Take a Break: H1 '22 Opt-out for New Teens

Raw text

Transcribed text as recorded, including headers and formatting fragments.

Take a Break H1 '22 Opt-out for New Teens Instagram Leads Review A/C Priv February 11, 2022 PLTF EX PLTF EX-01123 PLTF EX-01123_0001 Executive summary Strategy Reduce experiences of Problematic Use and Negative Appearance Comparison and land a steady drumbeat of expert-backed visible products that support teen well-being Our approach: Awareness & Control Identify and drive awareness of on-app behaviors that may lead to negative experiences and design bold interventions/controls. Goal for today's review: • Align on overarching principles and bar for success ahead of test decision Decide if we test (or test with intent to launch) "Opt-out for New Teens" Check alignment to execute Reels upsell Context: Take a Break priorities agreed at high-level in the strategy review on 12/6. Non-goals for this review: • Break Experience: Lower funnel plan is a fast follow (snooze, muting notifs, break duration etc.) • Status: Fast follow pending launch of Status PLTF EX-01123 0002 Proposal: Opt-out for New Teens PLTF EX-01123_0003 H1 Focus Goal Reach more teens who might benefit from taking a break to actually take a break North Star We want as many teens who need the upsell and Take a Break reminder to use it as possible in order to contribute to reducing Problematic Use via awareness and control. In the absence of PU metrics, we define "needing the product" based on key behavior patterns (based on internal and external research) - • • • Long sessions - 15% of teens have 20m+ session/day Frequent back to back sessions - 26% have 20+ sessions/day and half of sessions ≤ 10 min apart Late night time spent - 9% spent 10+ min on IG between 12 - 4 a.m. 35% of teens exhibit at least one of these; 2.6% exhibit all three. Hypothesis Defaulting new teens to Take a Break will help more teens actually take regular breaks from Instagram and upfront avoid or reduce Problematic Use (and NAC for those prone to it) over time. https://docs.google.com/document/d/1WEffBZ806pQXSV6WY6aSsZ20MO6pH-ss4MUWfJiXUIc/edit# Late night time spent / sessions Internal surveys on FB show an association with greater problematic use (1, 2) In a qual study on spirals, "teens say they need the most help during longer sessions that happen in evening hours" (1). Academic research finds that poor sleep is associated with night-time use (1, 2), and that late-night smartphone use impairs next day behavior (1) Very long sessions / overall time spent Internal surveys on FB show that greater time spent on Facebook is associated with more PU (1, 2) Academic research finds that greater time spent associated with more PU (1, 2) Sessions within 10 minutes of a previous session Our goal metric is exactly related to this. Identified in previous PU research (1) PLTF EX-01123_0004 Why should we consider this?: People considerations Teens Teens had a positive reaction to Take a Break at launch and view it as a means to give them agency & control. However research (2) also shows that self- regulation skills for young adults aren't fully developed - making it important to consider opt out as an approach. Post-launch data shows there is room to increase usage, with opt in at 4%; most teens prefer the (longest) 30 min break reminder option. Parents Spending too much time scrolling is a top concern e.g. could displace school work or lead to pressure to look perfect. (external source, internal research by request) Experts General sentiment among experts is that we need to prioritize age appropriate experiences including defaulting teens into those, rather than expecting them to opt-in themselves. Youth Advisors recommend that we default teens into this experience PLTF EX-01123_0005 Why should we consider this?: External landscape Safety (Youth Well-being) + Policy IG has committed to building age-appropriate experiences and youth defaults are an important component because of the Age Appropriate Design Code (AADC) and GDPR which requires we meet the bar of the "best interests of the child." Part of showing progress here is addressing concerns that young people may not yet have the skills or self-control to manage their engagement on our app. Press 'Take a Break' was received with positive to neutral press coverage, but journalists were quick to question why we weren't defaulting teens into it or take a stronger approach on encouraging them to use it (Tech Crunch, WSJ, CNN) Regulators There is additional legislation in progress across in several key countries and regions regarding nudge techniques and time spent. For example, the UK Online Harms Bill draft notes that social media companies should 'suggest well-being enhancing behaviours' (such as taking breaks). Defaulting teens into TAB would resonate well with regulators and set us up well for legislation that is to come. PLTF EX-01123 0006 9:41 < Similar Posts Q Liked by kenzoere and others 三 9:41 M Current State: Opt In for All Teens • • Available to all teens (users with stated age 13-17), new and existing Teens get an upsell for Take a Break on Explore Similar Posts surface to teens after they have spent 20 mins anywhere on the app They can then choose a break reminder frequency, which will remind them to "close Instagram" after they've spent that much time on app lil_lapisla golden hours lofti232 Great Shot! View all 5 comments Add a comment... 