Wikimedia Apps/Team/iOS/Personalized Wikipedia Year in Review

Background

Year-end reports with insights on user's habits and achievements have become standard in many apps. They have the potential to peak user's curiosity to learn more about their own habits, compare themselves to others, and inspire a deeper connection with the platform.

We've seen indications that providing either a collective or personalized wrap up of Wikipedia insights is something that would be received well. The Mobile Apps team has seen social media posts where users share screenshots of their open tabs, or reading histories, to illustrate their wide and varied Wikipedia reading habits. Each year, the communications team posts a blog about the top 10 read articles on English Wikipedia, which garners positive engagement from readers. Volunteers have also developed a personalized prototype of a year in review for editing, and it was received well, and improved at the Hackathon in 2024.

The Apps team plans to explore if a personalized Wikipedia Year in Review feature that displays insights about a user's reading history, editing history, and donation history is engaging for App users, and inspires them to make a donation to the Wikimedia Foundation.

As a part of new engagement experiments the product and tech team also plans to create a Wikipedia Year in Review site that showcases collective insights about the year. The apps team plans to leverage insights from this into a version for app users that do not have any personalized insights.

Why the apps?

The apps have the unique capability to provide personalized insights while respecting privacy, due to the app's ability to store information locally on the user's device. Data including what articles users have read, or saved, is all stored locally on the users device, and reading history can be cleared by the user at any time. We can process this data for display within their Wikipedia Year in Review without sending pageview data to our servers. The Apps teams have created internal prototypes proving out the concept in 2023.

The Mobile Apps provide a natural place to experiment: the app audience is much smaller in scale than Mobile and Desktop Wikipedia, and there have been recent investments in improving the donation experience by adding native ApplePay and GPay, based on community requests. Additionally, the technical nature of the apps affords us the opportunity to run quick experiments and get feedback in a way that is not as disruptive as on Web.

User stories

User stories are simple statements that product teams use to focus on what we're trying to accomplish. These are some that we're thinking about for this feature:

As a Wikipedia reader, I want to be visually struck and leave the experience remembering how cool and useful Wikipedia is —inspired and in awe at the breadth and depth of human knowledge that can happen over the course of a year, and ready to give back with a donation.
As a Wikipedia reader, I want to see a breakdown of my reading habits in the App, so I can share my quirky Wikipedia reading habits with friends on social media.
As an editor on Wikipedia, I want to see insights about my editing contributions to Wikipedia, so I can understand my overall impact this year and share with friends.

How does this work fit into the Wikimedia Foundation's Annual Plan?

Wiki Experiences 3: Consumer experience (Reading & Media)

Under the Wikimedia Foundation's Infrastructure Goal, and within the group of objectives focused on Wiki Experiences, is an objective related to improving the experience of consumers:

Wiki Experiences 3: Consumer experience Objective: A new generation of consumers arrives at Wikipedia to discover a preferred destination for discovering, engaging, and building a lasting connection with encyclopedic content.

Wiki Experiences 3.2 Key Result

Under the consumer experience objective, is one key result focused on integrated donor experiences: Wiki Experiences 3.2 (WE3.2) Key Result: 50% increase in the number of donations via touch points outside of the annual banner and email appeals per platform.

Several Wikimedia Foundation teams are committed to working on projects under the WE3.2 Key Result, you can see all active projects on the Reader and Donor Experiences page.

iOS team hypothesis	Timeline	Phabricator epic
Wiki Experiences 3.2.5 If we create a Wikipedia Year in Review experiment in the Wikipedia app, to allow users to see and share personalized data about their reading, editing, and donation habits, we will see 2% of viewers donate on iOS as a result of this feature, 5% click share and, 65% of users rating the feature neutral or satisfactory.	October 1 - December 31, 2024	T371946

Wikimedia Foundation teams are approaching annual planning more iteratively this year, so rather than committing to larger year-long projects, our first hypothesis is fairly narrow in scope. This should allow us to deliver value in smaller increments throughout the year, while also ensuring we have the flexibility to pivot as we learn. So WE3.2.5 is just a first step. We will evaluate results and decide next steps after seeing results from this initial hypothesis.

