= Summary
Page previews is a beta feature that has been tested by 10s of thousands of users across wikis. It shows a preview/summary of a page when a user hovers over a link. The user can move the mouse to dismiss the preview or can click on the link or the preview to continue onto the article.
This is a feature intended to improve the experience for any reader who normally would have clicked on a blue-link in wikipedia because they needed an overview (definition) of that entity. We know from preliminary survey results that ~1/3 of readers come to wikipedia for an overview, but once on site, when reading a page, the number is surely higher. The goal of hovercards is to:
# better serve the need when someone is looking for a definition,
- quicker
- less context switching
# not inconveniencing a user who wants a deeper dive into a subject
- only shows up after a pause
- is easily turned off
== Goal Visibility
Graduating page previews on is committed externally here: [[ https://www.mediawiki.org/wiki/Wikimedia_Engineering/2015-16_Q4_Goals#Reading | https://www.mediawiki.org/wiki/Wikimedia_Engineering/2015-16_Q4_Goals#Reading ]]
https://www.mediawiki.org/wiki/Wikimedia_Audiences/2017-18_Q3_Goals#Readers
== Current Status
As of January, 2018, page previews are live on all wikipedias but German and English.
= Rationale
- well tested
- readers of catalan/greek wikipedia for whom this was turned on liked the feature 2:1
- relatively well liked by editors (recent english wiki RFC [[https://en.wikipedia.org/wiki/Wikipedia:Village_pump_(proposals)/Archive_131#Proposal:_Enable_Hovercards_by_default| shows mild preference against in it's current state]]
- leads to more learning! (quick glance at greek/catalan data shows that pageviews did not drop measurably, but the use of hovers was a significant additional action.
- arguably visually appealing
= Success Metrics
Our next step is to a/b test this on a smaller wiki. Our metric of success is that overall engagement pageviews + hovers increases significantly.
There is an issue to consider, which is that hovers can be accidental. To try and identify the impact of this issue and characterize the types of hovers to discount, we will be measuring hovers of a control group (without hovercards enabled) to identify what hovers look like and how many naturally would have triggered a hovercard if they were enabled.
The results of A/B tests and qualitative testing on the feature can be found here: https://www.mediawiki.org/wiki/Page_Previews#Success_Metrics_and_Feature_Evaluation
Results from the most recent A/B test on English and German wikipedias are expected in early 2018
= Product Plan
(in rough order)
- [- ]Generate rough timeline
- [- ]Rename
- [-]Team instruments and familiarizes with code - soon
- [-]Fix breaking bugs and move to RESTbase
- [-]Rollout on small wiki
- [-]Improve disable/enable workflow - in progress
- [-]Qualitative analysis
- [-]Quantitative analysis
- [-]Image fixes
- [] Improving summary endpoint
== MVP
No new features: hovercards feature does not
- is scalable
- is net positive for users
- does not break existing workflows
- does not break anyone's experience
- can be measured
- is easily opted out of
- does not have misleading images
- does not interfere with fundraising
- community on any particular wiki is okay with it
== User Stories
As a reader of a wikipedia article looking at a blue link I am not familiar with, I want to learn what this [entity] is without losing my spot or waiting for another page to reload
== Delivery Estimate
We are expecting to deploy to English and German wikipedias in Q3 of 2018 (Jan - Mar)