Deep & actionable SKAN 4.0 breakdown with Eran from Singular. What to expect from SKAN 4.0? (#55)

[Music] hello everyone uh this is no gaming podcast two and a half Gamers um special scan session today so we are sharing actionable insights dropping Knowledge from our day-to-day user acquisition game design and AD monetization jobs so today it's only day to-day user acquisition job we're definitely not discusting latest news uh but we are having so much fun so let's not forget this is still 4: a.m. conference discussion Vibe so let's not take it too seriously so again as I said this is special scan iOS session with Aaron fredman so as you know as you might notice I replaced Felix and yakob with Aon because well he's definitely more Su most more suitable for any iOS or scan uh discussion and it's me Matel andich so eron welcome thank thank you for having me m

and yeah I'm HED to join as the one of the players uh there one of the gamers and excited to uh talk about no insights for scan nice yeah yeah perfect that that's what we need because we uh so we have all the all the Articles out there talking about how scan 4.0 is coming nobody knows when maybe we know already let's see and you are kind of you know in the in the let's let's go back Channel because well why why don't you just um give us some like quick introduction from your s uh who you are how did you end up in gaming as well that's it's always import important and interesting to know yeah for sure um okay so first of all uh my name is Iran I'm the CTO and co-founder at singular um you know we started the company almost 10 years ago now as a nice yeah so that's been a while started as kind of the first cost segregation solution in the mobile industry as we C it

originally and then very quickly we got into the MMP space we wanted to kind of build the ultimate unified solution to provide all the UA data from all your networks with all your attribution data have in single platform um you know we're uh ranked uh just recently in G2 as the number one product for MMP in the uh in the category and we work with amazing gaming companies like nice EA and glue and red games and Rio and Warner Brothers and many many others of course and we're kind of putting a kind of a big focus on Next Generation uh attribution methodologies like SK Network and privacy sandbox and Med mix modeling and all of that jazz um then so that's kind of I guess where we're are today and you're asking like how did they get to the gaming business so honestly I started from kind of the mobile industry in my career so before singular me and my two other co-founders G and Susan we all worked at a a subscript I app I would say not exactly

gaming but the consumer app calledo that was way way back when this was 2009 so really when the mobile industry kind of exploded um so we kind of saw the challenges around growth and growth data from there and I think that's what inspired us to uh start singular um but yeah I mean since then we've like been we've had like customers who are also uh uh gaming as part of like Theo we had like a small B2B business that we're interacting with them and then we kind started a new company and want to focus on you know solving growth data challenges basically so I guess that's what brought me into the gaming industry nice okay that's pretty cool that's pretty cool and uh so why you are here so as soon as at happened and all these privacy changes happened singular was like like one of the most vocal mmps out there so I I remember you were like always talking about what's what's happening just sharing some some insight and some know uh so this is good I I I

love it and I have some some really like let's let's call them direct questions because um I read a lot of articles but I still getting I'm I'm getting also questions like okay so what should we do when should we start uh using the scan 4.0 before we go there can you just maybe quickly uh give us some like quick state of the iOS like what's happening now of course everybody knows like it all this fun stuff but like now okay so we need to move from scan three to scan 4 do you have any maybe timelines or something that you can share you can you know from the from the back background and also maybe I I'm I can ask after all these iOS fun stuff privacy send books as well you mentioned it so maybe we'll go there we'll see we'll see we'll see but what you what you are seeing now from the like the iOS point of view okay so let's uh take a quick kind

of step back and what's the state of the industry with iOS at the moment and you are already familiar with the at changes you know the lack of idfa or the you know discussion like probabilistic veristic and all that things and then there's SK Network and again probably everyone is already familiar with the fact that Apple kind of designed SK Network to work across all iOS users without requiring the ATT popup or without you know demanding kind of either they give consent or not doesn't matter everything works on a SC Network and that was kind of the initial you know they originally released it like esan from 2.0 like two years ago with a lot of Dramas as you've mentioned and yes we're pretty I think vocal about the need to kind of pray for it and they've made changes through time they've added like published new versions and added more features like esan 2.1 and then 2.2 and then 3.0 like things have evolved since then and if I kind of fast forward