Feature Requirements

Share at least 2 personalized insights on App usage or editing for logged-in app users
Share at least 1 insight collective insight for each category: reading trends, editing trends, foundation Impact & Fundraising effectiveness for logged-in and logged-out users
Allow users to share individual slides
Ask satisfaction at the end, and collect text feedback
Present opportunity to donate for countries where we accept donations from
Ask users to opt-in to more robust tracking for a future iteration in the App
Prompt logged-out users to create an account, or log in
Where data comes from should be transparent to the user (link to media.wiki page). It should be clear that reading data is only from a user's App usage, and does not contain reading insights from Web
Edits should be confined to Wikipedia edits, and include those made outside the app
API calls for Editing Data should not cause strain on core APIs
Users should be able to view their Wikipedia in Review multiple times
Allow users to dismiss/clear their Wikipedia in Review
Feature is fully accessible using dynamic type and VoiceOver

Audience

Version 1 of this experiment will go to app readers in Mexico and Italy.
If results from Version 1 are promising, we plan to release a Version 2 to all users of the iOS app.

How will we know we are successful?

Validation

2% of Year in Review viewers donate on iOS as a result of this feature
5% click share on one or more screen
65% of users rate the feature neutral or satisfactory
Users that were shown personalized slides rate the feature Satisfactory at a higher rate (more than 10 percentage points higher) than users who were only shown collective slides.

Guardrails

No more than 2% of feedback includes reports of NSFW, Vandalism or Offensive articles referenced in Year in Review.
Completion rate: 20% of users who launch Year in Review view the last slide

Curiosities

How the average donation amount compare for donations through Year in Review vs Banners?
What % of Year in Review donors were repeat donors vs new donors?
On which slide are users most likely to drop off (close feature using done)?
Did account creations increase as a result of this feature?
Did App installs increase after release of the feature?

Version 1 Designs

Our initial designs for V1 of the feature are below. Please note, the illustrations on these screens are placeholders and will likely be replaced. We hope to continue evolving the feature, and adding slides.

The Year in Review item is added to the profile menu, accessible at the top right of 'Explore' or in the article.
Users will see a teaser for the Year in Review when they open the app
The introduction screen provides even more information about the experience. It contains a ‘Learn more’ link that leads to an FAQ page.
This personalized slide features the amount of articles the user has read.
The collective slide displays Wikipedia’s page views on the iOS app.
The number of times the user has edited Wikipedia.
How many times Wikipedia has been edited per minute.
If the user has donated in the iOS app before, they see this slide.
Users who haven’t donated before see this a variant of this slide with different copy and a call to action (Learn more about our work).
The ‘Learn more about our work’ link leads to MediaWiki page with more information about WMF’s mission and an option to donate.
The satisfaction survey is shown after users tap ‘Finish’ or ‘Done’ on the last slide
A toast confirms that the feedback’s been submitted.
A login prompt to improve the user's experience is shown when they’re not logged in.
A ‘Year in Review’ item is added within Settings.
The feature can be disabled.
Share step 1: The current slide can be shared via the ‘Share’ button at the bottom. It triggers the native share dialog from iOS.
Share step 2: The image that is shared feature’s the current slide’s content, as well as the hashtag and the username (when logged in).
Share step 3: In addition to the image, editable text is pre-populated.

Version 2 designs

Explore Wikipedia in Review (small)
Explore Wikipedia in Review (big)
We spent 2.4 billion hours reading
English Wikipedia’s most popular articles
We had over 31.6 million reading lists
Editors made nearly 82 million changes this year
3.4 billion bytes added
Wikipedia was available in more than 300 languages
We have viewed Wikipedia articles 1.5 billion times
Editors on the iOS app made more than 452,257 edits
Wikipedia was edited 342 times per minute
Your generosity helped keep Wikipedia thriving
0 ads served on Wikipedia
Explore your Wikipedia Year in Review
Explore your Wikipedia Year in Review
You read 350 (variable) articles this year
You read most on Wednesdays (variable)
You saved 25 (variable) articles
You edited Wikipedia 150 (variable) times
Your edits have been viewed more than XXX (variable) times recently
Red dot on profile menu informs users where to find the Year in Review again after they tap “close” or exit the core flow
Profile entry point (logged in)
In the profile menu, logged-out users are prompted to create an account
Prompt to login
Creates an account/loggs in

How to follow along

We have created T371946 Wikipedia in Review on iOS App as our Phabricator epic. We encourage your collaboration there or on our Talk Page. There will also be periodic updates to this page as we make progress on the experiment.

