Confluence Should Stop Being a Page Editor and Start Being a Memory Layer

[#1]

Let me ask you something honest.

When was the last time someone in your organisation said Let me check Confluence and actually found the right answer, trusted it, and moved on with their day?

For most teams, the reality is different. They search, find three outdated pages, guess which one is current, and either ask a colleague or give up. That is not a documentation problem. That is a memory problem and Confluence, as powerful as it is, has not fully solved it yet.

This article is for admins, engineers, and everyday users who feel that Confluence could be so much more. I want to share a bigger picture view of what Confluence should become and what many of us in the community have been waiting for.

The Core Problem:

We Store Knowledge. We Don't Preserve Context.

I love Confluence and it is excellent at storing content. But organisations do not struggle with storing information, they struggle with trusting it, finding it, and understanding why it exists. 

Think about these common scenarios:

• A new engineer finds a technical design page. Is it still valid/latest? Who owns it now?
• A product manager searches for a decision made six months ago. Was it documented at all? Which space? Which page?
• An admin looks at their Confluence instance and sees 4,000 pages. Half are untouched in two years. Do they really needed?

These are not edge cases. These are daily frustrations in almost every team using Confluence at scale.

The Bigger Picture:

Confluence as an Organisational Memory Layer

Here is the shift in thinking I'd like to propose.

Confluence should not just be the place where you put things. It should be the layer that preserves, connects, and surfaces what your organisation knows automatically, intelligently, and in the right context.

That means evolving from a page editor into something more powerful. 

What is Confluence Today	What it Should Become
A place to write and store pages	A living layer of organisational memory
Manual search	Smart search/Intent Aware
You know what exists if you know where to look	Discoverable knowledge, even across permission boundaries
Page ownership by whoever created it	Governance with accountability and lifecycle management

Long-Awaited Features !!

That Would Change Everything

Here are the capabilities many of us in the community have wanted for years and that would transform Confluence from a document tool into a true knowledge platform.

The Self-Healing Knowledge Lifecycle
Historically, Confluence required manual governance. If an admin didn't set a reminder to review a page, it sat there forever. The future of Confluence relies on automated, algorithmic lifecycle management.

• Auto-Expiry & Verification Workflows: Pages shouldn't live forever by default. New governance tools allow space admins to set an expiration date or verification cadence based on page labels.
• The Trust Score Indicator: A subtle, visual badge at the top of every page indicating its freshness. A page viewed 500 times this month with recent commits linked from Jira gets a green 98% Trust Score. A page unedited for 18 months drops to a amber 40%, warning users to tread lightly.

What Admins Need Most

Admins are the ones keeping the lights on in Confluence. Yet the tools for governance, hygiene, and quality are still surprisingly limited. Admins should have:

• Bulk ownership reassignment when someone leaves
• Automated archiving policies based on inactivity
• Space health scores visible from the admin console
• Usage insights per space, per team, and per content type
• Clear audit trails for sensitive or compliance-related content

Without these, admins are left doing manual audits of thousands of pages and not a sustainable model at any scale.

What Engineers Need Most

Engineers want to trust what they read. For them, Confluence fails when:

• A technical design page has no clear status (draft? approved? deprecated?)
• There is no link between the page and the Jira epic or PR it relates to
• They cannot tell if the architecture diagram reflects the current system or one from three years ago

Engineers need status labels, version context, and Jira integration that goes both ways and not just a link, but a live relationship between the work and the knowledge about it.

What Users Need Most

For everyday users, the ask is simple: help me find the right answer fast, and trust that it is current. 

That means better quality search, smarter suggestions, proactive surfacing of relevant content, and clear signals about whether a page is up to date. The best version of Confluence is almost invisible because knowledge finds you, rather than you hunting for it.

Closing Thought

The best Confluence is one where people do not have to remember where things are stored, because the right knowledge appears when and where it is needed in a Jira issue, in a search result, in a notification, in a teammate's comment.

We have the foundation. What the community is waiting for is the evolution from *document storage to organisational intelligence.

Would love to hear from others:

• What feature or change would make the biggest difference ?
• How your team uses Confluence? 

[__ Jimi Wikman]

These are all valid points, but one of the things that is most often overlooked is structure.

Almost all Confluence spaces I have seen are created without any form of structure. There is no information architecture in place, no standards, just notebook styles of the mad professor style.

