Suche
Suche Menü

Five years of using Confluence: Seven lessons learned

We started using Confluence almost five years ago. The tool quickly became our central collaboration platform and eliminator of the internal email. Today, our Confluence is used by around 50 users as a central hub for collaboration and documentation.

Here are seven lessons I have learned as a Confluence administrator over the years:

1. Theme It!

The out-of-the-box Confluence comes with a rather tame design and very limited room for customization. That’s why we have been using RefinedTheme since day one, as it allows us to use an attractive, CI compatible design – and even more importantly, we now have full control over the dashboard: our customized version includes, for instance, highlighted blog posts, a microblog, latest meeting notes, upcoming conferences, and calendar events.

 

An additional benefit is that spaces can be organized into categories and sub-categories:

2. Understand your Users

In my experience, the feedback Confluence administrators receive is very situational. For a better understanding of how users use Confluence, we performed a two-step survey:

The fist step was to conduct guideline-based interviews with twenty users, asking them about topics such as typical use cases, pain points, improvement ideas, or required information input to get work done.

Based on these findings, we compiled a survey with 28 statements to determine improvement potential for each of those items. Users could rate their assessment of overall importance and their perception of the status quo, allowing to calculate an improvement score to visualize the users‘ pain points. It became clear that slow load times were by far the most critical issue, followed by poor quality of search results, and overall comprehensibility of the site structure:

With these findings in mind, we were able to take numerous specific improvement measures, such as

  • splitting the largest spaces into several small ones,
  • shortening click paths for popular pages by reworking page tree structures,
  • introducing automated archiving (see also lesson 5),
  • creating topic-related portals (see also lesson 6), or
  • taking measures to increase label usage and improve search results, e.g. label visualization using the TagCloud AddOn.

3. Understand your Content

Apart from asking the users about their perceptions, we gained valuable insights by analyzing usage statistics. We found out that, despite having a total of over 150 spaces, over 80% of total views and 80% of total edits took place in 13 key spaces.

Knowing this, we wanted to find out more about the content in these spaces, and manually analyzed the top pages (or blogposts, respectively). Each page/blog post was checked for certain attributes such as hierarchal level, number of labels and comments, age, and „content tags“ (e.g. “internal announcement/update”, “technical documentation” or “request for feedback/discussion”).

This delivered valuable insights, such as the finding that “hot” pages were pretty badly located with an average hierarchical level of 3,66.

4. Dynamic Content is King

Let’s face it: No matter how hard you try to keep content up-to-date, much information is outdated. That’s why it makes sense to maintain certain content in a central place and dynamically embed it where needed. This is why we are in a continuous process of making more and more content dynamic – as an example, we maintain our team structure through the Muti Excerpt AddOn. The respective excerpt is then embedded on each team space dashboard. Another example is the PocketQuery AddOn that allows us to embed dynamic database queries like an overview of available disk space on our servers.

5. Automate Archiving

Being the wiki gardener is not exactly the most popular job, and a rather small company like ours doesn’t have the resources for a separate wiki gardener role. On this account, we introduced the Archiving AddOn in late 2015. Starting this experiment in two spaces, automated archiving is now activated for over 90% of all spaces.

The underlying concept is pretty straightforward: For every page, there is a view age and an edit age – the date when it was last viewed or edited, respectively. Once a certain threshold is exceeded, the page is automatically being moved into an archived copy of its original space. 10 days before expiration, the page’s author and the last modifier are notified via email, giving them possibility to prevent the page from being archived.

To keep things simple and transparent, the threshold values are set on a global level. Users can give one-click feedback (“too high”, “ok”, “too low”) via the Polls AddOn. Today, almost 25% of overall pages have been archived, increasing search results quality and overall comprehensibility.

6. Create Portals

In every wiki, there is a range of similar content that can be organized into portals. Linked to the corresponding templates, a structured framework keeps all content in place.

In our Confluence, examples of portals include upcoming conferences, procurement and travel requests, or an overview of current projects. We found the PageProperties Macro to be very helpful, as it allows to embed pre-defined excerpts from a set of pages onto a central overview page. The Table Filter AddOn offers a great possibility to handle larger sets of data.

7. Keep it simple!

Last but not least, this is a lesson we learned the hard way: For our initial Confluence, we set up a flexible but confusing permission scheme and a complex ownership concept on space and category level. However, in line with our open corporate culture, all users had the same level of access anyway, and the ownership structures quickly faded into obscurity. These examples taught us that overambitious attempts cause a lot of work for admins, while they don’t deliver any value for the users.

What are your Confluence learnings and best practices? We’d love to hear them from you!

Schreibe einen Kommentar

Pflichtfelder sind mit * markiert.


Agile Softwareentwicklung