Edit Review Improvements/New filters for edit review/Research

From mediawiki.org

This page provides an overview of the iterative user experience research performed for the New features for edit review. Each step of the research is outlined and the corresponding research reports are linked. Our goal is to give our users a more in-depth look into how user research is done, how it informs product direction and decisions, and what product possibilities have been considered thus far.


The New filters for edit review became available on Recent Changes as a beta feature on May 2017, and as default in October 2017. The new filters are intended to help editors to find their edits of interest to review. Some filters provided were not available before, and some are based on new machine learning capabilities.

We wanted to provide users with an experience that was intuitive and at the same time more powerful so that use cases that were not well supported in the past would be better addressed. Working in this area presented unique challenges, where research was key to learn about our users and how our proposed solutions worked for them.

Feel free to peruse the details in the research summaries and linked research reports below.

Key research questions and findings[edit]

  • Question 1: Did participants find utility in the new filters and functionalities for edit review on Recent Changes?
    • What we learned: Yes, overall the participants had very positive reactions to the new filters and functionalities.
    • How it informed the project decisions: Due to the positive feedback, we tested additional filters and functionalities, and made iterative tweaks on existing edit review elements based on user feedback.
  • Question 2: To what extent did participants discover, understand, and intend to use the new filters/functionalities on Recent Changes?
    • What we learned: Each round of testing yielded different results, but at each stage there was feedback to evaluate and improvements to be made to facilitate discoverability, understanding, and, as a direct correlation to that, how much participants intended to try the beta in their normal workflows on the wikis.
    • How it informed the project decisions: At each stage in the research process, we were constantly adding, removing, and editing both new and existing elements of the Recent Changes page based on our evaluation of the participant sessions and also their direct feedback/recommendations.
  • Question 3: How did the success of the filters/functionalities translate to the Watchlist page?
    • What we learned: The participant cohort did not express much interest in bringing these functionalities to Watchlist and indicated that they would not find utility in this notion.
    • How it informed the project decisions: For the moment, this is not moving forward, but further investigation is warranted as the participant cohort were primarily users of Recent Changes, not Watchlist. The new filters on the Watchlist will not be exposed by default, remaining as a beta feature as more feedback is collected.


  • Our macro level goal is to make the wikis and its tools more intuitive and useful for everyone, but for these sessions, we limited our participants list to those who were (relatively heavier) users of edit review pages. It ended up that *most* of our participants worked primarily on English Wikipedia, but we did have representation from other languages, other language wikipedias and Wikiversity.
  • Change can be difficult, regardless of utility, intention, or results. Though any particular change won't always make everyone delighted, we generally aim to make as many product decisions as possible based on our users' feedback and what will help the most users.
  • Usability on a relatively wide scope is a large undertaking, especially when generative research elements are mixed in as well. Our goal is to keep both details and overall look, feel and utility in mind as we make product decisions.
  • We need more participants in general! If you're interested in the Contributors space and would like to participate in upcoming research studies, please contact Daisy Chen with your areas of interest and we will keep you in the loop!

What’s next[edit]

  • Although the results have been encouraging, our wikis support many different usecases and we want to hear from you how the new filters work for your activities. Feel free to share your experience on the project talk page.
  • Are you interested in the Contributors space? Would you like to participate in upcoming research studies and provide your feedback? If so, please contact Daisy Chen with your areas of interest and we will keep you in the loop!

Recent Changes Filters Usability Round 1 2016.09-11[edit]

Research Questions[edit]

  • Concept and Intent
    • Do users understand what the filter categories represent, specifically the Quality, Intent, and Experience filters?
    • Do subjects understand what the different ORES (Objective Revision Evaluation Service) levels mean?
    • Do users understand the difference between filtering and highlighting, and how they might use these singly or in combination?
  • Usability
    • Do users notice and/or understand the default filters?
    • Are users able to find the correct filters and use them in combination effectively?
    • Do users find and learn to effectively use the highlighting to isolate targets?
    • Can they manage and interpret the highlight colors?
    • Do they find the “ignore highlights” button and understand its use?[1]
  • Effectiveness
    • How do users feel the new tools will affect their work and workflows?
    • What are users’ reactions regarding the new interface and its functionalities?
    • Do users feel any differently about the new interface as compared to the current Recent changes page?
      • Do they see that the new interface as being functionally neutral, inferior or superior?

Curious about the more nitty-gritty details on the research set-up? Take a look at the 2016 Recent Changes Filters research.


Overall, the new Recent Changes filters features were well-received. Users generally find that the prototype Recent Changes are improvements over the current and a majority think they would incorporate these features into their work.

Though features can still use some clarification, refinement, and ongoing iteration/improvement, using them is pretty intuitive overall and even when there are user hesitations, is a learnable process.

Over the course of this first round of usability testing, two prototypes were used with the second prototype building on findings from the first prototype.

Next Steps[edit]

Discuss existing issues and corresponding recommendations for improvement. Prototype 2 is recommended to go to beta for further user feedback and iteration.

Research Report[edit]

Presentation of findings.

Recent Changes Filters Usability Round 2 2017.05-06[edit]

Research Questions[edit]

After introducing the new filters for edit review to positive feedback, the Collaboration team’s next iteration aimed to include support for advanced functions.

