Meta & Children's Online SafetyH2 2'23 PU Youth Industry Practices & Defaults Strategy (v2)
Raw text
## H2’23 PU Youth Industry Practices & Defaults Strategy A/C Priv Well-being Leads Review June 1, 2023 PLTF EX-04321\_0001 1 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) META3047MDL-277-00008364 METANMAG-127-00008364 PLTF EX PLTF EX-04321 * * * ## Summarize opinion from team – takeaway H2'23 YIP Strategy Well-being Leads Review In the 4/18 IG x CSI YIP alignment review, we aligned on YIP definition and status, and agreed to follow up with our overall default strategy for PU as a next step. In today's review we share our plan to meet YIP by end of year across IPs with discussion focused on how we meet YIP by defaulting teens into a break reminder control. Our plan to meet YIP in H2 is as follows: 1. \[Meet PU-2\] Continue time spent accuracy improvements 2. \[Meet PU-6\] Ship a bedtime reminder 3. \[Meet PU-3\] Central resource on how to use our tools - led by CSI 4. \[Meet PU-5\] Default teens into break reminders i.e. Take a Break focus of discussion today Based on guidance that meeting YIP is our highest priority goal for H2 followed by potential new ED/SS focus, and late night use reduction, we are looking to discuss: 1. \[Align\] Revisiting prior guidance on defaults and opting teens into Take a Break by default 2. \[Feedback\] Our approach to defaulting U18 teens into Take a Break by testing conservative (new U18 teens) & aggressive options (all U18 teens) to gauge impacts on late night time spent and the IG ecosystem 3. \[Guidance\] Trade offs - How should the team prioritize YIP vs. late night time spent reduction in the H2 roadmap given current eng bandwidth PLTF EX-04321\_0002 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) * * * | Goals | | --- | | 01 Align on overall approach to meet Youth Industry Practices | | 02 Align on strategy for defaults & proposed product framework | | 03 Discuss next steps & feedback | | Non-goals | | Design & measurement feedback | PLTF EX-04321\_0003 * * * Context PLTF EX-04321\_0004 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) META3047MDL-277-00008367 METANMAG-127-00008367 * * * On 4/18, we aligned with IG WB & CSI leads on the following PU IPs and their status (canonical): \| Youth Industry Practice for Well-being(YWBIP) \| Product Translation →Translation of how this industry Practice applies to Instagram's products \| Status \| Next Steps/Recommendations \| \| \-\-\- \| \-\-\- \| \-\-\- \| \-\-\- \| \| PU-1. Provide tools that control teens' total time spent on our platforms. \| →Provide tools that enable teens to set a daily limit for total time spent on Instagram and receive a reminder when they have reached that limit. \| MEETING Time Spent includes daily limit but we don't default teens \| N/A because we are meeting with Daily Limit \| \| PU-2. Surface insights on usage patterns and durations to teens. \| →Surface insights to teens on daily time spent on Instagram and how long time was spent. \| PARTIALLY MEETING Time Spent exists, but prelim measurement shows its -75% accurate \| Prioritized for H2 \| \| PU-3. Offer age-appropriate information to teens on how to manage their time online. \| →Provide teens with information about how to use the tools to manage their screen time, within time spent tools flow. \| NOT MEETING We only have help center articles and upsells today. \| Prioritized for H2 \| \| PU-4. Provide options to mute notifications specific to nighttime \| →Provide settings that mute notifications with a night-time specific range of time. \| MEETING Quiet Mode exists but we don't default teens nor mute notifs by default \| N/A because we are meeting with Quiet Mode \| \| PU-5. Interrupt long periods of watch time with reminders to take a break and stop scrolling, defaulted to on, and/or move to other topics. \| →Provide teens with a nudge to take a break, triggered by long usage. \| NOT MEETING Take a Break and Topic Nudge exist, but we don't default teens. \| Prioritized for H2. Today's review: IG to review approach for defaulting TaB as part of meeting YIP. \| \| PU-6. Send notifications to teens before their bedtime to remind them to wind down \| →Provide teens with a reminders to wind down before bedtime, triggered by late night usage \| NOT MEETING \| Prioritized for H2 \| PLTF EX-04321\_0005 * * * Slide 5 1 What does IP stand for? Carly 5/31/2023 1 industry practices - at times also abbreviated YIP for "youth industry practices" (since they apply to youth) Elena Davis, 5/31/2023 PLTF EX-04321\_0006 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) META3047MDL-277-00008369 METANMAG-127-00008369 * * * Previous defaults proposal (TAB) In H1'22, we proposed defaulting new teens into Take a Break based on positive feedback from teens, parents, regulators, and press. Why we did not pursue it 1. Leadership shared that the bar for defaults is high and should consider policy/legal requirements, experimental data showing strong adoption, and behavioral data that a majority of users would support a given product. 