A post that fails to publish is worse than a post that publishes badly, because nobody notices it for hours. The calendar says scheduled, the morning passes, and the announcement that was supposed to land with the shop opening is still sitting there marked failed.
Almost all of these failures are predictable, and most are preventable. They come from a short list of causes, and once you know the list you can tell within seconds whether a failure is about this post, this channel or this network. This guide works through the list, from the most common to the most obscure, with what to do about each.
The five families of failure
Nearly everything belongs to one of five families.
- The connection is broken. The tool no longer has valid permission to post as you.
- The media could not be fetched. The network went looking for your image or video and could not get it.
- The post breaks a rule. Too long, no image, images and video mixed, a video too long, a field missing.
- A cap was hit. The network limits how much an app may publish for you in a window.
- The app itself is not approved yet. A review gate on the tool’s own developer app, not on your account.
Knowing which family a failure belongs to tells you how urgent it is. A rule break affects one post. A broken connection affects every post on that channel until someone fixes it.
Family one: the connection expired or was withdrawn
This is the most common cause by a wide margin, and the most misunderstood, because people read a reconnect prompt as a fault in the tool.
Networks do not hand out permanent access. They issue a token with a lifetime, and they withdraw it the moment anything looks like it might have changed.
| Network | How long access lasts | What ends it early |
|---|---|---|
| Facebook and Instagram | Page access does not expire on its own | A password change, removing the app, or losing admin rights on the Page |
| Around 60 days for a personal connection | It simply runs out, and most apps cannot renew it without you | |
| Threads | Around 60 days, refreshed before it runs out | Missing the refresh window, or removing the app |
| Around 30 days, refreshed continuously | Revoking access in Pinterest’s settings | |
| TikTok | A day at a time, renewed for about a year | A year of no renewals, or revoking the app |
| X | Short-lived, renewed constantly | Revoking the app in your X settings |
| YouTube and Google Business | Renewed in the background | Revoking access in your Google account’s security settings |
| Bluesky | As long as the app password lives | Deleting the app password, or changing the account password |
What to do. Treat a reconnect prompt as a five-minute job to do today, not a notification to dismiss. In Picifer a channel that loses access is marked for reconnection, the workspace is notified, and the daily digest lists broken channels, precisely because the cost of not noticing is a week of silence on that channel.
What to avoid. Do not connect a network with a personal account belonging to someone who might leave. Connect Pages and business accounts, with more than one admin, so a departure does not take the channel with it.
Family two: the media could not be fetched
This one looks impossible when you first meet it. The image is right there in the editor. The preview shows it. And the post still fails.
The reason is that several networks do not accept an upload from the tool at all. They take a link to your media and download it themselves. Instagram, Threads, Facebook video, TikTok, Pinterest images and Google Business all work this way. If the file is not reachable from the public internet at the moment the network goes looking, the post fails.
Three practical consequences:
- Media hosted somewhere private will not work. A file behind a login, on an internal server or on a local machine is invisible to the network.
- Format matters more than usual. Instagram only accepts JPEG images by link, so a PNG that displays perfectly in a browser can still be refused.
- The host has to allow the network’s crawler. A host that blocks unknown user agents blocks the network, and the failure looks like a broken file rather than a blocked request.
What to do. Keep your media in the tool’s own library rather than linking to files elsewhere, so the hosting is the tool’s problem and not yours. Convert images to JPEG when Instagram is a destination. And test one post per network when you first connect a channel, rather than discovering this on a campaign.
Family three: the post breaks a rule
This is the family most people assume is the whole list, and it is the easiest to eliminate entirely, because every one of these is knowable before the post is scheduled.
The frequent ones:
- Over the character limit. Especially on X, where a link counts as 23 characters however short it is, and on Telegram, where attaching a photo drops the caption limit from 4,096 characters to 1,024.
- No media on a network that requires it. Instagram, TikTok and Pinterest all refuse a post without an image or video, and YouTube needs a video specifically.
- Images and video mixed. Instagram, Threads, Telegram, Discord and Tumblr allow it. Facebook, X, LinkedIn, Bluesky and Mastodon do not.
- A video outside the allowed length. An Instagram Reel needs at least 3 seconds. A Facebook Story is capped at 60, even though a Reel on the same Page allows 90.
- A poll with media attached. Rejected on every network that offers polls.
- A required field left empty. A Pinterest pin with no board. A YouTube video with no title. A LinkedIn document post with no document title. A Google Business event missing its end time.
- A format the account cannot use. A long post on X needs a paid subscription on the connected profile, and the post fails rather than being trimmed to 280 characters.
