Readers/Reader Growth/Minimal Minerva

Background

The default mobile skin of Wikipedia, Minerva , is in use by millions of logged-out readers worldwide. Minerva provides a simpler interface compared to skins like Vector , and it optimizes for mobile performance.

Readers who frequently visited Wikipedia in the past are now turning to other platforms like AI chatbots or summaries that provide convenient, easy-to-digest information. As a result, the percentage of the world that visits Wikipedia is decreasing 20% year over year. This harms the global information landscape and has decreased the number of potential new contributors to the projects.

We want to experiment with an improved version of Minerva to see if it improves the logged-out reader experience on articles and can encourage logged-out readers to come back more often. Based on Readers Foundational Research conducted by the WMF UX Research Team, readers felt they had to do “a lot of scrolling” to find information. By offering a clearer and more streamlined version of the Minerva skin, we think we will be able to more effectively serve readers’ needs for satisfying their curiosity.

This experiment does not change the experience for logged-in readers and/or editors.

Hypothesis

If we streamline the toolbelt on Minerva, we will see practically significant improvement to logged-out reader retention.

Experimentation

We will run an A/B test which tests a version of a mobile experience that updates the navigation and information architecture of the Minerva toolbelt (against Control, the current experience). Please below for mocks.

The success metrics will be:

[Primary] Retention
[Guardrail] Constructive edits

Nice to have/secondary metrics include:

Talk page referrals
Language switcher usage
Watch usage
Account creation

Design

The design strategy for "minimal Minerva" is to turn "chrome" into content on Wikipedia articles for casual, logged-out readers. You can view variations of the design in a Figma preview.

This is a mock of the first test we want to run, tightening up the “toolbelt” in Minerva so it (1) takes up less vertical space and (2) shows the actions available to logged-out readers more clearly (compare the current state on the left with the version we will test on the right). If we see retention improve in the treatment group, we will move forward with rolling out this change to all logged-out readers on Minerva.

The design strategy behind Minimal Minerva is to turn the navigation section into content on mobile web, with the hoped-for benefits being a more usable, flexible, extensible navigation bar for logged-out readers. By making the metadata chips cleaner, less mid-grey, and more actionable, we think it might make the functions more usable. It collects actions, entry points, and trust signals all into one flexible, extensible design. It also brings the lead paragraph a bit higher up the page.

Future design explorations could potentially include a more prominent search bar, as well as expansions to add metadata highlighting good article status, edit counts, entry points for explore, suggested edits, and more.

Toolbelt

Challenge

The chrome pushes content further down the viewport and is too high in an article page's information hierarchy. Casual readers have told us as much. For example, a participant in a recent qualitative study said:
- I expect to see more main information appearing on the screen instead of the tools. The tools here are not the primary thing I should do first when I get to this screen. The first thing I get on the screen should be reading the content. So the content should occupy more space on the screen than the other secondary tools.

Interventions:

Put primary page actions (edit, watch) in line with the article title and replace the article/talk tabs with one entrypoint for talk pages.
Lighten the top navigation bar to make it less prominent.

Rationale:

This brings the lead paragraph ~ 40px closer to the top of the viewport and decreases the number of different navigation types in the view.

Language switcher

Challenges

The language icon has no label. Unlabelled icons are less usable calls to action for casual readers.
The icon is also a static chrome element which doesn't contribute to the article content which makes it more likely to be ignored and/or seen as extraneous.

Intervention:

Transform the unlabelled icon for languages into a navigation chip with the number of languages available for each article.

Rationale:

This creates a trust signal for readers, offers a more usable, more prominent affordance for switching languages, and makes the UI control contribute to the content.
When only a few languages are available for an article, we hope the new treatment will invite content translation contributions.
The Movement has made an immense investment in language diversity, content translation, and supporting marginalized language communities. Showing how many languages editions exist for each article surfaces a unique, important, and valuable feature of the wikis.

Talk tab

Challenge:

Casual readers ignore the current "Talk" tab, while adding visual heaviness.

Intervention:

Transform the "Talk" tab into a navigation chip that lists the number of discussions on an article's talk page.
Most wikis do not display the Talk tab to logged-out users on mobile. These wikis will not see any navigation chip.

Rationale:

"Discussions" is a much more normative, familiar mental model for casual wiki readers.
Talk pages already list the number of discussions available on the page, the intervention just surfaces the metadata to readers.
Even if readers don't read a talk page, seeing that discussions exist for an article creates a trust signal through social proof and implies the relative prominence of an article.
In qualitative testing, casual readers who would normally ignore the "Talk" tab noticed and engaged with the "comments" navigation chip [in this early testing, the chip said "comments" – this was changed to "discussions" in the actual experimental feature.]
We're hoping that by reframing talk pages in a more reader-centric way, readers will understand talk pages better, be more likely to read them, and be more likely to create accounts for contributing to the wikis.

Information architecture

Challenge

The current information architecture is limited and inflexible. We can only use unlabelled icons for new features and we can never show more than 5 icons.

Intervention

Flexible, extensible metadata chips

Rationale

Framing actions with metadata contribute to the reading experience by showing relevant information about an article that readers haven't seen before
Labelled icons are more usable than unlabelled ones
In the future, we can add more contextually relevant actions to this part of the UI. For example, we have concept designs for extending the metadata chips to potentially include Good Article, Featured Article, and Vital Article signposts, suggested edits, recent changes, citations, and more.

Download

Challenge

Download is too prominent in the mobile web IA.

Intervention

Remove the download button

Rationale

Downloading pages on mobile browsers is no longer a relevant feature and many mobile users already do not see the download icon by default.
For the relatively small number of readers who still want to download article pages, they can do so natively in all modern mobile browsers.
Offline reading use cases are better supported by the Wikipedia apps.