Wikimedia Apps/Team/Why 5
This page is obsolete. It is being retained for archival purposes. It may document extensions or features that are obsolete and/or no longer supported. Do not rely on the information here being up-to-date.
- 1 Introduction
- 2 Goals
- 3 Major Changes
- 4 Participate
Major content platforms on the internet have largely seen a shift from desktop web traffic to a mix of mobile web, desktop web and app traffic. Wikipedia on iOS has not seen this same migration to apps, despite releasing well reviewed official apps in recent years.
Additionally, when we have had influxes of new users to the app, for example from being featured in the App Store or through PR efforts, we have not ben able to retain those uses and grow the daily overall audience for the app.
Our goal with this release was to redesign the app as inviting and compelling experience that encourages re-use and regular engagement with Wikipedia.
Given our current app's good reviews and positive press, but inability to build and retain a meaningfully large audience, the team decided to effectively reset the app, and invest in new user experiences rather than new functionality. The goal was to rethink how people access written content (especially medium or long form written content) via apps 7 years into the iOS app store's development, updating the look-and-feel and navigation of the app around core use cases of discovery and reading.
Our main goal is to create pathways that encourage users to open the app more than once. If we are truly providing value to our users, and making it easy and inviting to read Wikipedia, we should see more people who are willing to try the app use it again.
We're using an industry standard "7-day" metric for user retention with a target of 15%. This means our goal is to get 15% people who download the app to re-open the app 7 days later. See our teams goal pages for specific targets and quarterly goals.
Our team has a few guiding aspirations we're using as we developed these ideas and weigh trade offs. We asipre for the next version of the app to be:
- Sticky - Our app will engage users and keep them engaged with Wikipedia not just once, but as a habit.
- Mobile-y App-ish - We will focus on features and user experiences specific to native apps and iOS devices. We won't re-create the browser.
- Responsive - The app will respond quickly and smoothly to user interactions and load content as quickly as possible.
- Polished and On Brand - The app will look visually consistent and use standard Wikimedia branding, such as icon forms and terminology.
- Privacy Friendly - We will advance our mission value of privacy by being transparent and "opt-in" oriented about data collection.
- Feature-able - Our app will be good enough to be featured by Apple in the App Store.
- Well Engineered - Users will have an experience which is crash-free, and volunteers and staff will be able to easily contribute patches to the project.
- Accessible - We will further our mission and values by making an app that is easy to use for the disabled and non-English speakers.
The feed is an experiment in creating a new, mobile-friendly way to experience that joy of discovery we all love about Wikipedia.
The feed was designed as a way to enable discovery via a popular touch based app paradigm. However, the feed also drives part of our larger goal of retaining users. In “How to Build Habit Forming Products” Nir Eyal provides a 4 part model for how/why products are habit forming. The basic process is a behavioral loop consisting of:
- Trigger - In our case an internal trigger of boredom and curiosity.
- Action - Physically scrolling the feed looking for something interesting, relevant or inviting.
- Variable Reward - One key to this model is that the reward for the action must be variable. That is, sometimes you find an interesting fact or picture to satisfy the trigger, sometimes you do not.
- Investment - By reading articles and reloading the app regularly new items are added to the feed (such as recommendations based on recent reading) which sets it up to be new and interesting when the loop is next repeated.
A centralized, diverse and dynamic feed is a key component of this loop, which we hope will get users exploring the diversity of content Wikipedia not just once, but as a regular habit.
Content: Global + Local + Personal
In order to increase the chances of you finding something that personally connects and sparks that moment of serendipity, we've combined several sources from across Wikimedia sites, and selected via diverse approaches.
- Global - Content curated by our community and shared by everyone across the world. Featured articles and pictures, for example.
- Local - Articles tagged as near your device location
- Personal - Similar articles based on articles you read or save
On first install, the feed includes:
- Featured Article of the Day from en.wikipedia.org
- The Top 5 most visited articles from the previous day in your primary language
- Quick link to Wikipedia Main_Page in your primary language
- Random Article from Wikipedia in your primary language
- Picture of the Day from commons.wikimedia.org
- Wikipedia Articles tagged near your GPS location, their direction and distance (if location permission granted)
As you read or save articles, the feed will begin to include recommendations based on your reading. Note that we do this by running simple "more like" queries, not through building user profiles.
Additional triggers vary by content type (location changes, for example). But the goal is to provide a stream of diverse Wikipedia content that might get and keep you reading.
Currently the feed is created and updated in a very simple deterministic way. In order to make a truly efficient serendipity machine there are a number of potential improvements that could be pursued. Eventually such a personal, mobile recommendation system would be an ideal project to engage our technical community and researchers in AI. Because this is a new and unique way to explore Wikipedia, we want to get some basic feedback and baseline data before we invest in a more sophisticated approach, if we ever do. However, it's fun to speculate:
- More Main page content across more languages: news items, this day in history, ...
- Many more data sources from across wikimedia: wiktionary word of the day, wikivoyage tours near you, etc.
