; "If someone adds several people to a group within a short period of time, that person may be warned or temporarily blocked from adding more people". The advice from the article about blocks: "Only send friend requests to people you know"

Facebook describes the behaviour for which it punishes, but does not give numbers: "We can't provide additional details on the rate limits that are enforced". The separate article "Blocked from sending friend requests on Facebook" (help/320587401313728) did not open in the time allotted, its content is not transferred here

source, checked 2026-08-04

They exist, they are not named by a number. Word for word: "Facebook has limits in place to prevent behavior that others may find annoying or abusive. These limits restrict the rate at which you can use certain features on the site. If you were warned for sending messages too quickly, you'll need to stop this activity temporarily to avoid being blocked from sending messages." And a prohibition in substance: "Unsolicited messages of any kind aren't allowed"

The help centre article "Limits to sending messages on Facebook". Neither the number of messages nor the duration of the warning is given; for programmatic access a separate count applies — 200 calls per 24 hours per engaged user in the Messenger API

source, checked 2026-08-04

API access

Capabilities. the Pages API lets you publish and schedule content, read statistics, moderate comments and receive events through webhooks

source, checked 2026-07-28

How it connects. a PAGE access token is required, obtained in four steps: request permissions → user consent → exchange the code for a user token → request the page token

source, checked 2026-07-28

the permissions are separate and requested individually: pages_show_list, pages_read_engagement, pages_manage_posts (publishing and scheduling), pages_manage_engagement (moderation), pages_manage_metadata, pages_manage_ads, pages_manage_cta

source, checked 2026-07-28

you can publish ONLY to a page (published or unpublished) that carries the CREATE_CONTENT task; publishing to a personal profile through the Pages API is not provided for

source, checked 2026-07-28

Rate limit. the app limit is a rolling one-hour window given by the formula «Calls within one hour = 200 * Number of Users», where Number of Users is the number of unique daily active users of the app. This is a limit PER APP, not per user

source, checked 2026-07-28

there are two kinds of limits: platform limits (for app and user tokens) and business use case limits, BUC (for system tokens and page tokens); where they overlap, BUC applies

source, checked 2026-07-28

metrics

A page video view (page_video_views) verbatim: «The number of times your Page's videos played for at least 3 seconds, or for nearly their total length if they're shorter than 3 seconds» — a threshold of 3 seconds, or nearly the whole length for shorter clips. A content impression (page_impressions) verbatim: «The number of times any content from your Page or about your Page entered a person's screen» — with no time threshold

www.facebook.com/help: ClaudeBot has its own section in robots.txt (help is not forbidden), but the help articles are served behind a sign-in — the Graph API was used instead. Note: Meta has two definitions of a «view» side by side — 3 seconds for video against an on-screen impression with no threshold; most Page Insights metrics will be switched off by 15.06.2026

source, checked 2026-08-07

Page reach (page_impressions_unique) verbatim: «The number of people who had any content from your Page or about your Page enter their screen» — unique people, with the metric marked as estimated; the difference from impressions is that impressions count every appearance on screen while reach counts people once

The metric is marked deprecated above Graph API v25 and falls under the metrics shutdown by 15.06.2026 — the definition may disappear from the documentation

source, checked 2026-08-07

page_post_engagements verbatim: «The number of times people have engaged with your posts through reactions, comments, shares and more» — reactions, comments, shares and more, counted as occurrences (not unique people); the platform does not define an engagement rate

What «and more» consists of is not spelled out; there is no ER formula

source, checked 2026-08-07

Catalogue section: all similar See also: catalogue index · region: USA and Canada · scheduler comparison · find by situation · platform restrictions