What is Confluence: a fair look at Atlassian's wiki

TL;DR. Confluence is Atlassian’s wiki — a hosted knowledge base of pages, spaces, and Jira links that has been the default corporate wiki answer since 2004. It’s the most-deployed tool in the category, the slowest of the major options on a typical workspace, and the one most teams either love because everyone already uses it or quietly leave because page loads have stopped being something they want to argue with. This post walks through what it is, what it does well, where it falls short, and how to tell which of those camps you’re in.

Most posts about Confluence are written either by Atlassian or by the company that wants to sell you the migration off Confluence. This one is written by the latter, with a real effort at the former’s fairness, because the truth about Confluence is more interesting than either side normally lets on. Confluence is the incumbent. Incumbents are usually right about more than they get credit for.

We’ll cover what Confluence actually is, how it works, what it does well, where it struggles, the Jira-and-Confluence pairing that earns it most of its budget, and the test your team can run before deciding whether to stay or migrate.

What Confluence actually is

Confluence is a wiki. The full sentence is Confluence is a hosted, multi-user wiki and team-knowledge-base product made by Atlassian, sold as part of the same suite as Jira, with deep integration into both Jira and the rest of the Atlassian catalogue, and a large macro / app ecosystem on top. The short sentence does most of the work; the long sentence explains why the short one undersells it.

Atlassian shipped Confluence in 2004, when the team wiki category was about a decade old (Ward Cunningham’s original wiki ran from 1995). Twenty-two years later, Confluence is the corporate-wiki product most companies have deployed somewhere, even when it isn’t the one their teams talk about. That gap — deployed everywhere, talked about reluctantly — is the whole topic.

For the broader category that Confluence sits inside, see our corporation wiki explainer and the knowledge base explainer. Confluence is one shape of the category; the category is bigger than Confluence.

How Confluence works

The mental model is small: pages live in spaces, spaces live in the workspace, the workspace lives in your Atlassian cloud tenancy. Almost every Confluence question collapses to one of those four levels.

Pages. The unit of content. A Confluence page is a rich document with a tree of nested pages underneath. The editor ships with the usual things — text, headings, tables, images, attachments, callouts — plus macros, which are embeddable widgets that render dynamic content (a Jira issue list, a table of contents, a status indicator, a chart). The macro story is load-bearing; most of the deep customisation Confluence supports runs through it.

Spaces. The unit of organisation. A space is a container of related pages — Engineering, Product, HR. Spaces have their own permissions, their own homepage, their own templates. Most teams have one space per business function, which scales fine until the org chart re-orgs faster than the wiki does.

Templates and blueprints. Confluence ships a long catalogue of page templates — meeting notes, decision record, project plan, retrospective — and a smaller catalogue of blueprints, which are page templates that bring their own indexing macros so a team can list every retrospective in a space without writing glue. (The shape is good; the implementation has been mostly unchanged for a decade.)

The Atlassian glue. This is the part most explainers under-cover. Confluence isn’t really sold as a standalone wiki anymore; it’s sold as the documentation surface for the Atlassian suite. Jira tickets link to Confluence pages. Confluence pages embed Jira ticket lists. Bitbucket pull requests reference both. The integration is the moat. (See the Confluence and Jira section below — the pairing is most of the answer to why is Confluence still everywhere.)

The full explainer with screenshots and the editor demo lives on Atlassian’s product page; we are not going to be the post that re-types that.

Confluence and Jira — the pairing that pays the bills

The single biggest reason Confluence is still the default corporate wiki in 2026 is that Jira is still the default issue tracker. Atlassian spent twenty years making the two products feel like one. Most of that integration is genuinely good.

A Jira issue can link to a Confluence page describing the feature; the page renders the issue list inline. A Confluence decision record can reference the Jira epic that sparked it, and the epic page knows the decision. A project page auto-generates a status table from the Jira board. Bitbucket pull requests reference both. The shared user directory means SSO is one decision, not two.

The combined value is real. Confluence build — the part of Confluence buyers ask about most when they’re standing up a new workspace from scratch — leans heavily on this glue: pages that know about builds, builds that know about pages. The single biggest reason teams stay on Confluence past the point where the wiki itself frustrates them is that the team is on Jira and the Jira-and-Confluence loop is the part nobody wants to unwind. (We have made our peace with this; the migration story later in the post is about the teams whose answer changed.)