How data is used for Year in Review

In preparation for the Year in Review, we added two changes to the App’s local storage. These changes align with the existing privacy policy allowance to use local data storage to "provide you with a customizable experience".

We updated the app’s History tab, so that if users view the same article multiple times, they are recorded multiple times within the app’s local storage. T370216. This will allow us to let users know what their top-read article is. Users can still clear their history at any time.
We added local tracking for donations made in the App. If you make an donation within the iOS App, the app will store your donation amount & status on your device. This can be cleared at any time from settings with “Clear donation history.” No additional personal details or payment information is stored. T376219

Users will see a link within the feature to "Learn more", which will lead to a How your data is used page with information on how the insights are calculated for each slide, and how they can clear their data.

Updates

February 2025

We conducted an analysis after the first 15 days of the second version being released widely to logged-in and logged-out users in all languages, measuring our key results, guardrails, and key results.
- KR 1.1 2% of Year in Review those who engaged with the feature donate on iOS as a result of this feature
  - Not met: 0.2% of users who viewed 2+ slides of Year in Review donated through the feature.
  - Our goal was to outperform our app fundraising banner’s donation/impression rate. Our actual came in lower than the goal, but we are still pleased with this result because the feature had a much more subtle donation call to action, and the donations received from the feature amount to more than a 50% in number of non-banner donations from this time period.
- KR 1.2 5% of users opt to learn more about the Wikimedia Foundation
  - Not met: 0.5% of engaged users who viewed the final foundation-related slides opted to learn more about the Wikimedia Foundation by following the “Learn more about our work” link.
  - We heard a few requests in the survey for more information on what the Foundation does, and how donations are used. For next year, it might be wise to incorporate that information directly into the slides, as users are unlikely to follow a link.
- KR 1.3 5% of users who engaged click share on one or more screen
  - Not met: 2.9% of engaged users initiated a Share action.
  - We heard feedback in the survey that the sharing experience could be improved by allowing users to share a summary of all the slides / information instead of each individual slide.
- KR 1.4 65% of users who engaged rate the feature neutral or satisfactory
  - Met: 47% of engaged users took the survey, and the vast majority, 97.4% rated it as either neutral, satisfied, or very satisfied.
  - 54.7% rated the feature as very satisfied, 32.2% rated as satisfied. Only 1.2% of users rated it as very unsatisfied.
  - Open-text feedback in the survey was generally positive: 15% of feedback were complimenting the feature or asked it to continue, and 60% of respondents asked for more statistics about themselves.
  - Less than 1% of responses mentioned concerns with the privacy and data-collection of the feature.
- KR 1.5 Users that were shown personalized slides rate the feature Satisfactory at a higher rate (more than 10 percentage points higher) than users who were only shown collective slides.
  - Our hypothesis for this KR was that users would be more satisfied with a more personalized experience. This was somewhat supported in the overall satisfaction scores: Users who saw at least 1 personalized slide only rated the feature Neutral/Satisfied at 1.5 percentage points higher than users who did not see any personalized slides.
  - Not seeing any personalized slides, and only seeing statistics about the Wikimedia movement, or the iOS app did not cause a significantly worse satisfaction in our users.
  - However, of those users who gave us survey feedback, 60% asked for more personalization, either generally or with specific ideas. This shows a hunger for more personalized slides.
- GR 1.1 No more than 2% of feedback includes reports of NSFW, Vandalism or Offensive articles referenced in Year in Review.
  - Actual: We did not see any reports of NSFW, vandalism or offensive articles surfaced in the Year in Review mentioned in survey feedback.
  - The risk for this was relatively low; we only referenced users’ article titles for articles they’ve Saved to Reading Lists. We saw many requests for more information about the articles they read in the survey. If we add images or more information about articles they read throughout the year next year, we should still be vigilant for reports of this type.
- GR 1.2 No more than 5% of users who engaged complain about fundraising components from survey, Donor relations, or support email
  - Actual: In the open-text answers to the survey, only 0.5% of the respondents expressed negative sentiment about fundraising aspects.
  - The most common complaint was that the feature felt like an Ad for Wikipedia, and they asked for it to be more personalized.
- CR 1.2 How does the average donation amount compare for donations through Year in Review vs Banners?
  - The average donation from Year in Review was 46.6% higher than compared to recent iOS Banner Campaigns in the Netherlands and France ($19.71 compared to $13.44).
- CR 1.3 What % of Year in Review donors were repeat donors vs new donors?
  - 54.4% of donors who came in through Year in Review were new donors who have never donated to the Wikimedia Foundation in the past.
- CR 1.4 On which slide are users most likely to drop off (close feature using done)?
