Skip to main content
Publishing sends a finished article through the project’s configured destination. Check the selected project, review the draft and confirm the intended publishing setup before sending it. Afterward, inspect the public page or receiving build: a saved draft, approval or accepted request records a different step from the website displaying the article correctly. Open Publishing in the Content section of the sidebar. The page contains publishing presets and destination controls. The Draft & Approval Workflow explains the review decision before delivery.

Publishing modes

The current preset names are Manual, Draft for me and Full autopilot. A preset combines scheduled generation with the publishing mode. Check the saved setup and any existing queued posts when changing it; selecting Manual does not cancel every operation already underway.

No auto publish

Manual turns scheduled auto-generation off and selects manual publishing. Use the dashboard to decide which post to generate and when to send it. Publishing still needs an appropriate destination or supported API/build setup.

Requires approval

Draft for me enables scheduled generation and approval-required publishing. Review the generated version before choosing Approve or Reject. Approval moves a draft toward publication; check its status and available publishing action afterward.

Auto publish

Full autopilot enables scheduled generation and automatic publishing without that separate approval action. The automated presets require an active destination in the settings screen. Access checks, configured holds and delivery failures can still stop a post. Preset values save through separate requests, so a change can partially apply after an error. Reload and check the saved state before relying on a new workflow. Use Publishing Modes & Autopilot for preset controls and their limits.

Supported platforms

The destination screen offers WordPress.com, self-hosted WordPress, Ghost, Webflow, Zepio, custom webhooks and Shopify, which is labeled Beta. Choose the setup that matches the actual website and account you intend to use. A listed platform does not establish that your credentials or content fields are ready. Use Connect Platforms for each connector’s setup. Static sites can also use the Public Blog API with a deploy hook; the receiving site must fetch the published content and complete its build.

One active destination at a time

Configure the connection intended to receive posts and inspect which connection is active. The connection flows attempt to deactivate other connections when creating or activating one, but those changes are separate database operations. After a save error, verify the list instead of assuming every connection change was rolled back. The scheduled publisher looks for an active destination; it can fail when the stored destination is unavailable or inactive. Inspect the post’s message and configured connection before enabling automation. Keep API/build publishing distinct from a saved CMS connection.

Who can change connections

Connection changes require project editor access in the connection API. Publishing presets require project admin access, and the current sidebar exposes Publishing to admins. If the controls are absent or a request is denied, ask the project admin to check access rather than assuming every workspace member can change the destination.

Check the connection before relying on it

Some connection flows validate credentials before creating the connection, including self-hosted WordPress, Ghost and Zepio. Checks differ by connector and action; do not apply a universal pre-save test or no-partial-save promise to every update. Use Test Connection where available and read its result. A previous Passed test is history, not confirmation that the next publish will succeed. Inspect saved values after an error and keep credentials out of support screenshots.

How content is prepared before it is sent

Publishing preparation can remove duplicate title or featured-image markup and resolve internal references. The exact payload and supported behavior depend on the connector and publishing path. Review the resulting website as well as the editor.

Check the content sent to the destination

ContentRelevant handlingWhat to inspect
Title and featured imagePreparation can separate the title and featured image from body markup.The rendered heading, hero image and body images.
ExcerptWebhook payloads distinguish the full excerpt from a shortened excerpt field.Which field your receiver maps to its CMS summary.
Internal and relative linksSaved destinations resolve supported references; a valid website address supports root-relative URL expansion.Every important destination, anchor and any unresolved reference.
A resolved link is not a page-availability check. A valid website address is also needed to expand a root-relative path such as /contact. Follow Internal Linking to review saved destinations and unresolved references before publishing.

When a publish fails

Identify the action that failed and inspect the receiving website before retrying. A receiver can create or update an article even if the response never reaches theStacc. Read the post’s current status and message rather than treating every error as proof that nothing happened.

Temporary problems are retried

The scheduled worker retries failures classified as temporary. Its fast tier includes the first five attempts, with increasing waits and random jitter. If transient failures continue, a slow tier schedules further attempts after configured waits of 1 hour, 3 hours and 12 hours. Once those delays are exhausted, another transient failure is marked failed. These are eligibility delays; scanning and queue timing can make requests run later. A manually triggered publish or sync can instead return an error directly and does not always enter that scheduled retry sequence. Some connectors check for an earlier remote post before repeating a request. The matching rules differ, so inspect an uncertain result before retrying again. These checks reduce duplicate risk without guaranteeing duplicate-free delivery.

Problems you need to fix are not retried

The scheduled worker stops on failures classified as permanent, such as rejected credentials or invalid configuration. Correct the reported cause before requesting another attempt. A detail page in Publish Failed can offer Retry Publish; other states offer their own actions. Use Publishing Errors & Retries for platform-specific recovery, manual-versus-scheduled differences and uncertain remote outcomes. An error message does not establish every cause or describe internal incident handling.

Custom webhook events

A custom receiver handles the event and writes the article into the website or CMS. It must implement the appropriate create, update and removal behavior; accepting a request alone does not complete those jobs.

Webhook event responsibilities

EventReceiver responsibility
test.pingHandle a connection test without requiring article fields.
blog.publishedCreate or reconcile the intended article. Sample Payload can also send this event.
blog.updatedUpdate the existing article using stable identifiers.
blog.unpublishedApply the website’s supported unpublish behavior.
blog.deletedApply the intended deletion behavior without recreating the article.
Follow the Webhook Reference for payload, authentication and response handling. Return the public URL and stable CMS identifier when the receiving write supports them, and inspect the recorded result. A URL can also come from a configured template; it does not independently prove the page exists. Manual and scheduled paths differ in accepted responses and retry handling. For an asynchronous receiver or build process, check completion and the actual public page. Implement duplicate handling for repeated delivery instead of assuming a repeated request will create only one record.

Managing connections

Inspect the active connection, saved destination and previous test result before sending a post. Use the available test or activation control deliberately, then check the resulting list and status. Re-test after relevant credential or destination changes rather than relying on an older successful test. After publishing or syncing, use View Live when a URL is available and inspect the article. For API/build publishing, inspect the completed deployment as well. If delivery remains uncertain, use the error guide and share the blog, action and sanitized message with support.

Blog authors for custom sites

Blog author name, Editorial team or Person selection, and Save author

Existing custom-site author-setting example; confirm the controls available for your project.

For a custom webhook site, agree with your developer how the website will store and display the byline. Confirm the author controls and outgoing field available in your project before relying on the example above. After changing author details, check the saved value and the public byline separately. The Author field and receiver checklist describes an optional author field and receiving-site responsibilities. Confirm that your receiver supports the actual payload and updates its visible byline and structured data. Do not use an author name to imply that someone wrote or reviewed content without their participation.

Review the sources used

Sources used popup showing research links, dates, and highlights

Existing sources-review example; availability depends on the post and controls offered in your project.

Where Sources used is offered, inspect the recorded links and relevant passages before approving the draft. If the control or records are unavailable, use the sources you supplied and verify important claims directly. A listed source does not establish that it supports every statement in the article. Check business facts, quotations and offers against appropriate evidence, and arrange any required specialist review. Return to Draft & Approval Workflow when the reviewed version is ready for a decision.