For the broader software-roundup framing, see our knowledge base software shortlist — the Atlassian pairing earns Confluence the Atlassian-bound engineering org recommendation in our table.

What Confluence does well

Honest section. Things Confluence is genuinely good at, with no caveats.

Documentation depth. Confluence pages are dense and rich. Attachments, in-line images, version history per page, inline comments, page restrictions, a full macro catalogue — the kit is wide. For long-form internal documentation, Confluence has the deepest editor of the major options.
Templates and blueprints. A new space starts populated. Meeting notes, retros, decision records, project pages — the blueprint catalogue is genuinely useful, and a freshly-onboarded team can be writing useful pages on day one.
The Jira integration. Already covered above; the loop between issue tracker and wiki is the strongest in the category. Anyone telling you their two-product issues + docs stack matches the Atlassian glue is selling something.
Permissioning. Spaces and page-level restrictions handle most enterprise permission shapes well, including the messy engineering can read finance’s hiring plans but not the comp band tab case. Audit logs are competent.
Marketplace ecosystem. Twenty years of macro / app development. If your team’s idea of a wiki includes the embedded Gantt chart we bought from a vendor in Helsinki, Confluence is the wiki that can host it.
It exists everywhere. This is not a feature, but it is a fact. The IT team already has a Confluence licence. The legal team already has a Confluence space. The runbook someone wrote in 2019 is still in Confluence. We already have Confluence is the most-cited reason to stay on Confluence, and it is not an unreasonable reason.

Where Confluence falls short

Equally honest section. Things Confluence does badly, with the same lack of caveats.

Page load time. This is the one most teams notice first. A loaded Confluence page often takes one to three seconds to render on a warm cache; a cold load is worse. Compared to modern wikis answering in tens to hundreds of milliseconds, the gap is the kind users learn by clicking and waiting and then opening Slack while they wait.
Search. Did you mean? on an exact-match query is the Confluence search experience most users develop a quiet policy about. Atlassian Intelligence (their AI layer) helps some; the underlying index is the part that hasn’t changed shape.
Mobile. The Confluence mobile experience is functional for reading and rough for editing. Most teams stop trying on mobile after a week.
Editor pain points. The editor has improved meaningfully over the last five years (the legacy editor is now mostly gone). It is still, by some distance, the slowest of the major editors on long pages with many macros.
The macro tax. Macros are powerful and slow. A page with twenty macros is doing twenty server round-trips before it finishes rendering. Power-users hit this wall first; power-users also tend to be the ones whose pages everyone else reads.
Self-hosted / Data Center pricing. The Cloud product is the strategic one; the on-premise / Data Center product is the one teams who can’t move to Cloud are stuck on, and the pricing reflects what can’t move is worth.

We have a story for the falls-short side. Our internal Confabula export is three years of accumulating grievances written into a canon postmortem — the 22-second homepage, the Did you mean? on exact-match search, the editor silently dropping paste-formatted tables, the mobile app that turned out to be a screenshot of the desktop app. The grievances are canon; the grievances are also recognisable. (For the actually- true performance benchmark, our knowledge base software shortlist has the numbers.)

Confluence is not a villain

This is the opinion section, and we are going to land it gently because the SERP makes us look like we’re picking on the elephant.

Confluence is not a villain. It’s a twenty-two-year-old product whose category has changed underneath it, whose parent company spent a decade investing in adjacent products instead of the wiki, and whose user base is large enough that the upgrade cost would be measured in months even if Atlassian shipped a perfect rewrite tomorrow. The product reflects all three of those facts.

What Confluence is also is the right answer for a substantial fraction of teams — specifically, teams whose Jira investment is real, whose page count is in the manageable range, and whose tolerance for one to three seconds of page load is higher than their tolerance for switching costs. That fraction exists. We are not going to pretend otherwise. (Our imports page exists for the other fraction.)