2. Testing without confidence in launching was considered challenging, as it is difficult to walk back what many would see as a positive step for well-being. We still face this challenge, but it's more defensible if we move forward with some level of default (ie., U16 v U18). With recent announcements from competitors, as well as our company goal to be at Youth Industry Practices, we are reopening the defaults topic for further discussion and re-evaluation. PLTF EX-04321\_0007 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) * * * User needs & solving the problem • Perceived low control over Instagram use is a common problem (~20-40% of youth IG users) and awareness of opt-in control features is modest (23-34% aware of time management features) • 35% of late night time spent for p90+ teens comes from sessions over 15 minutes long. However, 97.5% of teens have not adopted Take a Break or Quiet Mode, indicating opportunity for intervention To more fully address PU behaviors when they are occurring in the moment, controls and nudges could be surfaced in context by default. Industry landscape & regulation • YouTube and TikTok both have taken a strong stance on defaulting teens into time spent controls • Youth Industry Practices specify the need to default a break nudge (default either break nudge or daily limit plus break reminders in 2 of 3 competitors) and a bedtime reminder (default in 2 of 3 competitors). • In addition, select competitors like TikTok have more strict defaults around overall time limits and muting nighttime notifications for teens. To respond to this external landscape, stronger defaults are expected from IG. This would also be in line with recommendations from emerging legislation. Experts The Surgeon General’s recent advisory on teen mental health and social media use called for tech companies to “ensure default settings for children are set to highest safety and privacy standards,” based on synthesis of research evidence and input from subject matter experts. Experts have recommended that default controls can generally help to facilitate “learning through experience”—so teens can try them out, reflect on their relevance, and then make adjustments. * * * Peer app examples \*If a teen turns off Daily Limit on TikTok, they’ll see an upsell asking them to turn on again if they spend more than 100 mins on the app PLTF EX-04321\_0009 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) META3047MDL-277-00008372 METANMAG-127-00008372 * * * H2’23 Plan PLTF EX-04321\_0010 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) META3047MDL-277-00008373 METANMAG-127-00008373 * * * ## Meeting YIP Goal: Move "not meeting" and "partially meeting" statuses to "meeting" through the following workstreams PL2.2 Surface insights on usage patterns and durations to teens. • Adherent by landing H1 measurement of screen time & scheduler accuracy. • Setting H2 goals to improve screen time accuracy & scheduler reliability (details) PL3. Offer age-appropriate information to teens on how to manage their time online. • AHNB is driving this cross FAA workstream (FMA: aethrew48) PL5. Interrupt long periods of watch time with reminders to take a break and stop scrolling, defender to me, quickly move for other topics. • Teen to address or defending Take a Break. • Implementation plan discussed in today's review. PL6. Send notifications to teens before their bedtime to remind them to wind down. • Team is committed to building a deadline reminder, and is exploring adding as standalone or part of Quint Mode (designs to come). • Will also address our core goal metric of reducing late night time spent. SCA PLTF EX-04321\_0011 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) * * * ## Options considered for Take a Break default | Options | 1: Default New U16 Teens | 2: Default New U18 Teens | 3: Default All U16 Teens | 4: Default All U18 Teens | | --- | --- | --- | --- | --- | | Description | Only U16(13-15) Stated Age Teens are defaulted into the feature at registration | Only U18(13-17) Stated Age Teens are defaulted into the feature at registration | All new and existing U16 recommended age teen users are defaulted into the feature (with an upsell to inform them about the feature) | All new and existing U18 recommended age teen users are defaulted into the feature (with an upsell to inform them about the feature) | | Late night impact | "0.1% of incremental LNTS impact per month" | "0.2% of incremental LNTS impact per month" | Potential for >3% impact on LNTS | Potential for >5% impact on LNTS | | Ecosystem Impact | Gradual ecosystem regressions that may not drive noticeable near term regressions (though these will have significant impact in aggregate) | Gradual ecosystem regressions (though these will still have