How do you export Confluence data for an AI data license?

A Confluence wiki is the closest thing many companies have to a written memory: how things are done, why a decision was made, what went wrong last time. It exports in several formats that keep different parts of that, so the useful question is which format keeps the reasoning. This page covers the routes and what each leaves behind.

What's my data worth?


What a Confluence site usually holds

Team spacesPages grouped by team or product, each with an owner. A natural unit to approve and export one at a time.
Decision and design pagesThe write-ups that say what was chosen, what was rejected and why. Often the most useful pages in the wiki.
Meeting notes and retrospectivesWhat was discussed, what was agreed, and what went wrong. They hold names, so they need the most cleaning.
Page history and commentsThe earlier versions of a page and the comments under it, which show how an answer changed and who pushed back.
How-to guides and runbooksStep-by-step pages that staff follow. Valuable for showing a process, and easy to share once reviewed.

How Confluence content leaves the site, route by route

ToolWhat an export gives youWhat is missing or needs an admin
Export a spaceA whole space as web pages, a document file or a data file, run by a space administrator.Some formats drop page history and comments, and restricted pages may be left out depending on who runs it.
Export a single page or a page treeA chosen page with its child pages, useful for a first small sample.It does not give you the wiki as a whole, and each page needs checking for the same details as the larger export.
The programming interface or an admin backupA full pull of content, versions, comments and attachments, run by a developer or a site administrator.The raw result is not meant for reading, and it needs someone to run it and turn it into something reviewable.

Tool names belong to their owners and are used only to say where records usually live. Export features change by plan and version, so check the current rules for your account.


Page restrictions are a rights question, not a setting

Some pages are visible only to a group, such as management or human resources. An export run by an administrator may include them, which is a reason to check restrictions before running anything. A restricted page tells you the company already decided who should read it.


Old pages are not free of problems

A wiki collects abandoned pages, drafts and copies. Out-of-date instructions that were never removed can mislead a reader, and a page copied from a customer’s document may carry that customer’s name. A short review pass for staleness and for copied material saves trouble later.


Pick the format that keeps what you need

If the page history and comments matter, check which format keeps them before you commit. A flat document export looks tidy and loses the trail of how a page changed. A data export keeps more and needs more cleanup. Run a test on one space first and open the result.


What stays out of a Confluence set

Restricted and HR pagesPages limited to management, finance or people teams stay out unless that group approves.
Customer-specific pagesPages written for one client, or copied from a client’s documents, belong in the client review, not the general set.
Security and access detailsPages with network diagrams, credentials or incident details that could help an attacker are removed.
Drafts and abandoned pagesUnfinished or contradicted pages are filtered, since they teach the wrong thing.

An example, invented for illustration

A payments software firm with 120 staff has a Confluence site with a dozen team spaces. The owner approves the engineering, support and product spaces, excludes human resources and the security space, and asks a space administrator to export with page history and comments kept. Each page is checked for copied customer text and staff names, the runbooks are separated from the meeting notes, and a sample of forty pages is shown for approval.


Questions owners ask about this

Which export format keeps page history?

It depends on the product version and your plan, and the answer has changed over time. Test a small space and open the result before choosing one for the whole site.

Are comments part of the page?

Often they are separate. Whether they travel with the page depends on the route, and they are worth keeping because they show where the team disagreed.

Do we need to ask the people who wrote the pages?

Pages were written as work, so the company usually holds them, but your terms and local rules apply. Staff are told what is happening, and counsel confirms the basis.

Is a wiki worth less if it is messy?

Messy is fine. What lowers its value is contradiction with no way to tell which page is current, so mark outdated pages instead of deleting the trail.

Didn't find your question? Ask us directly.


Keep reading

Related kinds of records

  • Project and task history: Jira, Asana, Linear, Monday, Trello, ClickUp and Airtable comments and status changes, not task titles.
  • SOPs and knowledge bases: Procedures with their edit history and comments; published help articles add nothing.
  • Jira data export: How to get issues with their comments and links out of Jira, and the secrets to remove first.

Industries that keep these records

  • Fintech companies: Customer money stays out; engineering decisions, post-mortems and procedures can go in.
  • Pharma companies: Compounds and patients stay out; procedures, deviations and training records can go in.