to now or more specifically to like November of just last year then suddenly Apple released the new version of SK Network called SK Network 4.0 or scan 4 short uh it was actually also kind of a surprise a bit of the industry because there like always would happen you know it's fun so they're kind of giving teasers on kind of it's gonna happen it's kind FL going to have some new fatures but then fin out of the blue they said I 16.1 is out and happy surprise it includes scan 4 in it and tons of documentation of changes and differences and new features and all that so that was just before the holiday season so it was kind of great uh you know great kind of interesting timing and of that so uh that was of again November of last year um so then when you asking okay so okay so what's the status right now right now we're in February 2023 okay so what's exactly the

stat so the status first of all scan for is out it's part of iOS 16.1 and above um it's important of course to know that uh because it depends on the OS version then you need iOS users to update to iOS 16.1 so they have scan for supported um already based on kind of the adoption rates that I see so we see check what kind of uh iOS I have at the moment because I'm always the last to yeah of course yeah 100% yeah I don't know what 6.1 okay 6.11 okay okay there you go so you're I guess there you go yeah so yeah it's definitely growing um based on our data we see that roughly 45% of iOS users have 16.1 and above uh so it's still not like the masses but you know it's pretty me for almost half yeah yeah yeah pretty decent right so you know roughly one of every two

kind of iOS users would have scan for builtin and it's increasing like pretty fast like from November it got to 40% we expected probably in the next I don't three months or so you'll have like probably like 80% of the users which is kind of what you usually see in these timelines um so that's kind of one part that you need for scan 4 to work yeah but then of course there's uh like there's multiple endpoints that need to work there's the M&P side to support all the like infrastructure for that advertisers the measurements uh I think by now all the mmps have released your scan for support and announce these yeah you also the you were also the first one right like oh well we have we are live on scan 4.0 but then I was I was reading the the fine print at the end like but you know like we are ready but it's not adaption rate is not uh very big but we are ready so as soon as every everybody else is ready you're good you're completely right you know it's the notebook podcast right right so

we basically announced that was in November right just shortly after the uh it was on and honestly we've been preparing for that kind of release by Apple so it's not like we Sally developed everything in like two weeks but bring out the ini things but yes you're right that the stages basically means that we have a scan for SDK that the advertiser can update to be ready for scan for it doesn't mean that obviously they can run scan for at scale because you need all the iOS users to add to scan 4 and you need the networks to support scan 4 but this was really important for many of our customers because if you can imagine I've mentioned some of the kind of Brands working with singular kind past like War Brothers in EA and these it takes a while for all the studios to update right for these big companies I'm pretty sure yeah it takes ages to get out with with the new exactly so it was really good news for them to understand okay how they're going to involve it in the release Cycles to make sure that once can for is actually live and like

production level they can be the first ones to kind of test it out right I think that was kind of the intention there um and think if again fast forwarding to now from November now we're actually seeing so again not production scale in scan for but we already see life PST backs some some ad networks are already testing it out and we're kind of coordinating with them we have some live advertisers customers who working with scan for actually seeing the data which is pretty exciting honestly yeah and you know starting to look how it's working in practice even if but cool yeah the question is like what's the what's the advantage of being the early adopter of all of this because it's so complex it's so complex y completely agree and you know it's a good question to ask not all of the advertisers and probably even not most of the advertisers it would kind of fit for them to kind of be an early adapter and like be a first mover but again you can of uh you know in the in

our industry see kind of various types of advertises some say you know I have like a I don't know an existing operation and got it to work in scan through or whatever I'm not really using scan which also happen to say yeah I prefer not to like test all the new features I want to wait for things to stabilize every want to kind of Ras everything give give it another like six months or a year and then kind of I understand I'll get all the learning through all the blog post and do what everyone else is doing right so that's definitely a a legitimate strategy and makes sense for maybe most of the advertisers but we do have some of our customers who actually want love being on The Cutting Edge of Technology right they want to be the first ones to test this they want to get the advantage they want to we know that there's going to be a learning curve like there was a learning curve for the initial version of scan and many were starting for a lot of time we saw that honestly the ones who started early got the struggles like everyone else like trying to figure it out but honestly they've uh now they feel I guess they got to a better place much