significant long term impact) | High potential for immediate topline regressions in ecosystem. | Likely topline regressions that are stronger than U16 Defaults | | IG and External Precedent | IG announced publicly that we we default new U16 teens to "less" on SCC, but we are soon moving to all U16 | IG defaults new U18 teens to safer TRGM settings | When IG moves SCC to all U16, we'll update comms. | | | TikTok mutes notifs from 9pm onwards for teens 13-15 | YouTube says they default all teens (13-17) for both bed + break reminders | | | | | TikTok mutes notifs from 10pm onwards for teens 16-17 | | | | | | TikTok says they set 60min Daily Limit for all U18 teens | | | | | - Recommended age teens doesn't have signal at registration, only new stated age teens can be targeted - QM/TaB effectiveness {approximate LNTS reduction per adopted user} is currently 6% - Defaulting our other core product, Quiet Mode, would not help us meet PU-5 as is not a break reminder. ## 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 ## ASSUMPTIONS: - New teens - - U16 (EU would be U18) - HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) PLTF EX-04321\_0012 * * * Slide 11 1 Brian: Challenges on option 2, relatively small audience. Exposure points will be critical. Loop in DS/DE leads early in the test setup to get it right in the first try. Be more formal in verification processes that exposures/QEs are set up accurately early on. Anastasia: Opportunity to combine these tests for efficiency Swithin: Defensibility of doing only new teens if that path works better. Sayed: Tiered enforcement, clear principles + tenets on U16 vs. U18. Align across PM pod on POV. Brian: Principles we need to uphold (new vs. existing) - external expectations, In touch with Recommendations leads to have principled alignment with IG leads on trade offs (eng vs. ecosystem). Will net out on this that will influence big swings (Sayed, Brian own AI) Gargi 6/1/2023 2 Will have guidance for ship decision but not impact test plan Gargi 6/1/2023 PLTF EX-04321\_0013 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) META3047MDL-277-00008376 METANMAG-127-00008376 * * * # Team recommendation: Test options 2 & 4 Our recommendation is to test both rolling & full U18 defaults of Take a Break in H2, allowing for more robust learnings on ecosystem and Teen P90 LNTS impact before shipping a version of defaults in late H2. ## Execution timeline: • Late Q3 '23: Start global test of new vs. new + existing defaults • Middle Q4 '23: Bring learnings to leads for review • Late Q4 '23: Launch defaults • H1'24: Explore additional default opportunities ## Rationale: Take a Break can be a powerful tool to reduce Teen P90 LNTS, but we are wary of the effects of full defaults on the IG ecosystem. This proposal will allow us to test our most aggressive option (default all U18 teens), while also gathering learnings from a lower-impact approach (default new U18 teens) that would still allow us to ship a default feature in H2 if we determine ecosystem impacts are too large for full defaults. PLTF EX-04321\_0014 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) META3047MDL-277-00008377 METANMAG-127-00008377 * * * Take a Break Default UX ## New teens defaulted during onboarding • Default new U18 teens to Take a Break (~1M stated age) • No extra education or call-outs are necessary for this default during the onboarding flow (in adherence with precedence set by Teen Safety defaults) • Default will be set to break reminders every \[20\] minutes PLTF EX-04321\_0015 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) META3047MDL-277-00008378 METANMAG-127-00008378 * * * Slide 13 1 @meta.com what is the approximate # of new U18 stated age teens each day? _Reassigned to @meta.com_ Brian 5/31/2023 1 ~1M Stated Age U18 Teens, ~550K Stated Age U16 Teens Taylor 5/31/2023 1 Is the rationale to use 20 mins instead of the other options rooted in FBs approach for consistency? Swithin 6/1/2023 3 Consistency with FB app Follow up async \_ GTM Gargi 6/1/2023 PLTF EX-04321\_0016 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) META3047MDL-277-00008379 METANMAG-127-00008379 * * * Take a Break Default UX New teens receive NUX for first break reminder • Create a NUX for the first TAB experience that new teens receive, triggered by \[20\] min session • Offer user education, control & transparency about how teens can manage and customize their break reminders • All subsequent TAB reminders will revert to the original/current experience PLTF EX-04321\_0017 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) META3047MDL-277-00008380 METANMAG-127-00008380 * * * Take a Break Default UX Existing teens defaulted with transparent upsell • Default existing U18 teens to Take a Break • Leverage the Anti Bullying Defaults Framework to auto-default existing teens into break reminders, with a heads up notice of \[X\] days • Transparent options for