Meta & Children's Online SafetyChat: Honestly the only insight I see in these charts is that teens are really into using IG at 11pm when they probably should be sleeping
Source PDF not yet released. Text is transcribed from the public filing.
Thread View
████ name redacted. Different bars may be the same person or different people.
quoted in an excerpt in the archive.
Friday 9 April 2021
09:01████How do you guys get screenshots of the app for decks? Do you just use your own accounts?
09:02████yup lol
09:02████I see - my direct inbox has no threads - haha and I only use messenger lite because it is way better than the real messenger . . .
09:08████Are we still allowed to download any apps we want onto our work phones?
09:08████I cannot - maybe it is a new policy? Or just a bug?
09:09████from the app store? yeah i think we're able to..
09:09████Yeah I cant lol
09:09████oof that's probably a bug then
09:09████Yeh thats deff a bug i dont even have a personal phone
09:10████you can contact techsupport@
09:10████If you need us to take a screenshot for you, lmk
09:10████feel free to add me to IG and send me bs messages too
09:10████Haha so far I only have a message from my test account :)
09:14████Yeah my work phone has some original issues (MobileIron not working properly) that I have tried the 3 different ways suggested by tech support to fix it & none of them have worked. Maybe this issue is related. Will contact tech support
09:15████Would be good if someone else could add screenshots for slide 33!
09:15████(There are 2 versions of the active now unit - one has stories merged in with the 'chat heads' - ideally we dont want that one - it lead to a decrease in sends but was being used to boost poor stories numbers - but if thats the only one you have then I dont think it matters)
09:16████Also has anyone looked at time spent on IG varying throughout the day (By local time)?
09:17████shared: 172154219_747480999255259_145976926424625_n.jpg
09:17████My numbers looking at things during the day are looking a bit weird and I didnt find anything on workplace - will debug today but would be helpful if you guys already have something
09:18████sounds like something Junia looked at :) at least in terms of sessions
09:26████indeed! here's a slide: https://fb.workplace.com/photo?fbid=862946417611426&set=pcb.753698851946804
09:28████shared: 169599777_1128519440907100_8326729633641490664_n.png
09:28████shared: 171317905_476724750137031_400771353584906162_n.png
09:29████various versions of timezone adjusted sessions over the day lol
09:42████@silent fyi does look like Remix was cannibalizing Stories Reshares (therefore net negative): https://fb.workplace.com/groups/2208900212692314/permalink/3023937387855255/ Makes me feel like the use-case was pretty different (it was kind of a stories reshare with more friction), the new slide from Junia on visual replies I think fits much better!
10:03████Still very weird though to see a cannibilization ratio of > 100% (Usually between 0 and 100%) - especially on a metric like DAP
10:03████Basically it implies something like - creating a reshare yesterday was so much extra work and friction that you are going to bother less with stories altogether today
10:05████I think they were trying to avoid ppl doing "Feed->Stories reshares" without adding any "original elements", something like forcing people to add a picture and or high-intent stuff on top (vs. just resharing the initial post). So in a way I think this feature was just adding some friction to the Feed->Story reshare flow so I'm not too surprised that the overall Stories DAU decreases!
10:07████so it killed F2S reshares a lot, but the subset of users that did end up remixing got categorized as Original, which is why we see the lift
10:07████Ah - it changed the original flow as well? Then that makes sense
10:40████FYI adam just said in Q&A that we're "wildly underinvested in direct"
10:40████aand is quoting our stats
10:42████He quoted the notif stat, something like 60% of notifs are direct. i think that was from my deck 1.5 yrs ago lol
10:42████he also said the 3X bigger for teens stat
11:35████Hey guys, figured out why my queries were giving weird results when I look at the time of day distribution - might be of general interest. Turns out the the IG sessions tables I was using - which are the official ones - extensively re-use the same session id for the same users. So if a user has 10 sessions over the week probably 5+ of them will be given identical session ids. (The IG sessions tables dont group by user_id, session_id, ds or similar so they still appear as seperate rows in that table) but part of my query was grouping by sessions at some point.
11:36████I checked the impact on my results I added to the deck - just due to luck in the way my queries were written the impact on those numbers is < 5% so I dont think they need updated
11:36████But I wil reach out to the IG DE for the sessions data domain because that is a pretty bad table design IMO
11:38████Haha - I wasnt expecting to find so many problems in your tables!
11:38████it's not our fault :(
11:41████we keep the same session_id if users switch accounts during a session (same app_device_id, different userids, same session_id) -- although you're saying the same userid will have duplicate sessionIDs over the course of a week?
11:49████Yes correct
11:49████The same session_id appears again and again
11:49████shared: 171580239_204443134449215_9206330877762746610_n.png
11:50████Straight from time_spent_session_metadata:measurementsystems
11:56████Hmm or maybe I did something wrong actually - like inserting into a table might have accidentally cast them as the wrong data type or something
11:56████TBD
11:56████Wont post anything unless I am totally sure it wasnt me!
11:58████Nope it definitely isnt me! Its the tables!
11:58████Try running this sql for example:
11:58████[Redacted - Source Code]
11:58████In the measurementsystems namespace
11:58████Will make a post about it later today
12:11████shared: 119226974_370708470600099_2090971459120095006_n.png
12:54████Im second guessing myself
12:54████why is the 44% stat not relevant?
12:54████(IG reply rates are 44% slower than msgr)
12:54████Feels like a relevant supporting insight to include no?
13:02████I wonder if it could fit in the "competition" section as add'l rationale for why we don't have "primacy"? it's supported by research showing that Direct is less urgent than snapchat/imessage (messenger is likely comparable for adults)
13:02████we could prob dig that research up
13:03████My understanding was that someone else (I forgot who you said) was already going to mention that stat? And then you were asking about including the additional CI research about response times?
