The Pillars of Eternity wiki was one of the very first Gamepedia wikis, originally established in 2014. The wiki gained traction thanks to the efforts of Ausir, one of the original Gamepedia wiki managers, and developed a respectable base of content – enough to become one of the key wikis on the network.
I asked to be assigned as the wiki’s manager in January 2017, a year and a half before the sequel’s release. My first move was to take stock of the current state of the wiki, including content, design, and identifying potential areas of improvement.
The Plan
The wiki generated over 20 million page views in 2015, after the first Pillars hit the market. I was confident I could break that record: There was a lot of content, but there was still room for improvement!
Taking into account the wealth of content present on the wiki already and my experiences as online market researcher, content marketer, and – of course – wiki editor, I’ve prioritized the following:
- Overhauling navigation to make existing content easier to find.
- Upgrading the wiki skin to give it a more modern look.
- Expanding content to cover everything about the world of Eora and the games’ content.
The first two points were intended to refresh the wiki’s appearance and communicate to visitors that the site is maintained and continuously improved. After all, the skin is the very first impression a user will get!
The third point was a comprehensive overhaul spread out over the next year or so (since this was not my only wiki), and was driven by a simple philosophy:
Great content attracts great people, who create great communities.
The Implementation: Navigation and Skin
The first step was achieved by overhauling the front page and the skin. The old one was replaced by a responsive layout provided by Gamepedia, organizing content to make it less crowded and help guide the reader towards the content.
The skin was rebuilt from the ground up using UI assets datamined from the Deadfire beta. This served two purposes:
- It signaled that the wiki was going to be updated for Deadfire and has active users.
- It created a sense of versimilitude, resembling the game’s user interface.
We iterated on the front page design together with users Pangearocks and CompleCCity, eventually settling on the two-pane responsive layout in early 2018, which remained in use ever since:
While it might appear a little more crowded than the three-pane layout, the goal here was to make better use of screen real estate and communicate instantly to the reader that this is the one-stop shop for all things Pillars.
Gameplay content on the left-hand side allows for navigating to top-level articles for major game elements, while the right-hand side provides easy access to articles that help understand the dense, complex lore of the world of Eora. The dividers (also taken from Deadfire‘s user interface) help break up the text and provide an additional navigational element.
The Implementation: Content
One of the most ambitious projects I implemented from start to finish. For Pillars, it was a three-pronged approach:
- Standardize the layouts of every article to make them easily understandable, searchable, and attractive to search engines and users alike.
- Fill in the blanks by covering every character, location, item, quest, and interaction in detail, to target the keyword long tail and increase user acquisition and retention.
- Upgrade existing content, incluing lossless compression for assets, game maps, and everything else.
Take, for example, the article on Thaos ix Arkannon, the antagonist. The article originally provided decent coverage, especially in terms of gameplay. My improvements focused on standardizing the layout and expanding the content. The approach here was simple:
- Anyone interested in the lore and plot of the game should be able to use the wiki to understand precisely who Thaos is, what’s his story, and easily piece together what happened (Pillars has a tendency for major exposition to be either packed into either huge walls of texts or squirreled away in remote dialogue nodes).
- Anyone interested in gameplay and quests the character is involved in should be able to immediately find that information either through the infobox or the Interactions template that unambiguously shows what roles the character plays, if any.
The same approach was repeated for virtually every one of the 300+ characters the game has, not to mention quests, locations, items, and lore articles.
This approach resulted in overhauled, quality articles that provided an excellent foundation for other users to expand and build upon. High-quality images helped user retention as well, especially maps of locations composited from raw game files using ImageMagick and clever spreadsheet use.
Simply put: Existing content (overhauled or newly added) attracted visitors, the standard layout and the right amount of content encouraged them to fill in the blanks.
For Deadfire, I adopted a “seeding” approach. In order to attract users and reduce the difficulty of making the first edit, I datamined the game, processed the data, and used AutoWikiBrowser’s excellent CSVLoader plugin to generate templated articles en masse.
For example: Storm Speaker Ikawha is a character found at Port Maje. The initial page was seeded with an easy-to-understand and fill-out template; the image name was auto-generated based on the article name, which allowed for quickly uploading images and having them automatically display.
It was filled-out a few days later, making it an MVP (Minimum Viable Page), and subsequently filled to completion by two absolute fans, Macklin and KFrisson, bringing it to its current, awesome state.
The same treatment was given to other pages as before: Items, characters, locations, quests, the whole nine yards. It was the same basic procedure.
- Start with an automatically generated article that acts as a skeletal structure or foundation (Port Maje, Humility).
- Fill out essential information to make it an MVP (Port Maje, Humility had all basic information filled out thanks to extracting data from files)
- Let users take over and fill out the blanks (Port Maje, Humility).
Users did take over and filled out virtually every article on the wiki after the initial content seeding was done. What’s more, when the DLCs for Deadfire arrived, they continued their work and added content of their own initiative, using the framework and tools provided to expand the wiki.
End Result
The scale of content added and long-term support achieved all the objectives I defined for it:
- A large basis of articles for Google to crawl, providing good coverage in keeping with the MVP principle (Minimum Viable Page means the page answers most or all basic questions, such as where to find X, how good Y is, or what’s the description of Z for those of us who love lore).
- A self-sustaining, active community acting independently. Without users such as Macklin, FurloSK, Pangearocks, CompleCCity, Arawn76, and scores of other registered and anonymous users it never would have become such a great wiki.
- Outperformed rival wikis and game walkthrough sites well into 2022.
In terms of raw traffic, the number of page views has almost doubled in the year of Deadfire‘s release (from over 21 million from January 2015 to January 2016 to over 38 million from January 2018 to January 2019), while traffic in “slow” periods between DLC releases has also increased substantially – users kept returning to the wiki to look up information about lore and gameplay. And that’s despite increasing competition in the online community market and the existence of rival Pillars wikis.
Topics announcing the wiki were also pinned on the official Obsidian forums, which doubled as a very useful source of feedback on wiki structure and missing content. In fact, the Pillars Gamepedia, as the wiki came to be known, became shorthand in the community for quality coverage and a model example of what a wiki should be.
After all, the underlying principle was, is, and will be
Great content. Great people. Great communities.