1 day ago 0: Want a break? Regular breaks can help you reset. You can now turn on reminders to take breaks when it makes sense for you. photosbyean Turn on Not now Roddy Ricch The Box Time for a break? You've set reminders for every 10 minutes. Take a moment to reset by closing Instagram. Take a few deep breaths Write down what you're thinking Listen to your favorite song Do something on your to-do list ♡ Q Done Edit reminder Upsell in Explore Break experience Take a Break is launched to 6 EN countries on iOS (9MM eligible for upsell). Android and Global testing is complete, launch on Feb 3 PLTF EX-01123_0007 Proposal: Opt-out for New Teens Turn on feature by default for new teens. Remind users that the feature is "on" when they reach 20 min in the app with a NUX designed to explain the mechanics of the feature, and how to change the reminder frequency, or turn the reminders off altogether. Key Considerations: • Proactively creates teen awareness when they spend long periods on app • Strong external narrative and proof point about our commitment to building age appropriate experiences • Behaves like other teen defaults (see slide 20) Risks: • While teens can opt out or choose different duration, this experience reduces upfront agency for teens • Could cause confusion, especially the first time. Leading users to turning the feature off because they don't know what it is 9:41 三 9:41 Time for a break? Regular breaks can help you reset. We'll remind you to close Instagram every 20 minutes. You can turn off these reminders below or anytime from settings. Take a few deep breaths Write down what you're thinking Listen to your favorite song Do something on your to-do list Not now Manage reminders Time for a break? You're getting reminders every 20 minutes. Take a moment to reset by closing Instagram. Take a few deep breaths Write down what you're thinking Listen to your favorite song Do something on your to-do list Not now Manage reminders NUX (Feature education and management) Steady state (Break reminder) Some more ideas: different CTA(s) (i.e. 'Got it' / 'Okay' instead of 'Not not' or whatever we use for affirmation), stronger contrast between text and background to make sure user is well alerted to the feature, remove break activity recommendations to bring more focus to the NUX text. PLTF EX-01123 -01123_0008 0008 Key Decision Considerations PLTF EX-01123 -01123_0009 0009 Product Principles Awareness and control We optimize for driving teens awareness and providing controls for them to manage their time on IG. Solves the user problem We prioritize designing for users who lack self-awareness or control over the amount of time they spend in the app, and balance this with empowering users who are self-aware to choose what's best for them. Resonate externally We consider the external landscape because building solutions for teens comes with additional sensitivity and scrutiny. Consider ecosystem impact We want to ensure our solutions help the key target segments that need it most; through this we ensure that any downsides (ecosystem impact) are sufficiently balanced by upsides (Well- being benefits). PLTF EX-01123_0010 Test Success Criteria Our bar to consider the test a success is: • Awareness & Control: Opt out rate within threshold Solves User Problem: Break usage within threshold (how often do new teens take a break after seeing the reminder) Considers Ecosystem: Material impact to IG critical metrics that is discussed/aligned on with leads Supporting data for additional signal : Standard support tool sun survey for understanding perceived value Qual feedback from lab sessions (North Star) Reduction in behaviors associated with PU (new metric) (Stakeholders have varying POVs, represented in discussion #1) PLTF EX-01123 3_00- 0011 Launch scenario grid: WB impact vs Ecosystem impact Impact on IG Critical Ecosystem Metrics Scenarios & Possible Outcomes Other parameters to consider: • • If the test is picked up by press/regulators, difficult to reverse IG ecosystem overall and room for additional budget External environment new regulations Age 13-15 vs. 16-17 Well-being upside (Users find value & solves user problem) Positive Neutral Negative Positive Launch Launch Neutral Launch (based on