Growth/Personalized first day/Welcome survey
This page describes the Growth team's work on the "welcome survey" project, which is a specific project under the larger "Personalized first day" initiative. This page contains major assets, designs, and decisions. Most incremental updates on progress will be posted on the general Growth team updates page, with some large or detailed updates posted here.
On November 20, 2018, the welcome survey was deployed to treatment groups in Czech and Korean Wikipedias, and initial data was collected and analyzed in this initial report in December 2018.
By March 2019, we had concluded that the design called "Variation A" is best, and that is now shown to all new users in Czech and Korean Wikipedias.
We are in the process of investigating which survey variant is best in Vietnamese Wikipedia.
- 1 Current status
- 2 Summary
- 3 Comparative review
- 4 Design
- 5 Questions, responses, and other text
- 6 User testing
- 7 Analysis and experiments
- 2018-11-20: deployed the welcome survey to Czech and Korean Wikipedias, with half of users getting Variation A and half in a control group.
- 2018-12-22: published initial report on survey results.
- 2019-01-16: after determining that receiving Variation A was not depressing activation rate, changed Czech and Korean such that half of users get Variation A and half get Variation C.
- 2019-01-24: deployed the welcome survey to Vietnamese Wikipedias, with half of users getting Variation A and half getting Variation C.
- 2019-02-27: switch Vietnamese Wikipedia so that half of users get Variation A and half get no survey. Meant to investigate concerns about abandonment rate from initial experiment.
- 2019-03-06: after analyzing experiments in Czech and Korean Wikipedias, started to give all users Variation A, because it performed better in all scenarios and all metrics. No experiments running in Czech or Korean Wikipedias.
- 2019-03-21: disabled survey in Vietnamese after seeing that desktop users who receive Variation A have a much higher abandonment rate than desktop users who receive no survey. Still investigating this result.
- 2019-07-12: removed the ability for users to add free-text responses to the survey. We have learned what we needed to learn from the responses we've received so far, and they have become a potential vector for abuse.
- 2019-07-15: deployed the welcome survey to Arabic Wikipedia, with half of users getting Variation A and half getting no survey.
- 2019-08-30: published results showing that welcome survey does not negatively impact Arabic Wikipedia users.
- Next: altering welcome survey questions to gather the information needed for newcomer tasks
We know from the New Editor Experiences research that newcomers arrive at the wikis with a specific goal in mind. If they're not able to accomplish it, they're unlikely to come back. With the welcome survey, we want to ask newcomers about their goals, so that we can give them experiences that help them accomplish their goals. We want to ask new users additional questions during their account creation process, such as their reason for creating an account, what they are trying accomplish, topics that they are interested in, or whether they want to meet a mentor.
To read more about the motivation for this project, see the "Personalized first day" initiative.
To learn how best to design the welcome survey, our team's designer reviewed the way that other platforms (e.g. Reddit, Coursera, Medium) ask initial questions of newcomers. While the experience we want to give newcomers is definitely different than other platforms (we want to give newcomers an optional, lightweight, non-invasive experience), we also recognize that there are best practices we can learn from other software. The comparisons are shown in this slide deck, and the main takeaways are:
- Modals were generally more in use or otherwise a full page overlay for forms.
- No more than 4 questions were asked during onboarding.
- Questions were either progressively disclosed on a single screen or otherwise shown 1 question per step with clear visual indication of the number of steps (eg. via a stepper UI).
- Multi-question forms tended to be skip-able entirely, but with mandatory questions if the user wanted to participate.
- An optional guided tour or further help was often provided afterwards once user was in the app.
- Friendly voice and tone - welcoming, informal language framed as asking for information to help orient the user ("Tell us about yourself", "Get recommendations", etc.)
Our evolving designs can always be found in these clickable mockups, and with additional contextual information in this Phabricator task. Our goal for the design is for newcomers to answer all the questions they want to, and to not depress the number of users who complete their signup process and get back to what they were doing.