much earlier than others and they it works to their advantage means they can you know buy uh idfa less inventory which is usually cheaper than Ida based inventory yeah and they feel that they can optimize there right or like do things that others can't do during the time yeah that that's true that makes sense like I mean uh I'm usually not as as early adopter as maybe I would like but uh depends on what kind of companies I work with of course if I work with smaller companies not everybody has the resources and everything to actually correct like what does it mean so it's like I'm sometimes I'm play playing the the wait and see game but so you you you talk about testing so is it possible to have Scan three in place and then start testing scan four on let's say partial traffic yes so the answer is for first of all yes and it's also necessary

because we understand that it's going to be a gradual process you're still always going to have like even when scan for reach like Mass adoption you have like 80% users in scan 4 right you're still GNA have like the 20% of scan 3 so even during the time you still need to maintain the scan three models in a sense now Apple was pretty smart about it they kind of put various uh kind of Backward Compatible features in every new features that they've included scan 4 so anything there or like almost everything there can kind of be equivalent to scan three just like let's features in scan three in a way okay so uh We've also designed our you know solution to basically have parallel scan three and scan four so that if you have a scan three user everything is going to work based on scan three of course and if you have a scan 4 user then you'll get kind of the additional features for them definitely complicates things but it's probably going to be like that for a while yeah so that's what I I always say to to

people in in gaming industry like it's not going to get any easier it's going to get even more even harder these days it's like but that's how it is you we just need to adapt but what's the then so we are already kind of comparing scan three and scan four so what's the biggest difference between scan three and scan four h wow what's the biggest difference between Scania and scan for um yeah so I think you know there's a lot of thing that I'm personally excited about let's so there's like several big uh you know differences like features I would say so uh if you think about it there's like the web to app support um you know which is really interesting because think about like until now they're like UA channels which were heavy from web that couldn't work with scan just inventory not coming in scan can't be measured in scan I think about search campaigns like I don't know there's like all kinds of you know ad netw works with heavy search traffic that you can imagine that maybe would be more based from browsers

basically and couldn't be measured by scan and it's possible which is exciting on its own right there's the there's features for getting like more granularity so kind of getting deeper reports it's for higher scale advertisers but potentially they can get like adset level data Maybe creative level data Maybe placement level data so that's kind of interesting on its own and there like the course values that we can probably talk about but the thing that the most excited about among other feature I think has the biggest potential for being a game changer is the multiple postbacks uh the fact that now for every user you can get up to three different postbacks different uh cord CS I think that's the most exciting ones uh for me okay so what are what are the different post BS and how does the the V like the conversion Windows look for these post Pikes because I was reading all these like uh first log like hm okay this is nice it's great but then

like right now I'm used to see the the conversions and Prospects pretty quickly so we are looking at day one so I can see very quickly everything that's that's happening this it kind of extends to eternity almost so how does that look how does that look like because now it's it's not one only so that's going going to be free right so how does it look like okay um I me it's great yeah no it's yeah so I think yeah you know if we kind of question wait is it good thing it's a bad thing definitely there's no question scan 4 is definitely an upgrade from what we had in scan three um but okay but how it actually looks at so let's uh talk a bit about like the timer mechanism like the the post backs and and all of that so until basically until scan version three you only get single post back right for every attributed user um and that could and if you pass kind of you know a certain volume then you know so get like

a conversion value and all that so we're kind of understand like the Privacy TS and all that so essentially if you have a conversion value then you can ideally calculate like your Ras for that campaign right so you get kind of that post back and you can calculate like the encoded Ras so if I try to uh um kind of make an example let's say that I know a user made a purchase after they inser the app and kind of paid the game for a bit and they like made a purchase like $5 then it's encoded as a conversion value five and includeed in the post back and you can calculate your R which is great but the big issue and in my opinion in my personal opinion one of the biggest issues in SK Network until now was the limited time window you could only yeah basically you there's like we can get into the technical details but the standard was to only uh record what happened in the first day of the game from the first launch right so you have 24 hours to see if any users made any in a purchase or if they had any it signals you know meaningful event