In these sessions, we utilized a prototype testing environment and lead users through a series of tasks. Through completing said tasks, we are able to evaluate if research participants were aware of the functionalities, found them intuitive to use, and whether they found utility in them.

  • What elements of the extended filters are most and least useful for users?
  • Do users notice the new capabilities?
  • If they find them, are they easy to understand and use?

New page elements/functionalities

  • Namespaces, tag, users, and categories filters
  • Filter set bookmarking
  • Display options (size and pagination of results)
  • Navigating and updating results

Read more details on the research planning and protocol at the 2017 Recent Changes Extended Filters research.


Generally, participants had positive responses to the new filters and functionalities. To different degrees/varying use cases, most of these new features have audiences ready to wield these tools.

However, there are some areas in the prototype that can benefit from additional clarifications, UI tweaks, and workflow considerations. The details on user recommendations can be found in the research report.

Next Steps[edit]

Discuss existing issues and corresponding recommendations for improvement.

If possible, integrate changes into beta feature to allow time for users to experience them. After period of use, users of the beta should be prompted for information on how much the feature was utilized day-to-day, what specific elements/functionalities they used, and their feedback and recommendations for how the beta has worked for them and where areas for improvement remain.

Research Report[edit]

Presentation of findings.

Recent Changes Filters: Usability Round 3 and Beta Feature Satisfaction Testing 2017.07-08[edit]

Research Questions[edit]

The new filters for edit review have been out as a beta feature on various wikis. Our goal is to learn from users directly how the experience of using the new filters over a period of time has been for them.

Since the initial rounds of generative and usability testing, the team made some tweaks from past usability testing and feature requests that were then applied to a new prototype, named ‘integrated’ filters, along with a set of filters made specifically for the functionality on watchlist. In this round of testing, we evaluated the usability of the latest iteration, the applicability of its watchlist mirror, and the live updates feature in conjunction with the beta satisfaction testing.

Rating users’ satisfaction with the Recent Changes beta
  • What do users think of the filters? Have their opinions changed over time?
  • Have they put the filters to use? If so, how? What are some of these uses?
  • Are the filters supporting their existing activities and workflows?
  • Are they creating new ‘existing activities/workflows’ as a result of having the additional capabilities?
  • How do they feel about the beta Recent Changes vs current Recent Changes?
  • More specifically, how do the new set of filters compare with the current set?
  • What does the page need to contain for it to reach peak effectiveness for the users?
Integrated filters/live updates usability
  • How are discoverability of the filters and general navigation for users? Overall, has there been improvement?
  • What reactions do users have to the live updates button and pagination options?
  • Do users know about / recall / use the resources listed in the related links section?
Watchlist actions usability
  • What actions do users typically take on their watchlists?
  • What filters would be useful? How do they feel about the watchlist filter set on the prototype? What filters are still missing?
  • Is the purpose of the watchlist-related filters provided in the Watchlist page understood, discoverable, and useful for watchlist users.
  • Do users create the same filter sets in the Watchlist they would for Recent Changes?
  • Are users able to identify which pages require their attention (are not visited) at a glance and why the distinction is made? Is it clear for users how to mark some/all of them as visited?

Read more research project details at the 2017 Recent Changes Beta Satisfaction and Integrated Filters research.


User Satisfaction with the new filters beta

Overall, the users feel that the beta feature provides a cleaner interface and a better experience than the current RC page. Though not without caveats; there are user suggestions for improvement, and adoption and use of the Content Quality and User Intent filter categories does not appear widespread. Improvements should be made to clarify the filters’ function, means of usage, and reliability, and other user concerns and suggestions should be evaluated and addressed.

The primary applications of the filters are related to vandalism patrol, and users are generally able to achieve their intended goals using the beta page.

Integrated filters/live updates usability

Discoverability, general navigation, pagination

  • Testing revealed no issues involving the discoverability of the dropdown menu
  • Though participants successfully complete tasks involving ‘advanced’ filters (users, tags, namespaces, categories), their initial discoverability on the page is still lacking
  • Pagination does not appear to cause much confusion for users even when testing an atypical use case (sorting oldest first), although most do not process the need for that functionality on the page.

Live updates

  • All 5 participants had overall positive reactions to the introduction of this feature, with some minor concerns and suggestions for improvement.

Related links

  • All 5 participants use the options provided in the related links to some degree.
  • Mixed reaction as to whether the section should be hidden/collapsible.
Watchlist actions usability

Among this cohort, the number of articles in their watchlists varied (along with the reasons for adding them) but their use of watchlist is rather consistent; they don’t use the page much if at all and if they do, it is not rigorous and more of a casual scanning for new activity. The recruiting of this cohort, however, was based on their use of Recent Changes page, so they may not be representative of the opinions of heavy watchlist users.

A majority of participants indicate that they do not and would not use filters on their watchlist page, consistent with responses when presented with the hypothetical posed during beta satisfaction testing.

All participants understood the bold/filled bullet represented a page with new activity/not yet visited, and it's an expectation that clicking on any part of the entry (diff, hist, or article link) should mark the entry as ‘read’.

Next Steps[edit]

Discuss existing issues and corresponding recommendations for improvement (this will always be a recurring element of our work).

Discuss the utility of further examining the idea of bringing the edit review filters to pages like watchlist, but with more seasoned/heavier users of watchlist pages specifically.

Research Report[edit]

Presentation of findings.


  1. That feature has been removed after first iteration