KR 3.1: Release two curated, accessible, and community-driven browsing and learning experiences to representative wikis, with the goal of increasing the logged-out reader retention of experienced users by 5%.

Hypothesis

If we create a feature to read articles out loud (using text-to-speech functionality built into modern mobile devices), we will make our content more accessible to a broader range of users, which would improve retention and increase interaction with our content for these audiences.

Must Haves

• Navigates to an article in the app.
• User is provided some affordance to initiate text to speech.
• User has the ability to control playback either in the app itself or in the media playback control.
• User must be able to navigate to a different article or flow within the application and be able to persist the listening experience.
• User must be able to have basic control of the listening experience from outside the application.
• When the user is on a different article to the one currently playing, they have the ability to switch to the new article.

Nice to Haves

• Indicate or provide the ability to use spoken or AI version.
• Be able to select system level voices.
• Entry points in other features

How will we know we were successful

Validation

• 1% higher app retention rate for logged-out users that engaged with the feature compared to those who did not engage
• 25% increase in average time spent on articles for users who engage with the feature compared to those who did not engage.
• 10% of unique users that engage with the feature use it more than once in a 30 day period.

Curiosities

• What are the overall average/median session lengths for text-to-speech playback?
• How many users engage with the system-level playback controls, instead of just the in-app playback controls?
• Assuming we offer an option to change the "voice" of the text-to-speech speaker, which voice is most commonly selected? Most commonly selected playback speed?

Resources

• Initial comparative review on Figma
• Original task (T126889) with logs of user requests for this type of feature, as well as initial design explorations.
• The WikiSpeech extension provides insights into massaging article content (i.e. removing elements that should be excluded from text-to-speech narration) that we can likely reuse.