confirmed ecosystem impact; extend test duration) Negative Maybe w/bias to launch based on research conviction/exper ts Don't launch Maybe (evaluate based on other parameters and scale) Don't launch; harder to explain from external POV PLTF EX-01123 0012 Product Private by default (new teens) Hide Like Counts aka 'Daisy' (all ages, new + existing) Remix/Tag/ Mentions/ Guides off by default (new teens) Take a Break (20 min reminder on by default; new teens) Benchmarks: Comparing Decisions on IG Controls Interaction Safety Privacy for Well-being Minors U18 Data Use Test Bar X X X X X X X Launch decision Tradeoff on regulatory & competitive expectations vs. decline in teen engagement "No ship" in 2020 Shipped in 2021 (U18 EU, U16 global) Ecosystem vs. WB survey results trade off: some WB upside specific to like counts 28 day test unblocked user expectations, neutral to safety settings, external perception, ecosystem Tested "always on" but not shipped due to high ecosystem impact and low upside. Launched as opt in Test in progress (decision ETA: Mar 22) Adoption and ecosystem data with supporting signal from survey data, qual sessions and long term behavior change TBD (This review) https://docs.google.com/presentation/d/1MOTjjSySP8pMdSesY59AQZRHUQ- yNkbBzQtj4gVG_Bc/edit#slide=id.gdfcd26b3cb_0_504 https://fb.quip.com/oS7bAbbcLte2 PLTF EX-01123_0013 Compared to Benchmarks, Upper Bound Projected Ecosystem Impact is Lower Overall, our rough estimate for the upper bound top line time spent regression is -0.08% (details in table). This is lower than benchmarks like teen Private by Default and Daisy by Default which coincidentally both had -0.3% time spent regressions (Private by Default showed ~2% drop in teen engagement and Daisy also showed -1% to -2.6%). We would likely see lower time spent regression than this estimated upper bound as teens may spend more time on the app in shorter sessions even when they do honor the break. All users Take a Break Funnel Top line share % of top line (min per day per user) 28.8 100% 13-17 stated age 2.63 9.1% Time spent past 20m in 20m+ 0.17 0.59% sessions Stay Opted-In 0.15 0.53% Honor the break 0.02 0.08% Data is a 1% sample from 1/30-2/5 (7 days). Opt-out and honor rates estimated from current launch product usage. Total time spent estimated assuming 1.7B WAU. Source: https://fb-my.sharepoint.com/:p:/p/stanleybrode/EUX7SSFwEF1Lrv-YknFAVCcBYHC8gpczSw5qkU8--Fe2YA?e=ArGEkV https://docs.google.com/spreadsheets/d/1ccMiDtgZ_y0nE6yUAD4lJqlGqvh-6uJl8Q6cWufk7cg/edit#gid=0 PLTF EX-01123_0014 Considering the External Landscape: Competitor Well-being Teen Defaults (Green shows where competitors lead) Feature f Daily Time Limit Default off, teens or parents can turn on X Yes, default off Yes, default X off Yes for YTKids, default off X Break Reminders Yes, default off X Testing Yes, default on (not a control) X Yes, default off X Do Not Disturb Yes (pause notifs), default off Yes, default off Yes, default off X Yes, default off X X Nighttime Mode X X X Yes -- notifs muted for all teens X Bedtime reminder, default off X Hide Like Counts Yes, default off N/A Yes, default off X X Yes, default off Has tested, not launched PLTF EX-01123_0015 Decision topics #1 #1: Do we agree with the proposed bar (slide 11-12) for testing Take a Break default compared to those for benchmarks (slide 3) for safety/privacy? There are two parts to the recommended decision bar: 1. 2. Similar to benchmarks: Based on ecosystem impact and expectations set by external landscape Additional: Measure positive impact for "solves user problem", using usage funnel metrics since we don't have PU measurement yet We believe this higher bar could help address trade offs between well-being upside and ecosystem while making a launch decision. There are different POVs across stakeholders: Upside: Team: Well-being upside through adoption measures is our best proxy in the absence of operationalized PU measurement JM (higher upside bar): Well-being upside through measurable reduction in PU behaviors to justify this intervention for all new teens Downside: Ecosystem: Can't upfront unblock launch without reviewing specifics of ecosystem impact Policy Comms (lower downside bar): Well-being impact might be hard to quantify, especially in the neutral cases. Need more flexibility in ecosystem impact as reversing a test will be hard to justify For reference, benchmark products have a larger impact on IG product experience - Private by Default limited connections/inbound follows, impact how connected new accounts felt on IG and Daisy impacts counts that are used as social proof and as a signal on what's worth time to look at. PLTF EX-01123_0016 Decision topics #2 #2: If we are aligned on the bar, then are we aligned on how our framework defines Go/No Go scenarios and the challenges for deciding not to launch after testing? There are two "No Go" scenarios to discuss: 1. 