  - The most-closed slide for non-donors was the final slide “0 ads served on Wikipedia”, which makes sense as it was one of the ways to exit the feature.
  - After that, the most commonly closed slide was the first one with collective or personalized information about how much we’ve read on Wikipedia.
- CR 1.5 Did account creations increase as a result of this feature? How many of these accounts constructively activated?
  - If users were logged-out, and entered the feature through the announcement, they saw a collective version of the Year in Review, with stats about Wikipedia’s overall impact. At the conclusion of the feature, we prompted them to log in or create an account for their personalized version. If they did, they were presented immediately with a personalized version,
  - If users were logged-out and tried to open the feature through its permanent location in the profile menu, we asked them to log in or create an account to view the feature.
  - We wanted to require an account to view the personalized version of teh feature for a few reasons
    - We can show personalized editing stats for logged-in users
    - We can make future versions of the Year in Review more robust and possibly cross-platform if users are logged-in
  - Because of these prompts to log in and create an account, we hypothesized that we’d see an increase in new accounts created, and that some of those new account holders would edit!
  - We saw a 74% increase in new accounts compared to the baseline. Accounts created from Year in Review constructively activated at a lower rate than overall accounts, at 3.0% compared to our baseline of 4.9%.
  - There were 368 new account holders who were constructively activated due to the feature!
- CR 1.6 Did App installs increase after release of the feature?
  - The sharing functionality of Year in Review included a link to install the app, and we also ran our first Apple App Store event to promote the feature. We expected to see a slight increase in app installs due to this feature.
  - We saw an 8.3% increase in new app installs compared to our baseline!
  - From the Apple App Store promotional event, we had 131,000 impressions on the event, and 871 downloads of the App.
Lessons learned from this feature
- Our hypothesis was: If we make improvements to the personalised and collective content of the iOS apps’ Year in Review, and scale its availability, we will learn if this is an effective fundraising method.
  - We consider this hypothesis supported: the Year in Review succeeded in being a subtle but effective fundraising method: we met our goal for the number of donations from the feature. 54% donors who gave through Year in Review were first-time donors, and we saw a higher average donation from the feature as compared to banners. 97.4% of survey respondents rated the feature as Neutral or Satisfactory, and only a small handful of users (0.5%) complained about the fundraising aspects of the feature.
- We learned a lot from this feature that will be applied to future iterations
  - Above all else, users want more personalization, and statistics about their reading and editing habits on Wikipedia. Of the users who were disappointed in the feature, most of them had expected it to be more personalized. 60% of survey feedback mentioned wanting more personalization.
  - What kind of personalization do users want?
    - The most requested was the user's reading history categorized by topic, theme, category, or genre.
    - Second most requested was which articles users have visited, particularly which articles they visited multiple times. Users would also be interested to see which articles they visited that are more niche or lesser read.
    - Users want to review a list of all articles they read over the year: we could create a reading list with all of the articles they read that year, and offer them an opportunity to save it to their reading lists.
    - Users want collective stats and personalized stats intermingled. They like seeing the collective impact, and want to be compared to the collective. Next year we should not remove collective slides for logged-in users.
    - Some users asked for a summary of their past donations.
    - Visualize and tell the story of their knowledge journeys: rabbit holes, paths of knowledge followed on the same day, journeys of links clicked.
  - It’s important to continue communicating where the data comes from, and its limitations (such as statistics that do not cover the full year, or why it’s not able to be cross-platform).
    - Some users were confused why the review did not include their reading from other devices or Web.
    - 0.76% of all free-text survey answers expressed concern about Wikipedia tracking their reading or didn't want more tracking. We should surface more prominently the option to opt-out of Year in Review (in settings).
  - Adding more personalized readership statistics would have a higher impact for app users than adding more personalized editing statistics. 3% of survey feedback asked for more editing statistics, or corrections to editing statistics. This is compared to 20% of responses asking for their reading history by topic, and 9% asking for more information on specific articles read and rabbit holes followed.
  - Requiring account creation to see personalized slides did not have a negative impact on the feature overall, and had a positive impact on account creation and activation! 97.4% of users who did not see any personalized slides still rated the feature Neutral or Satisfactory. We saw a 74% increase in new accounts, and 3% of those account-holders were constructively activated (368 new editors due to this feature!).