that makes you predict that they'll probably make an a purchase later on right yeah but uh as we probably all know in kind of the gaming industry especially for ones who uh rely a lot on in purchase there's barely any users who would be willing to make a purchase already in the first 24 hours right um and that's a big problem that's like a huge problem for scan mean for like for like mid for casual yeah majority of games that's a big issue um so um so it's true that you get a postback pretty early on but if you get like you know 90% zeros of like you didn't make a purchase and then like barely working with others it's like how can you understand the True Performance of your campaigns right so that's been and honestly I've heard from adver kind of saying yeah sitting on the fence and saying yeah not even sure if I can if I should invest in scan in general because I can't really see anything from Beyond like one day right so that was kind that was that's also only um like way of looking

at things obviously you want to know like what's uh what kind of performance you are getting from the campaigns yeah well in these days especially on iOS it's not going to be that easy anymore so maybe it's kind for okay I'm I'm going to well change my ter terminology about these things but then it's kind of you need to change the the way of thinking about the valuation of the iOS campaigns and of course like if you can't see then it's really hard to kind of justify the spend of course but then you have different approaches to you know to look at things like Blended approach of course if you're small company money goes in money goes out of course it's not ideal it's very far from ideal but what can you do yeah but yeah so if you're running you know with like 10 channels right for a large scale Advertiser now suddenly the Blended metrics are more difficult to what can you do but now that's the big news in

scan 4 you know because it's you know it feels like to me from my perspective it feels like apple understood recognized this issue and they came up with like a new a new proposal and basically said yeah now you're gonna have three specific windows and they chose the windows themselves said first post back yeah which was kind of interesting yes of course so said first post back sent after two days with some random there we can dive into the details then second post back after seven days some random and then third post back after 35 days and those can have a course conversion value themselves so now if I try to again like look at it think about it as from a gaming kind publisher standpoint now literally if I had any in a purchase in the first 35 days of playing the game which sounds more reasonable from many of the gaming companies out there now then s you can actually include you can know exactly how how many users made a purchase and whether this was high

purchases or medium purchases and really see kind of the like d35 R right basically not very different it's not like I'm saying like D9 but d35 yeah it's exactly exactly um so I think that can be I know also from speaking with our customers that's that makes a huge uh impact right on their analysis so I think that's that's kind of the big uh news there of course always like I tend to say like with with SK Network there's always good news and bad news so the good news are that you you you're getting three postbacks you're getting two days seven day and 35 days that's great the bad news are that the second and third postback have much more random to them so the post back is sent after if the first postback was we used to kind of get it between like one and two days basically of random now the second and third post back we can get it after between 1 and six days of random

again if the user can of made a purchase after seven days and you get that conversion value then it can take up to another six days until you get that post back which is pretty insane but you know that's what you have to work with yeah true so we need to know uh think about how actionable is this actually for the U managers because of course it's not only day one but then like if if I need to wait another like six well it's not only six days 12 days and then another x x days to actually see the the PO back well I need to be able to make decisions very quickly because you know I don't want to waste money I'm pretty sure that you are hearing this from from other advertisers as well so any any anything that you know can help you actually lock these windows or or um the conversion values or that because I'm pretty sure that there is something uh something in there it can't be just Doom and Gloom on

for sure yeah um so yeah there's also the locking mechanism that I can dive in in a second but I think if you think for a second about like the UA kind of methodology and practice and even before like scan and IDE PRT right the gold old days right usually you start the campaign yeah you can already in the first like few days probably you see like what's the initial performance right you look at D1 D2 D3 kind of performance and kind of make if you see like it's totally trapped like this campaign like not performing well you can decide to stop it if you see that it has some potential you probably you know maybe make some adjustments but kind of still monitor right you want to see what happened in day seven and day 30 and so forth right so kind of you r on the first initial indicators to kind of make more real time decisions or like daily decisions but still monitor what happens eventually right with with that campaign so if we think about it that way it's I don't think it's going to be very different now with scan 4 because you

can imagine you can still rely on the first post back we're going to talk about even deeper optimization with lock windows but if I don't change anything I can check what happened after two days okay how much information I got from from that like how does the initial conversion look like if I get like I see like a full conversion value I can see the fine conversion value I see a lot more details there and say okay if that compan seems to be performing well or not then you can decide to scale it or not but then after seven 35 days s before S as the old days you can take another look at that campaign for that install date and see okay did that campaign really continue to increase did the users really make a purchase as I thought that they would or did they increase in value as I thought that they would kind of validating your assumptions in in a way I think that way of that process is still valid in the new days it is but then how how the let's say reporting changes afterwards because now I'm I'm used to see the one but then and well yeah how how did the