existing teens to understand what is changing, turn on the feature now (or wait), and/or opt out if the default is not a good fit Note: Will need to consider strategy & framing for teens who already have TAB on, but have it set to a different time interval than the default (20min). Will also need to consider how this overlaps with any existing Supervision time limits. PLTF EX-04321\_0018 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) META3047MDL-277-00008381 METANMAG-127-00008381 * * * Discussion HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) PLTF EX-04321\_0019 META3047MDL-277-00008382 METANMAG-127-00008382 * * * ## FAQ: Potential H2 Roadmap Beyond Take a Break Defaults ## With current HC (7 eng) • YIP: Ship a bedtime reminder • YIP: Default teens into break reminders i.e. Take a Break • QM Armadillo compatibility • (Android Only) YIP: Continue time spent accuracy improvements ## Collaborations: • BrandX: Quiet Mode adoption, including Activity Feed QP & permanent entry point • CSI: PU-3, central repository of age appropriate information for teens • Others TBD With additional HC: We can prioritize late night use reduction levers outside of YIP • (iOS) YIP: Continue time spent accuracy improvements • Reels interruption i.e. awareness bumper (key strategy for late night use reduction) • Night time wind down timer • Quiet Mode effectiveness (audio off, dim screen) • Quiet Mode day of week customization • (WIP) Ranking: Adjust recommendations at night time • ED/SSI blunt tools - pending more context PLTF EX-04321\_0020 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) * * * 4 YIP in context of PU Blunt levers across discovery recs - digital cleanse/rec refresh Resources for ED/SSI (team has historically owned this) As part of roadmap reviews, WB will review top priorities and find ways to staff. either deprioritize other areas or HC for this Sayed with share ED/SSI context for blunt levers bake all of above priorities into roadmap Gargi 6/1/2023 PLTF EX-04321\_0021 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) META3047MDL-277-00008384 METANMAG-127-00008384 * * * | 1\. Based on the context shared, are we aligned on revisiting prior guidance on defaults and opting teens into Take a Break by default? | | --- | | 1\. Are leads aligned with our approach and UX to defaulting U18 teens into Take a Break? | | With tests of conservative (new U18 teens) & aggressive default options (all U18 teens), we will weigh the relative impacts on late night time spent & the IG ecosystem. | | 1\. Ahead of our goals review on 6/8, do leads have an early steer on how the team should think about resource allocation and goal prioritization? | | At current team size, we expect meeting YIP (4 workstreams) + Armadillo compatibility to utilize 100% of eng bandwidth. | | As a result, without incremental HC, levers to (1) reduce late night use - Reels interruption and Quiet Mode effectiveness (customization of hours and audio visual cues) and (2) ED/SSI blunt levers - will fall below the line. Rough roadmap here. | PLTF EX-04321\_0022 * * * Slide 18 9 Question about default 20 vs. 10 min Gargi 6/1/2023 5 Sayed: Aligned on the rec. Tab: Rationale makes sense across options. Relatively low risk option to see meaningful adoption and retention. Gargi 6/1/2023 6 Ashley: Using most recent, best in class research to inform implementation. Consistency across FoA. 20 min upsell- are you considering Gargi 6/1/2023 8 Sayed: H2 there is opportunity to take on even more ambitious goals. Defaults are a big swing in that direction. Even more optimized for test set up and excellence - fix issues seen this half Brian: This half other teams are regressing our metrics. LNTS is a tier 1 metric, but we don't have a robust method to ensure guardrails are being held. RI as benchmark for defense strategy (DS process oriented). Topline impact for defense strategy that can be in line with order of magnitude for next week. Trade offs: Scale of regression is important Gargi 6/1/2023 7 (1) How can we have meaningful impact in this space? SG advisory as consideration What happens after the default? what if users dismisses? Hard blocks/friction? Long term how can we affect time spent for teens? Feature limiting like for Explore or Reels? (2) Empowerment as a strategy - has it been effective? What roles could parents/guardians play? Identify core tenets that they can apply from settings POV for more stringent settings. Gargi 6/1/2023 PLTF EX-04321\_0023 HIGHLY CONFIDENTIAL (COMPETITOR) META3047MDL-277-00008386 HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) METANMAG-127-00008386 * * * Slide 18 (Continued) 2 Are we comfortable to defend keeping just new teens if we have other controls like SCC that may roll out completely to all teens? Swithin 6/1/2023 1 @meta.com @meta.com @meta.com Can we ensure Policy (liz and others) and Legal (Sonia or whomever on the IG side) PRIV PRIV PRIV