- Machine learning to personalize feed content as you use it
New Look and Feel
Updated for Today's iOS
Over the development of iOS user expectations for apps have evolved. While visual design remains extremely important (42% of users say they delete an app if they disliked the user interface), user expectations around interaction design have also grown and developed. Although the current app maintains consistency with our Android app, it hasn't maintained that consistency with the platform. As part of this re-design, we wanted to give iOS app users an experience that was built just for the current platform. Some specifics include:
- Swipe to navigate back and forth between articles and show the table of contents
- 3D touch to support previewing pages (on supported 6S devices)
- Adopt system font and built-in OS text styles where possible
- New flat, thin, icon forms
- New universal alerts that can be dismissed (including via swipe)
A major design goal with this update is to give the app a consistent character, reflecting what Wikipedia is. When your serious friend tells an off color joke, it is all the more shocking because it seems out of character. A major goal of this redesign was to embrace the character of Wikipedia and convey that through everything from icons to informational text.
- Serious (not cute or silly)
- Educational (not funny or social)
- A utility of the internet (not frivolous or trendy)
- Focused on the written word and image
We want users to feel that we are being our authentic selves and convey that with each word, pixel and animation.
In this version of the app we're using the Tab bar paradigm common across iOS apps. This type of navigation has a number of advantages over the existing "W" navigation menu approach, including:
- One touch access to lists of personally relevant articles: exploration feed, saved articles and recent article history
- Entry points for key stories: visual reminder of articles you may want to continue to read and explore
- iOS standard paradigm across many app types (everything from Cydia to Facebook, to the built in Contacts, to the Pokedex)
- Phone form-factor friendly as main navigation is near thumb
- As an app specific UI pattern, it makes app feel “appy”, and less like a browser.
First Time User Experience
Although people are likely familiar with Wikipedia, they are likely unfamiliar with interacting with it through an app. In the previous version of the app new users were prompted to Log in on first use, which was not particularly in line with what users expect of Wikipedia. We decided to use that first time experience to introduce new app features, and help the user with configuration.
Current first time user experience:
- What's new - Provides information about the new features of the app.
- Language preferences - Allow the user to choose top three language preferences, based on device settings. Also aids in discovery of other language wikis.
- Privacy opt-in - Ask the user to volunteer to sharing analytics data. We want to be transparent and up front about our respect for user privacy.
Sign up on this form to help us as a beta tester!
Or post on the Talk page here.
- See this (somewhat dated) breakdown showing how small the traffic is from iOS compared to other sources:http://stats.wikimedia.org/wikimedia/squids/SquidReportOperatingSystems.htm
- Currently 4/5 stars on the app store: https://itunes.apple.com/us/app/wikipedia-mobile/id324715238?mt=8
- App Analytics retention report is access limited, but available internally here: https://analytics.itunes.apple.com/#/retention?app=324715238 Based on opted-in users sharing data with Apple. Opt-in rate is 25-30% and is reported on that same page.
- This goal is somewhat arbitrary. Our August D7 rates ranged between 8% and 12%. Its not 100% clear D7 a "good" rate is. There are many published reports for what a healthy retention range is, but their is significant variance by app category, geography, install source, etc. and much of this data is out of date: https://mixpanel.com/blog/2013/11/04/trends-report-the-new-standards-for-mobile-app-retention/,+http://info.localytics.com/blog/app-retention-improves https://www.quora.com/What-are-typical-long-term-retention-rates-for-a-consumer-app https://www.theinformation.com/which-apps-retail-their-users-and-which-ones-dont?token=70bce9bb72a330a6f4ae89878597b25e
- What is “sticky”? http://www.webopedia.com/TERM/S/sticky.html
- I made this up. It means leaning on features or capabilities only a mobile native app can have, as well as being purpose designed for touchscreens, etc.
- The median user expects an app to launch in 2 seconds. No one is willing to wait more than 4 seconds. http://techcrunch.com/2013/03/12/users-have-low-tolerance-for-buggy-apps-only-16-will-try-a-failing-app-more-than-twice/
- More than anything this means consistency of design language, fonts, etc. And for the record: our brand, not Apple’s.
- Privacy is not just an issue for us, as a value of our movement, but also matters to users: “Increasing customer concerns over app access to their information; upwards of 54% of users have either not downloaded an app or deleted an app relating to privacy concerns” http://www.pewinternet.org/2012/09/05/privacy-and-data-management-on-mobile-devices/
- This has growth benefits to downloads, but is also about pride of work. It matters less that our app be featured (http://denzhadanov.com/2014/03/26/how-to-get-featured-on-the-app-store/) but that it be good enough that it could be featured. The current app has been featured several times, so we should continue that tradition.
- In addition to negative reviews hurting install rates and user perception of the app, crashes kill retention: 79 percent report that they would only retry an app once or twice if it failed to work the first time. http://techcrunch.com/2013/03/12/users-have-low-tolerance-for-buggy-apps-only-16-will-try-a-failing-app-more-than-twice/
- Apple and iOS provide the best support for the community of people with disabilities, and for many the iOS apps are there only way to easily access the internet. http://www.slate.com/articles/technology/future_tense/2015/10/tech_companies_should_care_more_about_customers_with_disabilities.html
- I highly recommended the cited YouTube video which gives a 10 minute explanation of this framework: https://www.youtube.com/watch?v=hVDN2mjJpb8