2. Strong negative ecosystem impact: We will need to weigh this against well-being signal and external factors. If the test is picked up by press/regulators, will be difficult to reverse without a strong user-based reason Neutral well-being impact: It is somewhat likely that we may not have sufficient signal on the well-being upside (due to common measurement challenges in new well-being areas like this). This will warrant further discussion based on ecosystem impact and might be similarly hard to reverse PLTF EX-01123 0017 Decision topics #3 + #4 #3: Which path should we decide today for new teen accounts? • A: Continue "Opt In for All Teens" status quo and improve • Pro: Least disruptive UX to teens, likely no ecosystem impact • Con: Doesn't create incremental awareness for teens, doesn't address external criticism B: Test "Opt-out for New Teens" with reactive comms • Pro: Allows us to evaluate trade offs and iterate with data before making a one-way door decision • Con: May set expectations to launch that we'll miss if we don't, which could amplify criticism vs. status quo C: Test "Opt-out for New Teens" with proactive comms (as a limited launch) • Pro: Creates upfront awareness for teens. Shows a commitment to building age appropriate experiences • Con: One-way door decision to announce now, before we have supporting data #4: If we're deciding option A, feedback on the Opt-in options? PLTF EX-01123_0018 Check Alignment: Reels Upsell PLTF EX-01123 0019 9:41 三 9:41 三 Upsell in Reels User experience Existing teens will see the upsell on Reels chaining after they have spent 20 minutes on the app Upsell will appear 1x/day (inclusive of upsell on Explore, whichever is seen first) Recommended Approach Short term: Creator collab reel. Modified Ul to support Break CTA. Start with 6 EN-speaking countries with highest scrutiny. Long term: Follow-up with a more scalable solution for rest of world. Use mid-card solution, which allows to have the text dynamically translated. Key considerations Trigger: We are reviewing chaining/consumption behavior on Reels (also in comparison to Explore) and will follow up with a recommendation by 2/4. instagram. with jazzyoh Regular breaks can help you reset...more jazzyoh Original audio Turn on break reminders Contribution to Goal: 8.4% of goal metric Test start: February 28 ♡ 823k Q 51 Reels Set reminders to take breaks? Regular breaks can help you reset. You can now turn on reminders to take breaks when it makes sense for you. Turn on reminders Q Short term editorial video, EN only Long term globally scalable Ul PLTF EX-01123 0020 Decision topic #5 #5: Execute Reels Upsell? We heard support from Communities on Dec 6 (and in recent Well-being email review, and email design review) that this is a straightforward opportunity. Do we agree on moving this into execution? PLTF EX-01123 0021 3_002 Appendix: More Opt-in Options PLTF EX-01123 0022 Additional opt-in options See visuals in slide 24 1. Same QP, updated logic Same experience as today in Explore, but with increased frequency and number of impressions. Can be potentially introduced in Feed as well. 1. Updated QP functionality, easier activation One tap conversion (instead of 3 taps). Can be introduced in Feed as well. 1. Full screen upsell in Reels + Story Full screen upsell on selected immersive surfaces (Reels and Story) 1. Full screen upsell takeover anywhere in the app [Not Recommended] Block the user from engaging with the app until they choose to turn on the feature (or dismiss the blocking upsell). Disruptive solution with no precedent. PLTF EX-01123 0023 9:41 三令 9:41 < Similar Posts < Similar posts ♡♡ M X Liked by kenzoere and others lil_lapisla golden hours lofti232 Great Shot! View all 5 comments Add a comment... 1 day ago C Want a break? Regular breaks can help you reset. You can now turn on reminders to take breaks when it makes sense for you. Turn on Not now Set reminders to take breaks every 20 minutes? Regular breaks can help you reset. We'll remind you to take a break when you spend this amount of time at once on Instagram. Turn on reminders Choose different frequency photosbyean Roddy Ricch. The Box photosbyean Roddy Ricch. The Box Q 1. Increased frequency/impressions of current experience Q 9:41 X 9:41 Set reminders to take breaks every 20 minutes? Regular breaks can help you reset. You can now turn on reminders to take breaks when it makes sense for you. Set reminders to take breaks every 20 minutes? Regular breaks can help you reset. We'll remind you to take a break when you spend this amount of time at once on Instagram. Turn on reminders Turn on Edit Not now 2. One tap opt-in 3. Contextual full screen upsell on selected surfaces (Story, Reels) 4. Blocking interstitial takeover anywhere in the app [Not