Josh Latshaw, 6/1/2023 PLTF EX-04321\_0024 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) META3047MDL-277-00008387 METANMAG-127-00008387 * * * FAQ: Did the team consider other default controls? Current suite of MWB controls that could be considered for defaults. Only Take a Break was considered since PU-5 clearly refers to break reminders. PLTF EX-04321\_0025 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) META3047MDL-277-00008388 METANMAG-127-00008388 * * * FAQ: How is the portfolio performing & how did that inform the team's choice? - Please note Daily Limit has not been actively updated/iterated on outside of maintenance since 2020 (ref/ref) - Quiet Mode has potential as a default strategy but only after we ship globally, optimize retention and hour customization (post H2) - Take a Break is our most mature product, ready for an H2 default strategy, especially with strong retention and equivalent effectiveness, include most recent encouraging results from IGD QP test. | | Quiet Mode | Take a Break | Daily Limit | | --- | --- | --- | --- | | Teen P90 LNTS Effectiveness per User (28D) | -6% per User(Yet to Observe Global) | -6% per User | Not measured | | Ecosystem Impacts | None observed, but not confirmed in Global Test | Confirmed SS Impact on US Core Foreground Sessions(TaB Upsell Test) | Not measured | | Teen Feature Retention(28D) | 63% | 96% | 92% | | Teen Adoption | 1.6% | 1.6% | 1% | | Teen Trigger Rate | | 19% | 13% | | Teen Honor Rate | | 45% | 40% | PLTF EX-04321\_0026 * * * Slide 20 2 @meta.com do we have a hypothesis on this one? This implies that spending more time on the App induces more future sessions? Could this be because increased engagement creates events (comments, posts, dms) that others react to and then bring the user back? _Reassigned to @meta.com_ Brian 5/31/2023 2 We don't have a leading theory yet--I want to validate and investigate more deeply once we have restarted test with tighter targeting, but am holding on additional analysis until we have that data. That said, in the initial test, we did observe SS regression in communication (DMs) in the US, so the communication theory is a viable hypothesis (particularly when we consider the idea that the QP is in DM). Potentially, the user enters DM inbox, sees the upsell and chooses to enroll in QM or dismiss rather than sending a message. This could lead to fewer "come-back" sessions. Taylor 5/31/2023 PLTF EX-04321\_0027 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) META3047MDL-277-00008390 METANMAG-127-00008390 * * * FAQ: How do the products compare for a general defaults strategy beyond PU-5? | Product vs. value of default | User value UXR | Late Night Time Spent | IG Ecosystem Impact | External POV | Product/Eng | YIP Alignment | | --- | --- | --- | --- | --- | --- | --- | | Daily Limit(assume 1 hour) | Feed Pause based on total time spent had positive impact (gen pop)(1,2,3summary here)High time spent strongenet association with perceived low controlGap between perceived effectiveness and awareness(report in review)Expert endorsed works better if limits are a personal goal | We do not have recent measurement of the products ecosystem or goal metric impact since we haven't iterated on it since2020. | | \*/Meets competitor standardTT already announced w/some criticismCould be stronger if we decrease default | Unlikely to meet H2 timeline as product requiresUX refresh&updated baseline measurement | \*/Daily limit default is not a requirement, but would be"beyond" | | Take a Break(assume 20 min) | \*/Upsell already shown to qualifying teens; though repeated break nudges will reach more usersGap between perceived effectiveness and awareness(report in review)Expert endorsed | Global Tests of TaBhave shown6% per adopted user effectiveness in reducing late night time spent | Global Tests of TaB showed55 impact on US Core Foreground Sessions(est,20% impact per adopted user)so defaults likely will impact IG ecosystem. | Meets/exceeds competitor standardAlso addresses"rabbit holing"etc | Product has strong retention,low frictionReminder is surface agnosticBreak honor rate is low | Nudge to take a break is required by default(PU-S)Feed Pause precedentwill make this more timely forIG to meet | | Quiet Mode(assume 11pm to7am) | Positive user feedback and perceived effectiveness from QM users(report forthcoming)Opt out might be high related to retention issue)/Preventing late night'pull'to app is only part of the evening/night problemExpert endorsed | US tests showed6% effectiveness per adopted user at reducing late night time spent,but global tests have yet to confirm these results. | US tests showed minimal impact onIG Ecosystem,but these have yet to be confirmed in global tests | \*/Meets competitor standardExternal scrutinyis currently lower on night use | \*/More investment needed inH2 to increase effectivenessNot yet shipped globally;adds risk due to ecosystem