We considered three main design approaches for this project. The new account holder lands at each approach immediately after clicking "Create your account" at Special:CreateAccount. And after completing each one, they are returned to the page from which they originally clicked to create their account. Please note that the question wording and text in these mockups does not reflect current drafts -- the mockups are more for the visuals and workflows. Question wording and text will be posted separately.
- Variation B: this option shows users one question at a time, and also removes the surrounding links from the left and top navigation of the wiki for a more streamlined experience. This is inspired by the interface for the Content Translation tool. We decided not to pursue this variation at all, since it is more work than Variation C, which we like better.
The team built and deployed Variation A in November 2018, and simultaneously built Variation C. We will be deploying Variation C during January 2019.
Questions, responses, and other text
Original set of questions
The questions that this feature asks to newcomers, along with their specific wording and the other text in the feature, is really important to getting useful data. The team worked on these things with help from several other people experienced with surveys at WMF. All the questions will be optional, and will be translated into the languages of the wikis on which the welcome survey is deployed. Below are the questions we want to ask, and we are still working on the wording. To see the current text of the feature, check out the current mockups.
- Why did you create your account today?
- Goal: if we understand what a newcomer is trying to accomplish, we may be able to show them help materials that help them accomplish it.
- Response options:
- To fix a typo or error in a Wikipedia article
- To add information to a Wikipedia article
- To create a new Wikipedia article
- To read Wikipedia
- Other (please describe)
- Have you ever edited Wikipedia?
- Goal: we want to know how many people creating accounts may have already made IP edits. This would also allow us to refrain from giving too much help material to people who know what they're doing already.
- Response options:
- Yes, many times
- Yes, once or twice
- No, I didn't know I could edit Wikipedia
- No, for other reasons
- I don't remember
- People can edit Wikipedia articles on topics they care about. We've listed a few topics below that are popular for editing. Select some topics that you may wish to edit.
- Goal: it may be possible in the future to connect newcomers with experienced editors who share their interests, or just give them recommendations on articles to work on.
- Response options (see this update for explanation of how these were determined):
- As checkboxes: Arts, Science, Geography, History, Music, Sports, Literature, Religion, Popular culture
- Behind typeahead: Entertainment, Food and drink, Biography, Military, Economics, Technology, Film, Philosophy, Business, Politics, Government, Engineering, Crafts and hobbies, Games, Health, Social science, Transportation, Education
- It will also possible for users to add their own topics.
- Email address
- Goal: Only if the user did not supply it on Special:CreateAccount. Email is important for engaging new editors and allowing them to recover their accounts, so we want to ask twice.
- We are considering starting a program for more experienced editors to help newer users with editing. Are you interested in being contacted to get help with editing?
- Goal: We want to learn whether newcomers feel that they need human-to-human help.
Update for Newcomer Tasks in September 2019
The original purpose of the welcome survey is to use the survey responses to personalize a newcomer's experience so they get what they need on their first day. The first feature that we will be personalizing is the newcomer tasks feature. Welcome survey questions and responses will change to meet the needs of that feature. Specifically, these are the planned changes:
- We will alter the responses in the first question ("Why did you create your account today?") to include response options that are common in the free-text "Other" field:
- "To add a photo or image to a Wikipedia article": this is one of the most common things that newcomers try to do.
- "I'm participating in a program, class, or event": many newcomers come through programs, and may want the ability to provide personalized content based on the program from which they are arriving.
- We will collapse two of the other responses in the first question into one that reads "To make changes to an existing Wikipedia article". We're collapsing the following two responses into that one response to make room for the new responses above:
- "To fix a typo or error in a Wikipedia article"
- "To add information to a Wikipedia article"
- We will remove the question asking the user's topics of interest from the survey entirely. This is because we'll be asking that question in the newcomer tasks feature on the homepage instead.
- We will add a question asking users to indicate the languages they read or write. This question may help us encourage certain newcomers to use the Content Translation tool.
During the week of October 22, 2018, we used usertesting.com to conduct six tests of our Variation A mockups with internet users unaffiliated with the Wikimedia movement. In these tests, respondents are compensated for trying out the mockups, speaking aloud on what they observe, and answering questions about the experience. As our team's designer described on the Phabricator task, the goals of this testing were:
- Identify improvements to Understanding of the survey (update copy depending on users' comprehension of the instructions and questions on the form).