recommended] PLTF EX-01123_0024 Appendix: More Context PLTF EX-01123 0025 What is Support Effectiveness? Support Measurement Framework is the conceptual framework we have been developing to holistically evaluate how well our tools reach, engage, and support our users. The original framework has evolved over time but essentially, over the past year, through tests with our bullying products, we've learned the following.. Resolution is not always possible BUT it can be captured through two other constructs (Support and Effectiveness) Support captures the process of WHAT we did to help someone with a need Effectiveness captures the results or HOW we did someone with a need We now have a well established set of tailored questions that we can tailor to a product solution. Two new constructs, resilience and relief may offer new ways to capture impact. Measuring Support and Effectiveness of Perceived Resolution and Resilience (Survey Outcomes of 5 Bullyiing ProductsAnalysis in Progress) SRT Approval for Survey Measurement Questions Survey Review Approval Sample Survey (Hidden Words) 5 Insights from Validating Support Effectiveness Measurement Framework (09/21) Product Brief: Support Measurement Framework (02/21) What is Support Effectivness (12/2020) PLTF EX-01123_0026 Why late night sessions and/or long sessions? The behavioral signals we recommend have been shown to relate to problematic social media use in internal quantitative and qualitative research, as well as in external academic research. Late night time spent / sessions • • Internal surveys on FB show an association with greater problematic use (1, 2) In a qual study on spirals, "teens say they need the most help during longer sessions that happen in evening hours" (1). Academic research finds that poor sleep is associated with night-time use (1, 2), and that late-night smartphone use impairs next day behavior (1) Very long sessions / overall time spent • • Internal surveys on FB show that greater time spent on Facebook is associated with more PU (1, 2) Academic research finds that greater time spent associated with more PU (1, 2) Note: passive sessions/time spent may be more problematic (1). PLTF EX-01123 0027 Performance Metrics (Top of Funnel - Setting up the control) Funnel Step # of Teens % of above Teens who got the upsell 3.72M Teens with break turned on 166K 4.5% opt-in Post-launch Performance Performance Metrics (Bottom of Funnel - Pausing app usage when reminded) Funnel Step # of Teens % of above WAU teens with break turned on 172K (223K non-teen) % that took 1+ break in past week (goal metric) 29.8K 17% Performance Takeaways Overall adoption from users seeing the upsell has been consistent (~4%), and 92% of teens retain the reminder after enabling it. However, break reminders decrease in effectiveness over time (20% down to 10% over a month, indicating some fatigue). We will follow up with additional optimization ideas for lower funnel. Source data: https://docs.google.com/document/d/1Hgy2E2z1q5Linx5qZHlckVs7A15PvUDuphZfdeAp2QU/edit# WAU here is IG WAP (iOS + Android) who have the break turned on (regardless of whether they actually honor the break or not). PLTF EX-01123 0028 Live experience 9:41 9:41 < Similar Posts < Q 口 Q Liked by kenzoere and others lil_lapisla golden hours lofti232 Great Shot! View all 5 comments Add a comment... 1 day ago Similar Posts Liked by kenzoere and others lil_lapisla golden hours lofti232 Great Shot! View all 5 comments Add a comment... 1 day ago M 9:41 三 Want a break? Regular breaks can help you reset. You can now turn on reminders to take breaks when it makes sense for you. Turn on Not now Want a break? Reminder to take breaks We'll remind you to take a break when you spend this amount of time at once on Instagram. 30 minutes Time for a break? You've set reminders for every 10 minutes. Take a moment to reset by closing Instagram. Take a few deep breaths Write down what you're thinking Listen to your favorite song Do something on your to-do list ♡ photosbyean Roddy Ricch. The Box Q Upsell in Explore 20 minutes 10 minutes Off Done Done Edit reminder Selector Break experience PLTF EX-01123 0029 Take a Break behaves like other teen defaults Like other products (ie tags, mentions etc), we won't let the teen know that their break reminders are on by default. They will see the NUX break reminder the first time they spend 20 consecutive minutes on the platform. 