This is for me more important to address than tools to keep track of information lifecycles, but I like that you are pointing out that information and collaborative information structures need lifecycle states and governance.

With Approvals, we are getting into this space, I think. With classifications in Guard, we have a second layer. We have ownership, and we can set automations for different things.  

What I miss is the ability to set more metadata and have it actually be useful... and that Approval feature so I can test it out :)

[Tanya Christensen]

@__ Jimi Wikman This is the Approvals feature:  Approvals in Confluence, now on Open Beta 

[Stavros_Rougas_EasyApps]

Another insightful post @Viswanathan Ramachandran. So well thought out. Atlassian should hire you!

You have a great technical take on the problem, but I will add that no solution works unless it takes into account reality - humans. This idea that we are a few features, automations or AI away from a breakthrough over simplifies (you are not saying this, but this is a tech dream to avoid unstructured information).

The easy black or white problems are few as we have limited them. We are left dealing with the harder stuff, why we have jobs, pay for tools like Confluence, hire Atlassian admins and pay for apps.

Review and clean-up can't be 100% automated, at best a few parts. I get your point about the "Trust Score Indicator", but it likely doesn't offer as much as expected, it can't solve larger underlying structural problems. I've seen it, gives a percentage, looks great, but too often doesn't offer much.

What features, apps, automation and AI can do is augment the human. Take action in less time and even better identify what needs to be reviewed.

- Bulk ownership reassignment when someone leaves
- Usage insights per space, per team, and per content type

Engineers need status labels, version context

Our app Space Content Manager deals with these and a bunch of other things. It helps identify and bulk change. We would build a Rovo integration that would in a query solve problems. We haven't because makes for a cool demo in a sandbox, but nobody would use that in production.

But as I often comment previews exist because in the end most things need a human review before submitting changes. Sometimes a few seconds. And sometimes realize, wow, this is not clear. I better review this before creating a mess. The ladder avoids making the problem worse, like can happen with an over reliance on tech.

Confluence content is a powerful driver for great AI results within your organization. But it's garbage in, garbage out problem. Work faster and smarter, but the challenge is that the work left requires higher skills, not a junior.

[Stavros_Rougas_EasyApps]

I want to second @__ Jimi Wikman point about structure. I will explain it like this. Confluence will become imperfect. If the basic structure is bad it can become a mess. It is much harder to clean up a mess then avoid the mess (how many of us as teenagers did not understand this).

The good news with AI is that Confluence content need not be perfect, I think if it's pretty good AI can interpret it well. Many Confluence instances are a mess and now Rovo/AI is highlighting how a mess kills productivity.

You need a decent structure. Use some automation. And ongoing human review/manual clean-up by people with content knowledge that can only to a point be helped by tools.

[Viswanathan Ramachandran]

Hi @__ Jimi Wikman

Really well said. Structure is the foundation and without it, lifecycle and governance can only do so much. I love your point about metadata too, that’s exactly where Confluence can become much more useful. Approvals and Guard feel like promising steps in that direction, and for the approval we have Comala Document Management which is great and it has something like auto-expiry and verification cadence on Approval. 

What Comala does not do and what Confluence still missing natively comes down to Intelligence, Integration, and Automation Friction. Ex:

• It operates on linear time, not Human Trust
• It doesn't solve the Duplicate Page problem (Semantic Intelligence)
• Comala lives entirely in Confluence UI and it doesn't bridge the gap to Code (Git/Markdown Sync)

In simple, Comala is like a security guard standing at the door, checking IDs and very strict. But what Confluence really needs natively is a digital janitor(always underestimated yet the key role) and editor who actively cleans the rooms, throws out duplicates, and connects the building to the outside world.

[Viswanathan Ramachandran]

Hi @Stavros_Rougas_EasyApps 

Thank you, that means a lot. I’m glad the post resonated, and I really appreciate you adding such a practical perspective. The human side is exactly what makes these discussions useful.

This is thoughtful feedback, and I really appreciate the depth of it. You’re absolutely right that structure and human judgment are the foundation, and tools only work well when they support that reality rather than pretending to replace it.
I also like how you framed the balance so clearly: automation and AI should augment people, not try to bypass review, ownership, or context. That is exactly the kind of practical perspective Confluence needs more of, and your point about garbage in, garbage out is spot on.