Family four: a cap was hit
Networks limit how much an app may publish on your behalf in a window. Ordinary posting never comes close. Bulk imports, migrations and enthusiastic first weeks do.
- Threads allows 250 published posts per profile per rolling 24 hours, and a carousel counts as one. That is also the real ceiling on long reply chains, since every part of a chain is its own post.
- Instagram documents a cap in the same range for posts published through its API, with a carousel counting as one.
- Facebook caps Reels published through the API at 30 per Page per 24 hours.
- YouTube works on a daily quota rather than a post count, and a single upload consumes a large share of a project’s default allowance. In practice that is a handful of uploads a day, not dozens.
- Tumblr limits calls per hour and per day rather than posts.
What to do. Spread a backlog over days rather than pushing it all at once, which is better for the audience anyway. If you are moving from another tool, import in batches and let the queue place them. Our guide to switching to Picifer covers what does and does not come across in an import.
There is a second kind of cap worth knowing: the reply window on messaging. Networks that allow private messages generally only let a business reply within a fixed period after the customer’s last message. That is not a publishing failure, but it is the same shape of surprise, and our guide to social media response time covers what it means for a team’s routine.
Family five: the app itself is not approved
This one is invisible from the inside, which makes it the most confusing of the five. Everything about your account is fine. The gate is on the tool’s own developer app, and it is a review by the network.
- TikTok keeps an unaudited app’s posts restricted until it passes a content posting audit, so posts can go out visible only to the account that made them.
- YouTube locks uploads from an unaudited project to private.
- Threads requires a reviewed permission for publishing, without which an app can only post for accounts it has registered as testers.
- Pinterest starts new apps in a trial mode where pins are created in a sandbox rather than on the real account.
- Google Business Profile starts with no quota at all until an access request is approved.
What to do. Nothing, as a customer, except ask before you depend on it. When you connect a network that matters to a campaign, publish one real test post and check it appears publicly on the network itself rather than trusting the tool’s status. That single check catches this whole family.
Partial failures, and why retrying is not always right
Not every failure is clean. Two cases need judgment rather than a retry.
A thread that fails partway. A thread is published part by part, each one replying to the one before. If part four is rejected, parts one to three are already live. Retrying the whole thread would publish those three a second time, which is worse than an incomplete thread. The right behavior is to stop, report how many parts went out, link the first one, and let a person finish it by hand.
A post the network says it accepted but cannot confirm. Some networks process video asynchronously and hand back a job reference rather than a post. A tool has to decide how long to wait before it treats the post as published. If it waits forever, the calendar lies; if it gives up too soon, the post is published twice.
Both are reasons to prefer a tool that does its own scheduling rather than handing your post to a network’s built-in scheduler. When the network holds the post, the tool cannot retry it, cannot cancel it, cannot report on it and often cannot even confirm it went out. Picifer schedules with its own worker for exactly this reason: one place owns the retry, the status and the record of what actually happened.
Prevention, in order of value
Validate in the composer, every time
Write with the network and post type chosen, so the character count, the media rules and the required fields are the right ones. The free character counter and post preview tools do the same job for a post you are writing by hand.
Fix reconnect prompts the day you see them
A broken channel is not one failed post, it is every post on that channel until someone acts. Put it ahead of writing anything new.
Keep media in one library
Media the tool hosts is media the networks can fetch. Linking to files elsewhere adds a dependency you cannot see.
Test each channel once, publicly
After connecting a network, publish one real post and look at it on the network itself. This is the only check that catches app-level review gates.
Schedule important posts with a margin
Put a launch or an event announcement in the queue a day early as a draft, not an hour early as the only copy. If it fails, you want hours to notice, not minutes.
Read the failure list weekly
Add it to the same weekly review where you plan the next week’s posts. Patterns show up there that single failures do not: one channel failing repeatedly, one format always rejected, one time of day that never works.
What a good failure looks like
You cannot eliminate failures, because the final decision belongs to a network you do not control. What you can insist on is that a failure is loud, specific and honest.
Loud means you hear about it quickly, by email or a notification, rather than finding it in a list. Specific means the message says which rule was broken and on which channel, in words you can act on. Honest means a post that will never succeed stops being retried and is marked failed, rather than sitting in a queue looking hopeful.
Those three properties are what separates a scheduling tool you can plan around from one you have to check. If you are building the routine that surrounds it, our guide to a social media content calendar covers the weekly session where failures get reviewed, and a social media approval workflow covers the review step that catches the mistakes a validator cannot.