9:41 < Time spent < Time on Instagram 1h 12m 2:18 Mentions Allow @mentions From Everyone People You Follow Daily Average Average time you spent per day using the Instagram app on this device in the last week No One Thu Fri Sat Sun Mon Tue Today Manage Your Time Set reminder to take breaks 20 min > Schedule a reminder to take regular breaks from scrolling. Set daily time limit > Limit the time you spent on Instagram each day by scheduling a reminder to close the app. Notification Settings > Choose which notifications to get or pause notifications for a little while Q Taß teen defaults Choose who can @mention you to link your account in their stories, comments, live videos, and captions. When people try to @mention you, they'll see if you don't allow @mentions. Q @mention teen defaults Θ PLTF EX-01123 0030 123_0030 Benchmark Ecosystem Regressions Results from 12/11-12/25 network test unless specified otherwise S IG Event Based Revenue (full test period) IG Event Based Revenue Ad Impressions, Feed Ad Clicks, Feed Pure Daisy -1.0% ± 0.9% -1.361% ± 0.854% -0.060% ± 0.189% -0.923% ± 0.295% Pure Daisy Teen Overall Appwide IG DAP 0.023% ± 0.156% -0.092% ± 0.067% IG App Sessions IG App TS -0.281% ± 0.213% -0.30% Consumption & Interaction Overall Likes -1.040% ± 0.396% -1.162% ± 0.320% Overall Comments -2.673% ± 1.026% -2.132% ± 0.696% Home Feed Post Impressions 0.019% +0.332% -0.022% + 0.183% Explore Media Impressions -0.278% ± 0.416% -0.395% ± 0.233% Outbound Follows -1.227% ± 0.638% -0.989% ± 0.444% Metric (Test vs. Control) Topline Today Topline in 1 Year Topline in 5 Years Topline Teen Impact in 5 years Overall DAP -0.02% -0.06% -0.3% -2.2% Overall Time Spent -0.02% -0.05% -0.3% -1.9% Overall Feed Impressions -0.01% -0.06% -0.3% -2.3% Overall Story Impressions -0.01% -0.05% -0.2% -1.7% Overall Feed Creation -0.05% -0.12% -0.6% -4.4% Overall Story Creation -0.03% -0.07% -0.4% -2.6% PLTF EX-01123_0031 IOS Reminder Adoption Trend IOS: Break Reminder Adoption Rate by Days Since First Reminder (Android Teen 6EN Take a Break Upsell Experiment, 2021-11-10 to 2022-01-10) Reminder Adoption Rate 25.0% 20.0% 15.0% 10.0% 5.0% 0.0% 66 46 0 60 8 5 5 5 50 4 4 4 4 40 33 35 34 33 30 28 25 24 2 20 18 15 14 12 10 8 6420 Days Since First Reminder iOS_EN_launch Days within 34 are iOS EN Teen public launch, beyond 34 are data from iOS EN Teen public test with small sample sizes The trend is similar across different reminder settings, though 20m has a slightly higher rate than 10m. Source: https://fburl.com/anp/cunbjc53 PLTF EX-01123 0032 Retention Rate Retention Rate of Take-a-Break Reminder (iOS 6EN Launch, 2022-01-10) 100.0% 94% 96% 93% 80.0% 60.0% 40.0% 20.0% 10m 95% 94% 20m 88% 30m 0.0% NonTeen Teen Teen vs Non Teen PLTF EX-01123_0033 > Trigger Rate & Adopt Rate Trigger and Adoption Rate of Take-a-Break Reminder, (Denominator = WAU with TaB Enabled) (iOS 6EN Launch, 2022-01-10) 71% 60.0% 47% 40.0% 43% 66% 40% 36% 29% 20.0% 19% 10% 0.0% NonTeen_10m NonTeen_20m NonTeen_30m Teen_10m 24% 15% 7% Teen_20m Teen_30m Teen/NonTeen and Reminder Setting 1_displayed 2_adopted PLTF EX-01123_0034 Teen vs. Non-teen Preferred Break Reminder Frequency • • Majority teens preferred 30 min followed by 10 min Majority adults preferred 30 min followed by 20 min ds reminder_setting Q n_users Q n_stated_teens Q n_stated_adults n_age_na Q 2022-01-09 10m 177214 55059 88092 34063 2022-01-09 20m 100730 44557 40085 16088 2022-01-09 30m 115977 63474 37487 15016 PLTF EX-01123_0035 About 30% of teens globally (44.6M) will have had at least one 20m+ session in the course of a week. 17% of teens (25.5M) will have had at least one 30m+ session. (There were 149M teen WAU from 1/20-1/26.) % teens with at least one session of a given duration in a week 1/20-1/26 60% 40% 20% 52.13% 30.05% 17.19% 0% 10m+ 20m+ 30m+ PLTF EX-01123 -01123_0036 0036 Should we opt teens in at account set up? Is it too disruptive? On registration day, only 8% of teens had at least one 20m+ session, 7d 23% and 30d 31% If we add a delay, teens might see the upsell or control on time spent and choose to ignore it (opting them in after X days would disregard that choice). Alternately, we may need to delay upsell by X days, adding complexity Team recommends that given a small % of teens will see this on registration day and adding a delay has more complexity than benefits, we opt teens in at account creation Users with at least one long session (cumulative) 50%- 40%- non_teen 28d: 53.8% 14d: 49.6% 7d: 45.1% 28d: 31.2% teen unknown 28d: 54.6% 14d: 50.8% 7d: 46.4% 28d: 42.8% 14d: 37.9% 7d: 33.7% Session duration 28d: 31.4% 10 20 30 30%- 14d: 26.9% 20%- 10%- 7d: 22.6% 28d: 19.1% 14d: 15.4% 7d: 12.2% 14d: 27.1% 7d: 22.8% 28d: 18.7% 14d: 15.1% 28d: 21.1% 14d: 17.6% 7d: 14.4% 7d: 12% 28d: 11.2% 14d: 8.9% 7d: 6.7% 10 20 10 20 0 10 20 Days since registration PLTF EX-01123 0037 • Current External Landscape: Safety: Committed to age appropriate experiences. Control is for all users. Teen need more protection and not have as much self- awareness. Defaulting them in is good for them. 8/10 teens retained their private setting Policy: Policy makers argue we put profit over youth well-being, and that we intentionally attempt to get youth addicted to our product. This was particularly