impact | _/Adding bedtime reminder toQM would be a good fit and addressVIP gap forPU-6_/Mutting notifs at nighttime would go"beyond"addressingPU-4(bot not far beyond as1 competitor defaults this) | ## 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 ## ASSUMPTIONS: - New teens - - U16 (EU would be U18) - PLTF EX-04321\_0028 - <PS: SCC moving to all teens not just new> * * * ## FAQ: How does this default strategy evolve over long term? Below is a rough outline of a multi-half defaults strategy, though we’d need to finalize resourcing before locking this in: ## Q3'23 • Ship Reels night time awareness bumper after testing in Q2 (beyond IP) • Test default on for Take a Break ## Q4'23 • Ship Take a Break to U18 teens (meet IP) • Prepare to test Quiet Mode default by investing in effectiveness levers (audio off, dim screen, day of week customization, retention) ## H1'24 • Test & ship (based on results) default on for Quiet Mode • Competitive & regulatory landscape to inform if Daily Limit or any other bundle of defaults is to be considered PLTF EX-04321\_0029 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) META3047MDL-277-00008392 METANMAG-127-00008392 * * * Appendix HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) PLTF EX-04321\_0030 META3047MDL-277-00008393 METANMAG-127-00008393 * * * | Expected External Impact | | --- | | Youth Industry Practices: Would move us from “not meeting” to “meeting” as we currently do not default break reminders like other competitors are already doing. | | Press: Potential for a medium-volume moment. We’d be seen as taking a meaningful step, but catching up to competitors. A larger moment would require doing something new (e.g. stronger break experience or not letting teens opt-out) | PLTF EX-04321\_0031 * * * | Launch volume | Low | Medium | High | | --- | --- | --- | --- | | Expected sentiment impact | Neutral/no impact | Moderate positive or negative | Strong positive or negative | | Impact to user | Only power users would notice | Noticeable, but table stakes and/or another competitor has launched something similar | Very noticeable and/or changes user experience. No competitors with similar feature. | | Potential activations\*Not guaranteed | Mention in bundled announcement (if applicable)Potential inclusion in Hidden Gems, TWOIMention at events | Bundled announcementPress pitching with medium-volume coverageSocial channelsMention at events | Standalone announcement (or highlight of bundle)Press pitching with high-volume coverageExec commsCreator activationMktg supportHighlight at events | PLTF EX-04321\_0032 * * * | Bedtime Reminders | | --- | PLTF EX-04321\_0033 * * * Competitor products [https://blog.google/outreach-initiatives/digital-wellbeing/improve-your-sleep/](https://blog.google/outreach-initiatives/digital-wellbeing/improve-your-sleep/) [https://www.technobaboy.com/2020/05/22/youtube-bedtime-reminders/](https://www.technobaboy.com/2020/05/22/youtube-bedtime-reminders/) [https://techcrunch.com/2023/01/06/tiktok-testing-sleep-reminders-feature-nudge-you-whenbedtime/?guccounter=1](https://techcrunch.com/2023/01/06/tiktok-testing-sleep-reminders-feature-nudge-you-whenbedtime/?guccounter=1) PLTF EX-04321\_0034 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) META3047MDL-277-00008397 METANMAG-127-00008397 * * * Changes since prior guidance # Parameters for targeting bedtime reminders Goal: Provide all U18 recommended teens with bedtime reminders, with opt-out control • Age: YIP requires offering all teens (U18) a bedtime reminder, triggered by late night usage. (YouTube defaults this on; TikTok offers sleep reminder feature). • Scope — New teens vs. All teens: Industry expectation is to offer this feature to all teens (including existing teens), but permits opting out of reminders. If this feature is defaulted on, the impact on goal metrics will be observed sooner. ○ Precedent is not uniform: ▪ SCC started with new teens now going to existing teens ▪ Anti-bullying team defaulted existing users into a Creator bundle with a 30 day notice and option to opt-out • Opt-out/customization: IG precedent is to allow users to update settings and opt out of defaults • Time: Based on usage patterns, we plan to test bedtime reminder variants at 10PM, 11PM, and 12AM and explore the opportunity for age-based timing. We would allow users to customize when they see the bedtime reminder. PLTF EX-04321\_0035 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) META3047MDL-277-00008398 METANMAG-127-00008398 * * * How to offer bedtime reminders? Spectrum of directions | Concept | Standalone Bedtime Reminders | Bedtime Reminders x Quiet Mode | Bedtime Reminders x Take a Break | | --- | --- | --- | --- | | Description | New Bedtime Reminder control added to the suite of MWB products (alongside TAB, QM, etc.). Default teens to seeing the reminder at night, with the ability to customize when it shows up. | Integrate Bedtime