For the teams that should migrate, the migration is the moat. Most of them know they should migrate; almost all of them are scared of the export. Markdown on demand is the right shape for export; talk to sales is the wrong one. Confluence’s export — for spaces, for individual pages, with attachments, into formats that survive a re-import — is better than its reputation, and importers in the category have caught up. (The ten-minute Confluence import into Raccoon Page is one example; there are others.) The migration is no longer the moat it was three years ago.

When to stay on Confluence and when to leave

The decision tree, briefly.

Stay on Confluence if:

Your team is heavily on Jira and the integration is doing real work.
Your page count is moderate (under ~5,000 active pages per space) and the speed pain is tolerable.
You have an Atlassian admin who already runs the workspace and the marketplace apps that matter to you.
The cost of disruption to the team is high enough that fix the speed in five years is acceptable.

Consider leaving if:

Your team types more than it clicks and the keyboard surface is broken in your daily workflow.
Your search experience consistently fails on queries you know exist in your wiki.
You are paying Data Center prices for an on-premise install and the cost trajectory is wrong.
Your team has switched issue trackers and the Jira-and-Confluence glue is no longer doing work.
You want a wiki that AI agents can read and write to through a real, audit-trailed API. (See our AI knowledge base post for the longer treatment of this constraint.)

The signal we’d watch for, more than any of the above: are your fastest writers writing somewhere other than the wiki? A team whose engineers draft postmortems in Google Docs and copy-paste them into Confluence has already left the wiki; the migration is just paperwork.

For the broader operating-discipline frame on top of any wiki choice, our knowledge management software guide is the companion read.

Things people actually ask

Is Confluence a wiki or a knowledge base? Both, and the distinction is mostly marketing. Confluence ships pages, links, search, version history, permissions — the wiki shape. It also supports the knowledge base framing: spaces of articles, runbooks, FAQs, decision records. The same software does both jobs; teams pick the framing that matches their adoption story.

What’s the difference between Confluence and Jira? Jira is an issue tracker — work in flight, tickets, sprints, boards. Confluence is a wiki — documentation, decisions, runbooks. They share an Atlassian login, a permissions system, and a deep cross-link mechanism, which is why Jira-and-Confluence is sold as a single answer for many engineering orgs. They are different products with different mental models, sold side-by-side because most teams need both.

How much does Confluence cost? Atlassian Cloud Confluence has a Free tier (up to 10 users), a Standard tier (~~$6/user/month at last published price), and a Premium tier (~~$11/user/month). Data Center pricing is per-user-per-year and starts an order of magnitude higher; Atlassian’s pricing page is the only authoritative source because the numbers move. Our pricing block covers Raccoon Page’s tiers if you’re shopping around.

What’s a Confluence build? Confluence build is the working term for the initial workspace setup — spaces, page hierarchy, templates, permissions, macro install — that a team does when standing Confluence up from scratch. It is a project. Atlassian partners exist almost entirely because Confluence build is non-trivial enough that some teams hire it out.

Is Confluence good for documentation? Yes for internal-team documentation, with the speed and search caveats above. For public-facing customer docs, the shape is wrong — Confluence is not optimised as a help-centre, and the URLs aren’t the friendly ones a public KB needs. Document360 / Helpjuice / Zendesk Guide are the better fits there.

Can I use Confluence without Jira? Yes. Confluence stands on its own as a wiki product; many teams use it without ever installing Jira. The trade-off is that you’re paying for an Atlassian product without the integration that justifies most of Confluence’s pricing relative to Notion, Slite, or Raccoon Page. Most pure wiki-only shops eventually migrate.

How do I migrate from Confluence to a different wiki? Confluence supports an HTML/XML space export; modern wiki tools ship importers that consume it. The Raccoon Page Confluence importer handles a typical workspace in about ten minutes, preserves the page tree and attachments, and converts the most common macros (info / tip / note / warning callouts, code blocks, status macros) on a best-effort basis. The hard part is not the export; the hard part is the team agreement.

Confluence is the wiki most teams have. Some of those teams are exactly right to stay, some are exactly right to leave, and the test is honest: are your fastest writers using the wiki, or are they using something else and pasting in? If the answer is something else, the ten-minute Confluence import is the shortest path to the wiki your team is already trying to write into. Raccoon Page Free is three users, one space, one hundred pages, no card — enough to find out.