13:04████I was skeptical because reported - perceived - response times could be very different from the 44% stat. Both are great data points but could be due to very different things. For example the CI research is probably around Direct being low stress or something - whereas the 44% number might have a more technical explanation in terms of notification types or whatever
13:04████i think they are related but a bit different
13:05████Yeh I think we could add it if we find it, no biggie tho
13:05████WDYT about verifying that stat and including it? worth it? not really? Reason I ask is Vishal has seen it and vishal sent this person to speak to me, so im wondering if he sees value
13:07████Up to you guys. I think if it is true it is probably some kind of mix effect (Like Android users in India are super responsive or something) but if it is true in the US and also true for US teens then I think it is an important data point and improving responsiveness somehow could be a promising area of research
13:20████Greg, can I say this:
13:20████shared: 170541771_777847189837253_4993314776182537414_n.png
13:27████WDYT?
13:27████shared: 169968338_248525163652449_7085981809494916575_n.png
13:27████sry in interview :)
13:27████ya no rush
13:30████looks good! for bullet 3 -- could it potentially be taken the wrong way by Messenger org in case it gets shared with them?
13:30████idk how ~ diplomatic ~ we need to be
13:31████Good point. Personally I would prefer to just be blunt. It is actually pretty ridiculous IMO that FB doesnt have a single person full time dedicated to direct (That is what my messenger contacts tell me) when it is such an important surface
13:33████Yeh I think we keep ti like this for IGLeads
13:33████We prob remix it a bit when we share it out
13:33████the main point is to land it with IGLeads so they can deal with it
13:35████Do you just copy & paste charts from excel @ @ - mine seem to lose a bit of resolution when I do that?
13:37████screenshot works better than direct copy/paste
13:51████Feel free to tweak you're the boss lol, F&F moments just feels a little confusing to me as we're mostly talking about CF it's a bit weird to reintroduce F&F
13:52████"Messaging is the only medium that is still growing its share of close friend moments on IG"
13:52████is that accurate?
13:53████James r u like "MY GOD IS IT LIKE THIS EVERY WEEK"
13:53████the answer is no :P
13:53████Haha a little bit
13:53████this is like twice a half i think
13:53████just for IGLead prep
13:53████Maybe Messaging is the only messaging that grew in share of close friend moments over last year ?
13:53████wanna make sure we have the suit n tie on
13:53████I'll be logging off in an hour (Well actually now & then come back soon for 1 hour) - but I trust you guys have it under control! My part is pretty trivial anyway :)
13:54████only medium** :)
13:54████(just because idk about the very recent days so idk if it's fair to say "is growing")
13:55████We're sure the FnF stuff isnt a logging bug or smthn
13:55████lol
13:55████I remember they changed the definition of FnF at some point
13:55████i think it was before lat year tho
13:56████shared: 171108926_458946405220499_7741511167623857271_n.png
13:56████Ah it was 2019
13:56████that is feed imps by fnf vs not
13:57████june 2019 there was a cahnge in def
13:57████but yours is looking at 2020 vs 2021 right Greg?
13:57████Ya i had the trend and looked fine!
13:57████awesome
13:57████shared: 170773545_154553416565303_2065219806704702256_n.png
13:57████MERCI!
13:57████DE RIEN MONSIEUR!
13:57████crazy chart
13:58████shared: http://www.facebook.com/1771598439878152 as 50366997_2300799350138882_4569562478131806208_n.gif
13:58████shared: http://www.facebook.com/1732905006870710 as 62449184_2316247935298279_5043837218204942336_n.gif
13:58████shared: http://www.facebook.com/232021058716423 as 50189975_415471175864065_9018811020948275200_n.gif
13:58████my god imagine a french will ferrel
13:59████will ferelle
13:59████is that an appropriate translation greg
14:00████That sounds correct oui!
14:00████I feel like not only are you a better analyst, but you've also gotten funnier over the last 2 years
14:00████William ferelle
14:00████.... thank... you..?
14:01████or maybe you've just increased your joke PR
14:01████used to be PRIVATE SHARING the jokes with Shnaa
14:01████soon BROADCASTING in IG data FYI
14:04████no yall have just become labeled as my CLOSE FRIENDS
14:04████it's lowered by BARRIERS TO SHARING
14:04████my
14:28████im gonna do a final read through now in case anyone wants to join
14:29████https://fb.workplace.com/groupcall/LINK:xtzhjmqarmta/ @silent
14:30████no need to join if you're sick of the deck
14:30████i'm in an interview!
14:31████clear no hire
14:46████It looks great to me lol I TRUST YOUR JUDGEMENT!
Only wondering if we should copy-paste the Product summary table just before enumerating the "what should we build" in case they already got overwhelmed by the first section and just want a summary of ideas?
14:50████ah yeh
14:50████good call
15:53████Some hopefully self explanatory charts! Enjoy your weekends!
15:54████shared: 170948991_463843588283858_3268126228045202760_n.png
15:54████shared: 171520814_488262745636174_1098131415548014070_n.png
15:54████shared: 171847436_302825771509673_223209435581480611_n.png
15:54████shared: 171406940_479540653236681_1408375484704792879_n.png
15:54████sick
15:54████Honestly the only insight I see in these charts is that teens are really into using IG at 11pm when they probably should be sleeping :(
15:56████Have a great weekend man, awesome first week
15:56████I guess technically next week is my first week :)
15:57████Great Week 0
15:58████im gonna add these to the deck they are cool
16:08████Gooood weekend [emoji]