Reminder as an option within Quiet Mode, and default on both Bedtime Reminder and Quiet Mode for teens to start at the same time. | Integrate Bedtime Reminder as a subset of Take a Break. Default teens to seeing the Bedtime Reminder part of Take a Break, with the ability to customize when it shows up. | | Pros | Able to more easily build and test efficacy of a standalone productTeens who don't resonate with the other MWB products have the option to turn on only Bedtime Reminders, which could reduce late night useMatches TikTok and YouTube paradigms | Seems like a natural product extension of Quiet Mode, as both are intended for night use and offer a set start timeBedtime reminders are a top-requested feature by QM users(28%) in survey responsesStrengthens one of our existing MWB features,rather than adding a new feature to the portfolio | Bedtime Reminders could be framed as a "break" experience, just intended for when it's time to go to sleepStrengthens one of our existing MWB features,rather than adding a new feature to the portfolio | | Cons | Adds another separate feature in addition to the MWB suite of products(more maintenance)Need overlap logic with TAB and QM if teens are opted into other features that may show up at the same time | Defaulting teens to Quiet Mode would require more product improvements and retention levers before it is feasibleTeens would need to find value in Quiet Mode in order to adopt the Bedtime RemindersIf we move towards Ad-Hoc Quiet Mode, this would impact implication of "bedtime" | Data impact analysis shows that defaulting teens to TAB could lead to negative ecosystem impact, so we'll need to identify trade-offs hereTeens would need to find value in Take a Break in order to adopt the Bedtime Reminders | | Directional Designs | | | | PLTF EX-04321\_0036 * * * Default control # Existing Teens: Upsell TAB Default (Short-term) • Continue to upsell TAB default break reminders to existing teens (following the Teen Safety framework) • Continue to leverage the 1-tap CTA to turn on the feature PLTF EX-04321\_0037 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) META3047MDL-277-00008400 METANMAG-127-00008400 * * * Default control ## Existing Teens: TAB Default (Long-term) • Concept to adopt the Anti-Bullying Defaults framework to auto default existing teens into break reminders • Heads up notice of X days, with option to turn on now or wait until the default updates PLTF EX-04321\_0038 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) META3047MDL-277-00008401 METANMAG-127-00008401 * * * ## A “full” default strategy is the best approach to maximize near-term impact, while a “rolling” default strategy may minimize ecosystem shock. The impact on both LNTS and ecosystem for a default strategy depends on several key factors: 1. Feature Impact: If a feature drives more impact at a per-user basis, the sooner the total impact will be seen (and the larger the potential impact will be). 2. Feature Retention: As with non-default implementations, low retention can reduce impact because of a “leaky bucket” of lost users. 3. Number of Default Users: If we default more users into a feature, the impact on overall goal metrics will be felt sooner. 4. Time: Lastly, for enrollment-linked defaults (such as new-teen users), more impact will be felt as the “defaulted” users make up a larger portion of total users. At current rates, a U18 rolling strategy that defaults new users at enrollment will reach "1M Users, while a U16 rolling strategy will reach "550K users per day. Based on these results, we could expect to see "1% of additional LNTS movement per half for a 6 months of rolling default. On the other hand, a full default push (where all recommended teen users are enrolled at once) could drive "5% of impact on Teen P90 LNTS, but also could drive significant impacts on ecosystem metrics. ## 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 ## ASSUMPTIONS: - New teens - - U16 (EU would be U18) - - <PS: SCC moving to all teens not just new> PLTF EX-04321\_0039 HIGHLY CONFIDENTIAL (COMPETITOR) HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) META3047MDL-277-00008402 METANMAG-127-00008402 * * * | Product vs. value of default | User value UXR | Late Night Time Spent | IG Ecosystem Impact | External POV | Product/Eng | YIP Alignment | | --- | --- | --- | --- | --- | --- | --- | | Daily Limit (assume 1 hour) | • Feed Pause positive impact (gen pop)• High time spent strongest association with perceived low control• Gap between perceived effectiveness and awareness | No understanding of impact on Late Night Time Spent. | No understanding of impact on Late Night Time Spent. | \+/\- Meets competitor standard- TT already announced w/ some criticism+/- Could be stronger if we decrease time | \- Will require refresh and maintenance; could be overhauled in future | \+/\- Daily limit default is not a requirement, but would be 'beyond' | | Take