In order to learn more about our users (specially learning more about our readers, from whom we know very little), the Reading Web team proposed to create a mechanism to poll users in-article.
We know little about our readers, and why they visit our wikis. What are they looking for? Are we serving them well?
In order to learn more about the millions of silent reading visitors we'd like to ask them questions designed to know more about the reasons they are reading the content.
Initial proposed solution
Outline and discussion in the [GOAL]: Develop a mechanism for quick surveys on desktop + mobile phabricator task
Create an extension to show surveys that will appear in-article to ask the users.
We will implement two types of surveys:
- In article banner with a question and pre-determined answers.
- Banner with link to external poll.
- Develop a banner with a link to an external poll that we'll be able to activate with configuration variables.
- Develop a banner with Q+Answers that we'll be able to activate with configuration and that will log to Analytics-EventLogging
- Surveys can be configured needing minimal development effort.
- Surveys can be bucketed to a % of users.
- Surveys can be shown in the mobile and/or desktop site depending on configuration.
- Surveys can be time boxed to a period of time via configuration. (Enabled and disabled via configuration)
Implementation will happen creating the QuickSurveys extension for the moment.
There's been discussion on integration with the CentralNotice extension but that was put on hold until this christmas fundraising and an in-progress refactor is finished.
Suggested sample mockups
This are only initial mockups, please don't focus on the actual questions/answers.
For the initial version the placement of the survey will be:
Place the survey after the lead section, before the section headings and the table of contents
Survey will be placed after the lead section, appearing on the right column above the infobox if there is one.
Survey placement is in discussion for further improvements, since for example on mobile depending on the length of the lead section the question will appear higher or lower, fact which may bias the answers of said survey.