Growth/Personalized first day/勧誘メール
|
Engagement emails
Use email as a vector to re-engage newcomers with information they need
|
このページで解説する「ようこそメール」プロジェクトの作業は、Growthチームが取り組む「個人化した初日」という大枠のイニシアチブ配下の固有のプロジェクトです。このページでは主要なアセット、設計、意思決定について述べます。進捗状況で増えた更新のほとんどは一般向けの Growthチームの更新ページに、そしてこのページには特定の大規模または詳細な更新をそれぞれ掲載します。
利用者にメールを送信する件は、重要な要素をいくつも検討する必要があることは承知しており、なかでも個人情報の保護とセキュリティの課題は大きいです。当チームではウィキメディア財団の法務部門ならびにセキュリティ部門と連携して、データを確実に適切に取り扱う方法を確認中です。コミュニティの皆さんからは引き続き、この機能を「ウィキメディアンにふさわしい」ものにまとめるように洞察を聞かせてください -- メールの良いところ、便利なところは取り入れ、不快だったりプライバシーを侵害するような側面は避けたいと考えています。
当初の計画では2019年3月末までに最初のメールを発信するはずでしたが、Growth チームではこの先数ヶ月にわたるプロジェクトの優先順位を下げると決めました。現在進行中の2件ヘルプパネルおよび新規参加者ホームページに集中するべきだと判断しています。チームはメール関連で進捗した作業を放棄せず、来年、再開する可能性を残してあります。
現状
- - コミュニティとの協議により新規参加者の初日体験をパーソナルにする件について意見を募集
- - 「勧誘メール」および「新規参加者ホームページ」の取り組みについて方向性を決定
- - このプロジェクトの優先順位を下げ、新規参加者ホームページならびにヘルプパネルを優先する結論に至る。
- - 財団のマーケティング・チーム、資金調達チームと連携し、私たち Growth チームは参加呼びかけのメール送信の準備を始めました。（Engagement emails） [1]
- - 歓迎メール実験の分析結果を公開
要約
調査により、新規参加者はそれぞれのウィキに到着したとき、何か特定のことをやってみたいという気持ちを抱いています。それができなければ去って行き、二度と戻ってこない事例は頻繁にあります。利用者には歓迎アンケート で何がしたくてウィキに来たのか、その人に必要なヘルプを案内して、私たちからその実現を後押しできないか尋ねました。そのヘルプを届ける方法の一つがメール送信です。
さらにまた、ウィキメディア財団にメールを開示した人には、ぜひ編集活動をしましょうと促したいのです。ご寄付の実績がありメールを開示した寄付者は寄付金の受領証が届くと期待しています。ご寄付に感謝を示した私たちから編集のお誘いメールを届けると、もしかしたら関心を持ってくれるかもしれません。
チームが目指すもの：
- 新規参加者に送るメールは、開いて読んだり、リンクをクリックしたくなるものにすること。
- 受け取った人の役に立ち、読んだら行動したくなる呼びかけをすること。
- メールの内容をウィキに置いた内容と結ぶには、新規参加者向けのホームページ を経由すること。
- 活性化と利用者の定着率を増やすこと。
以下の各項目は避けたい事柄です：
- メールが多すぎて迷惑（スパム）。
- 利用者がせっかくオンウィキで 学ぶ機会を奪う。（onwiki＝ウィキ空間で）
- 個人化が過剰で、（訳注＝利用者は）メールに脅かされてしまう。
- 活動が及ぼす影響が低いと示しては、やる気が削がれること。（Demotivate）
- ウィキで最初に開く場所がわかりにくくなること（landing place）。
- Growth チーム内外の他の介入と衝突すること
メールを受信できるか：ここで述べるメールを私たちから利用者に送信するには、先方が検証できるメールアドレスを持っていて、ウィキメディアでメール受信を受け取るよう選択する（オプトイン）ことが前提です。その選択肢を設けるほか、メールアドレスの登録、正しいかどうか検証する方法を簡易化できないか検討中です。利用者にメールアドレスを登録しろと圧力をかけたいのではなく、メールアドレスを登録するとこういう有利な点がある、よいチャンスはたくさんある、どうすればメールアドレスの検証が済むのか、もっと明確にお伝えしたいのです。（Emailability）
Why this idea is prioritized
Once a newcomer leaves the wiki after their first day — whether they do or do not accomplish their goals — it is likely that they are never encouraged to contribute again. This is because all existing communication channels are on wiki, even though the newcomer is no longer on the wiki. We can use email as a proactive communication vector to reconnect with new users off-wiki to help them to make their first edits or to continue making edits.
- Community members were positive in their feedback on the idea.
- Reach users no longer on wiki: emails enable communication with users who may likely no longer be active on wiki (and would therefore not see activity on Talk pages and Notifications).
- Communicate with new users unfamiliar with Talk pages: automated communications exist in the wikis in the form of bots that post on new user talk pages. While in some cases, those postings generate emails, email is not being fully utilized to get the attention of newcomers, who are usually not accustomed to how Talk pages work. Wikipedians who have experience mentoring newcomers have had success using email to communicate, since it is a medium they understand.