a Break (assume 20 min) | \+/\- Upself already meets this 1x/day, though repeated break nudges might work better• Gap between perceived effectiveness and awareness | Global Tests of TaB have shown 6% per adopted user effectiveness in reducing late night time spent. | Global Tests of TaB SS impact on US Office Foreground Sessions (est, 20% impact per adopted user) so defaults likely will impact IG ecosystem. | \+ Meets/exceeds competitor standard\* Also addresses "rabbit hotting" etc | • Product has strong intention, low history• Reminder is surface agnostic- One way door- Meaningful adoption is low | \+ Nudge to take a break is required by default (PU-5)Find Pause procedure will make this more timely for IG to meet+ TaB upsel currently meets a minimum bar, defaulting TaB itself would be "beyond" (but not far beyond) | | Quiet Mode (assume default) | • Positive user feedback and perceived effectiveness (for those who use it)• Opt out might be high (related to retention issue)+/- Preventing late night 'pull' to app is most effective the evening/night problem | US tests showed 6% effectiveness per adopted user at reducing late night time spent, but global tests have yet to confirm these results. | US tests showed minimal impact on IG Ecosystem, but these have yet to be confirmed in global tests | \+ Meets competitor standardDoesn't go beyond baseline | \+ Implicit pivot to sleep mode+ Track minutes notifs at night by default- One way door | \+/\- Adding bedtime reminder to OM would be a good fit and address YIP gap for PU-6+/- Muting notifs at nighttime would go "beyond" addressing PU-4 (but not far beyond as 1 competitor defaults this) | | Reels night time nudgeNot in scope: Muting not | • Positive user feedback and perceived effectiveness• Teens use Reels to wind-down at night, but often lose track of time+/- More effective triggering could account for surface-switching behavior• Gap between perceived effectiveness and awareness | No understanding of impact on Late Night Time Spent (Upcoming tests should provide this) | No understanding of impact on IG Ecosystem (Upcoming tests should provide this) | \+ Teens cannot opt-out- Only address night use on single surface, would be stronger if it expanded | \+ Less interruptive UX, TikTok does thisTargets existing users as wellTwo way door as it's not a setting- Currently surface specific | • Not specifically related to a YIP gap; this would be another example of a defaulted nudge to take a break• Would not fully meet the bedtime reminder YIP gap | ## 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 ## ASSUMPTIONS: - New teens - - U16 (EU would be U18) - - <PS: SCC moving to all teens not just new> PLTF EX-04321\_0040 * * * | Product vs. value of default | User value UXR | Late Night Time Spent | IG Ecosystem Impact | External POV | Product/Eng | | --- | --- | --- | --- | --- | --- | | Daily Limit (assume 1 hour) | • Not much in what was already documented • Is this less harmful to the ecosystem than default TaB, as an overall time management tool? • Are nudges/reminders at longer intervals of time spent more effective for adherence? • What was TikTok’s criticism noted externally (that it was too long of a time interval?) • Would bundling this with TaB be more meaningful? | | | | | | Take a Break (assume 20 min) | • Defaulting into repeated nudges could be more effective; need to find ways to promote meaningful adoption (ways for users to try it out and check in? varying the messaging? requiring an action to dismiss?) • Updated targeting for addressing ‘rabbitholting’ • Rising to a new bar from Feed Pause; more cohesive narrative? • How to keep it from being a one-way door? | | | | | | Quiet Mode (assume default) | • Improvements to core functionality to improve retention • Adhoc functionality to not only address late night, but broader time management use cases • Having a bedtime reminder feature, to meet YIP • Having muting of notifs at night be a potential default option in the future (would address PU-4) • Bundle with other features to make it easier to stop using at night/step away from IG (like visual cues to wind down) could be more of an external ‘moment’ • Bundle could have the largest ecosystem impact | | | | | | Reels night time nudge | • More effective triggering could account for surface-switching behavior • Only covers a single surface - so it addresses a narrow part of the time management problem, though that might make it additive/unique on top of TaB (part of ‘beyond’ YIP and might address concerns of Feed Pause showing up more persistently than TaB - this would be a new and more tailored nudge) • Could this be a Reels time spent nudge not exclusive to nighttime? | | | | | ## 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 ## ASSUMPTIONS: - New teens - - U16 (EU would be U18) - PLTF EX-04321\_0041 HIGHLY CONFIDENTIAL (ATTORNEY'S EYES ONLY) - <PS: SCC moving to all teens not just new>