highlighted in the recent US Surgeon General's Youth Mental Health Advisory, which also recommended social media companies invest in features that "remind children to take breaks and limit screen time." Regulatory: US: Strong interest in experience for minors, legislation around privacy, mental health impact, BR: Draft bills that mandate how much time kids can spend online, PH: Age management legislation that ask companies to insert logical stopping point 30 mins a day for social media). UK: In the UK, the Online Harms Bill says that social media companies should 'suggest wellbeing enhancing behaviours' (such as taking breaks). DE: Germany's Youth Protection Act requires us to take appropriate precautionary measures to protect minors and cites 'default settings that limit usage risks'. Defaulting teens into TAB would resonate well with the regulator. Experts: General sentiment among experts that we need to prioritize age appropriate experiences including defaulting teens into those, rather than expecting them to opt-in themselves. For example, UK-based group 5Rights said: "Instagram suggesting to teens that they take a break from a platform that is designed to harness their attention is like asking an alcoholic at the bar if they would consider herbal tea." <TBD Youth Advisory feedback tomorrow> Press: 'Take a Break' was received with positive to neutral press coverage, but journalists were quick to question why we weren't defaulting teens into it or take a stronger approach on encouraging them to use it (Tech Crunch, WSJ, CNN). More generally, we're under intense criticism with regard to the 'addictive nature' of our platform (e.g., new lawsuits in CA and OR) and we can expect this to continue. <Don't have much to point back to> <lip service over committing and showing we care about protecting teens> Competition TikTok has a different product - shows it every time you hit 40-45 mins. Gets props for their features. Guardians/parents: Spending too much time scrolling is a top concern (e.g. could displace school work or lead to pressure to look perfect). There is a lack of knowledge around available tools to help address this. (external source, internal research by request) Teens and creators: Both groups shared enthusiasm about the feature and said they would appreciate ongoing reminders to take breaks, but we know awareness of the features is low at this point. UK Parliament has recommended that the Online Safety Bill should have a Code of Practice, that require would social media companies to mitigate the following (through e.g. default on features/tools for teens): auto-play; endless scroll; and rabbit holes. https://committees.parliament.uk/publications/8206/documents/84092/default/ UK Age Appropriate Design Code suggests for 13-17 year olds companies should: Suggest wellbeing enhancing behaviours (such as taking breaks). https://ico.org.uk/for-organisations/guide-to-data-protection/ico-codes-of-practice/age-appropriate- design-a-code-of-practice-for-online-services/13-nudge-techniques/ PLTF EX-01123_0038 Competitors & Market Landscape: Well-being Teen Defaults (Green shows where competitors lead) Feature Daily Time Limit Default off, teens or parents can turn on X f Yes, default Yes, off ► X default off Yes for YTKids, default off X Break Reminders Yes, default off X Testing Yes, default on (not a control) X Yes, default off X Do Not Disturb Yes (pause notifs), default off Yes, default off Yes, default off X Yes, default off X X Nighttime Mode X X X Yes -- notifs muted for all teens X Bedtime reminder, default off X Yes, default Hide Like Counts Yes, default off N/A X X off Yes, default off Has tested, not launched PLTF EX-01123_0039 Existing Reels patterns (In Testing) Reels team is testing half card .. 9:41 三 9:41 三 < Reels mejuri Sponsored When you're taking the pawfect photo Ed Sheeran Castle on a Hill Ed Learn More Ads 823k 51 > Create reels with effects Join amalmuzaffar, vanessaqin and 123 other friends who are creating reels with Audio. ♡ Browse Effects Full screen card PLTF EX-01123_0040 +20 Minute Variant IG Criticals Ecosystem Impact +20 Minute Variant Ecosystem Impact Materiality Threshold Topline Impact Description Good User Experiences Fully Qualified Name Pillar (Budget) Lower Bound Topline Impact Upper Bound Topline Impact Stat-Sig? Meets Budget? instagram:perf good user experiences Product Foundation -0.20% -0.22% 0.06% No No Overall Objectionable Prevalence community_integrity:cep:viewer:ig_holistic_prevalence_world_combined Wellbeing 1.00% -0.77% 0.90% No Yes Foreground Sessions Reels Intentional Plays Stories Impressions Feed Impressions Direct Sends instagram:sessions:foreground_session_cnt:overall Feed & Relevance -0.20% -0.11% 0.05% No Yes instagram:clips_media:video_playback:intentional_play_count:overall Reels -0.75% -0.28% 0.12% No Yes