- Identify improvements to the Usability of the survey (check whether users are able to correctly input and submit their intended responses).
- Do users know how the visual design and layout works? (Eg., do they understand they should be selecting one answer from a radio button group)
- Are users able to navigate through the form?
- Any feedback missing that could help users complete the form?
- Gauge user Reactions to the survey and Expectations of how the information will be used.
Summary of findings
- Survey was clearly optional, and seen as short, low-effort, and non-intrusive to complete
- Generally seen as data capture for research (though a few testers thought more specific explanation about how responses may be used might make them more inclined to fill it in)
- "Mentorship program" was the one aspect not clearly understood, with about half of participants mistakenly assuming they would be providing the help rather than receiving it.
- A couple of users had misgivings about providing email and feared it may be misused for marketing or given to 3rd parties (also as they expected it to be mandatory for account creation if it was needed for recovery)
- Users liked the post-submission message with more information about "Getting started with editing"
- Several users assumed that upon completing the welcome survey, they would be directed to their "dashboard" to get started with editing.
- Add a full Thanks message after survey completion
- Rephrase “Mentor” checkbox question and clarify expectations for how users may be contacted if they select the mentor checkbox
- Revise phrasing of Q3 (Wordsmithing so that the the second sentence “We've listed a few below popular for editing” is more easily understood.)
- Add a tooltip beside or assistive text under the “Add more topics” field for no-js users explaining they can comma-separate multiple topics
- Add more information about how the email is used
- Potentially be extra clear by placing a tooltip with the following message (taken from Help:Account_management): “If you choose to give an email address, other users will be able to contact you by email. This feature is anonymous—the user who emails you will not know your email address. You don't have to give your email address if you don't want to (but doing so is required to reset your password if you forget it)”
Analysis and experiments
There are two kinds of analysis we'll be doing with respect to this feature. To read about those plans in-depth, see this page containing the "experiment plan".
Analysis of the responses to the survey
The main objective of the survey is to collect information about what new editors are trying to accomplish, so we can figure out if it's possible to personalize their experiences based on their responses. We are doing straightforward analysis on responses, broken out by elements like platform (mobile vs. desktop) and context from which the account was created (homepage, reading, editing).
The initial report on survey responses can be found here. Community members should feel free to translate this report into their languages. In-depth reporting will be created by March 2019. Some topline results:
- The survey has high response rates: 67% in Czech and 62% in Korean.
- Many people create accounts just to read articles -- a potential opportunity to engage new editors.
- High numbers of people indicate they are interested in being contacted for help: 36% in Czech and 53% in Korean.
Analysis of the impact of the survey on activation
Although we would not have intended it, it is possible that the welcome survey could depress the number of newcomers who make edits. Community members first brought up this concern, saying that by making the sign-up process longer, and keeping newcomers away from their objectives, we could cause them to leave the site. Therefore, we conducted an A/B test to find out whether being presented with the survey decreases activation rate (the rate at which new users make their first edit).
From November 19 to January 15, half of newcomers in Czech and Korean Wikipedia were given Variation A of the survey, and half were given no survey. Our experiment results are published here, and show that having the survey does not significantly decrease the rate at which newcomers make their first edit. Based on this analysis, we also tested Variation A against Variation C to see if Variation C has increased response rates (see next section). In August 2019, we ran this same test on Arabic Wikipedia, and again found that the welcome survey had no negative impact on new users.
Analysis of the two survey variants
After seeing that Variation A did not cause a decrease in activation rate in Czech and Korean Wikipedias, we moved on to comparing Variation A to Variation C in those wikis. During the month of February 2019, half of newcomers received each variant. After comparing the variants on activation rate, abandonment rate, and response rate, we saw that Variation A performs better or equal to Variation C on all languages and all platforms. We therefore ended the experiments in Czech and Korean Wikipedias and started using Variation A with all newcomers starting on March 6, 2019. In-depth results are forthcoming.