- Email campaigns are a proven means of engaging users of software: email is used by virtually all other community contribution platforms (e.g. TripAdvisor, Google Maps/Locals, Wikia) to engage users.
- Newcomers care about impact: many interviews with Wikipedians have surfaced that new editors first become enthusiastic about Wikipedia when they realize that others are reading their contributions. Email can be a way to make users aware of their impact. There have been example of success in showing Impact data in other Wiki projects such as the Outreach dashboard (see comment here).
Comparative review
To learn how best to design engagement emails, our team's designer reviewed the way that other platforms (e.g. TripAdvisor, Wikia, Google) send emails to 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 recorded in detail on this Phabricator task.
- Takeaways from looking at emails
- Show impact via celebration of milestones/achievements of a recipient’s contribution statistics or social impact (x people read your review) to motivate continued/re-engagement
- Appeal to prosocial behavior (your contribution helps others)
- Appeal to social proof encourages emulation (testimonials or examples of others participating)
- Provide personalized recommendations that target specific interests of individuals reduces effort for someone to find out how to participate according to their interests
- Personalized messaging in general is important (e.g., subject lines, copy, when the email is sent)
- Provide onboarding help anticipates issues faced by newer recipients on how to get started, as well as remind them of the value of the sender’s product/service
- Time messages to a relevant or meaningful event or timeframe (for example, an ‘impact’ email for the year sent at the end of the year)
- Subject lines are direct, succinctly encapsulating the email’s intended purpose/message
- Short, text-lean messages
- Personalized, informal language is generally used
- Usually contain a single, main call to action that is specific and prominent
- Requests for specific follow up and feedback are sometimes shown towards end of messages
- Optimize for mobile - commonly by using simple, single column layouts with centered text
- Branding is prominent shown at the top of the email
- Scorecards and ranking illustrations (leaderboards, badges, etc) are often used in impact emails
- Minimal use of graphics, predominantly visual clean messages on white
- Use of photography depicting people used in appeals to social proof and prosocial behavior
- Takeaways from looking at industry metrics and benchmarks
- Welcome emails are generally the best performing in terms of being the most opened (82.57%) and clicked (22.76%) type of email type.
- Triggered emails such as an email when a user expresses interest in being contacted for mentorship perform second-best to Welcome emails, and have a much lower unsubscribe rate.
- Autoresponder emails like a weekly or monthly “Impact” email have lower open and click rates, but also have a much lower unsubscribe rate.
Design
In order for emails to help users accomplish their goals, they need to be tailored to the things that we know about them from the welcome survey and from their editing history. Preliminary designs have been made for six email campaigns, but we will start with three, which are in bold:
- Welcome: a generic orientation email that links to their newcomer homepage.
- Impact: tell newcomers how many pageviews the pages they've edited have had.
- Mentorship: send newcomers an automated message from a mentor they can contact for help.
- Task recommendations: send newcomers a set of tasks related to their topics of interest or past edits.
- Neighborhood activity: send newcomers a list of editing activity that has happened on articles related to their topics of interest or past edits.
- New article: send newcomers who said they intend to create a new article a couple of pointers and link to tutorial.
- Quiz: send newcomers a quiz with a series of questions to determine "what kind of editor are you?" Here is an example from Codeacademy.
General requirements
- Should be possible to include rich content.
- Be sent based on rules that incorporate welcome survey responses, edit history, and time.
- Allow people to unsubscribe.
- Allow us to know how often the emails are replied to.
- Have a noreply@ from address or allow users to reply somewhere that their message will be seen.
Emailability
In order to send engagement emails to newcomers, they need to have verified email addresses associated with their accounts and they need to opt-in to receiving this content. The analysis our team did in this Phabricator task shows that a remarkably low number of users end up with verified email addresses. See the table below.
| Stage | Czech | Korean |
| Have email address (of all newcomers) | 70% | 55% |
| Verified email address (of emails) | 35% | 25% |
| Newcomers with verified emails | 25% | 14% |
Only 25% of Czech and 14% of Korean newcomers have verified emails. When we wondered why these numbers are so low, we realized that there are many moments during account creation where we could increase a user's understanding of what's going on, making it more likely they'll add an email if they want to. Here are some ideas we're working on:
- Adding language to Special:CreateAccount that explains that email addresses are required for account recovery. T216087.
- Adding a notification to Special:CreateAccount letting people know to look for a verification email. T216087.
- Re-designing the verification email itself so that it is clearer what to do with it, and is less confusing. T215665.
- When users try to create their account without an email, ask if they are sure. T215633.
Measurement
As with everything the Growth team does, it is important for us to be able to measure how users do and do not engage with the feature -- that's the only way to stick to strategies that are working and abandon the ones that aren't. We want to be able to measure how often users open and click on the emails we send them, but we know that instrumenting emails is new territory. We'll need to careful to safeguard privacy and security, and we will work with other Foundation teams to do so. We may need to come up with detailed solutions to learn what we need to learn without recording too much data, like the approach our team took with the "Understanding first day" project in which we obfuscated the names of articles viewed by newcomers.