instagram:media_vpvd_impressions:stories:home Stories -0.30% -0.07% 0.21% No Yes instagram:post_vpvd_impressions:feed:home Feed & Relevance -0.50% -0.14% 0.03% No Yes instagram:direct:sends:overall N/A -1.00% -0.18% 0.25% No Yes IG Time Spent instagram:time_spent:overall N/A -0.20% -0.12% 0.10% No Yes Infra - Power instagram:efficiency: normalized_power:overall Infra 0.40% -0.10% 0.03% No Yes Infra - TAO instagram:efficiency:tao_fbobj_get:overall Infra 0.40% -0.11% 0.03% No Yes IG DAP instagram:growth:instagram_classic:dap:overall Growth -0.20% -0.04% 0.01% No Yes Original Stories Creation instagram:creation:producer_17:original_media:stories Stories -0.50% -0.11% 0.11% No Yes Reels Producer DAP instagram:creation:clips_media_producer_dap:overall Reels -2.00% -1.51% 1.06% No Yes Creator Watch Time instagram:video_consumption viewer:creator_watch_time_60s_capped excluding_stories Creators -0.50% -0.18% 0.09% No Yes IG Revenue instagram:abp:ad_metrics:revenue_usd_cents:overall IG4B -0.20% -0.07% 0.20% No Yes IG DAU instagram:growth:instagram_classic:dau:overall Growth n/a -0.04% 0.01% No Yes Source query: https://fburl.com/deltoid3/u679t6ji Color Key: Reach: Doesn't violate topline threshold, stat-sig 8.70% Violates topline threshold, stat-sig No stat-sig difference between topline threshold and topline impact Source: https://docs.google.com/spreadsheets/d/11WEGaWnyozNTcU_ttklsPYTpnyPIK35ae1yeZwb ARs/edit#gid=1621687515 PLTF EX-01123_0041 Test Success Criteria Principle Success Definition Success Criteria Awareness & Control Teens that were defaulted in are finding similar value as those who converted via the upsell • Opt out rate within threshold NUX break usage within threshold (May be lower than current opt in model) Solves User Problem Ongoing retention and corresponding reduction in (behaviors associated with) problematic use Resonates Externally Positive feedback from experts and policy makers Considers Ecosystem Metrics indicate a valid trade off between well- Impact being upside and any ecosystem regressions Since there's no single clear Problematic Use metric, we will primarily look for signal via support effectiveness survey Supporting data via: Qual feedback in lab sessions (North Star) Reduction in PU related behaviors N/A since the test will not be announced publicly • Ideally, no material impact to IG critical metrics. Warrants further discussion depending on scale Test should run long enough to be able to project long-term effects (possibly 4-6 wks) Further product improvements needed to more effectively target PU behaviors PLTF EX-01123_0042 9:41 X Opt-out (easy disable) Turn on feature by default for specific cohorts. Remind users the feature is on when they reach their time in the app. Main CTA to turn the feature off altogether. Time for a break? We'll remind you every 20 minutes to take a break from Instagram. Pros Cons Proactively creates teen awareness when they spend long periods on app Strong external narrative and proof point about our commitment to building age appropriate experiences and taking a stand Easy opt out of the product Due to easy ways to turn the feature off from the reminder screen, it could make it harder for us to re-upsell user in the future if they chose to opt out. Could cause confusion, especially the first time, leading users to turning the feature off because they don't know what it is Take a few deep breaths Write down what you're thinking Listen to your favorite song Do something on your to-do list Turn reminders off Edit reminder Main CTA turns feature off Only 10m and 20m variants were associated with an overall decrease in daily time spent https://docs.google.com/presentation/d/1bnfLZ-qIz7a163N17aDGIE7Dd3UW9VZeGZy1D- dBSVk/edit#slide=id.g1037589bc4e_0_0 20m variant was also associated with the highest likelihood of ending a session. More recent analysis indicates that this continues to hold, although the differences are not significant. https://docs.google.com/presentation/d/1bnfLZ-qIz7a163N17aDGIE7Dd3UW9VZeGZY1D- dBSVk/edit#slide=id.g1061a3380a2_0_8 https://docs.google.com/document/d/1Hgy2E2z1q5Linx5qZHlckVs7A15PvUDuphZfdeAp2QU/edit#bookmark=id.bbt2fmbxn81u Highest retention rate: see slide 19 PLTF EX-01123_0043 IG Benchmark Controls & Similar Past Launches Intentional Time U18 Data Use Expectations For Spent/Well- Top of Privacy Funnel/Interaction Safety/Harm Minors Being Prevention Private By X X Default Embeds-Off By X X Default Take A Break By X Default DM Reachability X X Restrictions By Default Daisy By Default X Remix-Off By X X X Default Tag/Mention X Restrictions By Default Past Default Launches Daisy by default - • Private by default - Default for teens (28 day test approved) - Research confirmed user need, neutral on teen safety/integrity metrics, improved external perception, ecosystem impact PLTF EX-01123_0044