January 2025

We released the Personalized Wikipedia Year in Review to production! (T379406) This included a version for logged out users, and new personalized slides for logged-in users. We’re monitoring results

January 2025 Special Update

We have released a second version of the Year in Review to Beta testers. It can be accessed by downloading the most recent app version from Test Flight.

A video walk through of the second version of Year in Review

What's new in Version 2?
- The feature is widely available across languages & countries.
- We added 3 new personalized insights: which day of the week you read the most, articles saved to reading lists, and views on edits recently. More details can be found on the How your data is used page.
- Illustrations have been updated, and match the Wikipedia Year in Review web page.
- Only logged-in users are eligible for the personalized insights. Logged-out users will see a collective year in review. Logged-out users will see a prompt to log-in for personalized insights.

December 2024

The first version of a Personalized Year in Review went to production, released to all eligible users in Italy and Mexico, with either Italian, Spanish, or English as their primary language in the app.
We conducted an initial analysis after the first 15 days of the experiment, measuring our leading indicators, guardrails, and key results.
- Our first iteration of Year in Review showed 0.3% of engaged users donating, 2.3% of users clicked share, and 95% of users rated the feature neutral or satisfactory.
- Goal: At least 20% of users who launch Year in Review view the last slide
  - Actual: We saw a completion rate of 48.1%
- Goal: 2% of Year in Review those who engaged with the feature donate on iOS as a result of this feature
  - Actual: 0.3% of users who engaged with the feature donated on iOS through feature
- Goal: 5% of users who engaged click share on one or more screen
  - Actual: 2.3% of users clicked Share on one or more screens
- Goal: 65% of users who engaged rate the feature neutral or satisfactory
  - Actual: 95.3% of users who submitted feedback rated the feature Neutral or Satisfactory
- Overall learnings:
  - Users are satisfied with the feature overall, but interested in more personalization. 95.3% of viewers who submitted feedback rated the feature Neutral or Satisfactory. In examining the text feedback, the unsatisfactory answers come from users who want more personalized information about themselves. The most popular request was for insights on their reading history by theme or topic.
  - The donation rate is lower than our feature-specific goal, but the total donations received shows we can achieve movement towards the overall 3.2 key result goal. We set our donation rate goal at 2%, higher than our typical donation/impression rate for fundraising banners on iOS in 2023. We surpassed our donation goal, and received a total count of donations which amounts to a 50% increase in app donations for Italy and Mexico within the iOS App during a 15 day period.
  - Viewers did not complain about fundraising aspects: of the 4.7% of survey responses rated the feature as unsatisfactory, 0 mentioned the donation buttons or donation options.
  - Viewers are interested in both learning more about the foundation, and donating. We had a comparable engagement rate for donations (0.3% of users engaged), as we did for clicking on the “learn more” links that led to the MediaWiki page about the Wikimedia Foundation (0.2% of users who saw the slides).
We are continuing development on a second, improved version of the personalized year in review that will include more personalization, and collective insights about English Wikipedia. Our work is coordinated in this epic: T379406

November 2024 Special Update

The first version of the personalized Wikipedia Year in Review has been released to Beta testers in Italy and Mexico, on Italian, Spanish, and English Wikipedia. Users who have reading history on their device, or logged-in users with editing history are eligible to see their Year in Review. Find out more about how insights are calculated on the How your data is used page.
We value any feedback from those who have been able to try the feature, it will inform our iterations and improvements moving forward.

Video demonstration of Year in Review V1 on the Wikipedia iOS App

Our work for the first version of Year in Review was coordinated on this Epic: T376037.
We began working on iterative improvements for a second version of Year in Review, coordinated on this Epic: T379406

October 2024

We began development on Year in Review, completing data set-up and configuration. Our work can be followed from this Epic: T371946
We updated local storage tracking for this years & future Year in Review features
- Updating the history tab so that it can contain the same article multiple times, so that we can show users their most-read articles T370216
- Added local tracking for Native ApplePay donations, so that we can thank donors who have donated through the app T372343