reporting change uh thanks to the the different conversion right or post so think about yeah we when we're trying to think about our reporting for our customers and how to make it as easy as possible we're everything is already so complicated and that exactly yeah yeah so we're trying to okay keep it simple for the customer so how it would show up in the reporting is what they can see is already they see the day two basically performance or like the first postback performance which is essentially day two and let's do you short in that window you can see day seven performance you can see day seven Ras basically and then you can see day 35 Ras right so essenti anything in between or only day seven and day 35 so directly from the postbacks you can get like the exact window but if you get a bit more sophisticated you might be able to see a more things in between um and we can

talk about about the Locking windows and all of that but okay um but uh yeah so essentially just to kind of to simplify you can see kind of the reporting you see the Deep gr as much as can for will offer and then they two day seven day 35 and we saying wait but I want to see day 10 that just is an example right day three for examp that's honestly more difficult with scan uh today um you can do what you can do is you know one feature that was kind of included scan for you all the hidden surprises was the locking mechanism the lock window uh which is really interesting and cause a lot of room for discussion and kind of brainstorming in the team and with the Media Partners out there so the lock window gives the option for The Advertiser to decide that they don't need to they don't want to wait for the end of the window they want to shorten it like in real time say okay I'm done I'm not I don't need to wait for any more information about the user to send the to send the post back so just as an

example let's say a use case would be if the user made a purchase I would regard it as a high value user just as an example right because I'm looking for the high value user now so if the user made a purchase already in day 10 then I'm going to lock the window and send don't no need to wait for day 35 you can already send the post back to the network right after 10 days the advantage would be that the ad Network would get the postbacks ear on they don't need to wait 35 days for the third post back the disadvantage is that it becomes more difficult to understand what's the install date for where this post back belongs right so let me kind of try to Jes Christ okay so yeah as you said so there's good news and bad news sake ah okay exactly so if you you know if you basically say yeah I'm going to send the post back whenever the user made a purchase so the user can make a purchase

anywhere between like day eight to day 35 right the third postback so how do you know that literally you have like a range of like a full month from where the user kind of belongs right which kind of create a problem so I guess my point is there's a lot of potential there's things that you can do maybe around modeling which is something that we're heavily investing in like trying to figure out those things um the at the end of the day we want to provide like the install base like when did you spend money what were your results in as many cohorts as possible in kind of as real time as possible with the limitations that exist there right of course there there are going to be limitations so so okay so a user installs uh the game today he or she makes purchase in well say first 24 hours is the first post pack then uh this player makes purchase again in five days and then again in 20 days so how can I track this evolution

of his LTV in this like new can 4.0 right so um the first thing to kind of keep in mind we're trying to answer that question is Apple made it very very uh um hard to track that specific user and understand okay that's I got a specific user who made the three purchases right yeah instead they're kind of thinking they say thinking from the other way around uh for this campaign how many users did you get who made that amount of uh you know purchases right so if you try to sum up that this is kind of one kind of classic way to approach this is if you sum up those purchas three purchases within like you said like 20 days sounds like a pretty good user right especially if com first already in the first 24 hours they already made a purchase you know so intuitively I would mark this is probably a pretty high value user right that means okay you got some good a good user there so probably what I would do is create a model a schema conversion

model schema basically says and again if they made more than a total amount of purchases I if it's all Aggregates to like more than I'm thring out like $10 or if they made more than one purchase for example during the time period You' probably mark them as a high value user in the third postback and if if they made also a purchase in the first day in the kind of the second window then they probably be considered a high value user in the first day and in the first post back and second PST back so what you'll get in the reporting is you s show that you had one install in theory and there are high value in both like D2 D7 and 35 and you can probably also model the the RWS basically based on that kind of estimating okay how much average on average revenue you got from high value user right in C day7 and d35 so you're using some estimations to say okay on average I think because I saw that you had like a one install with like high high high there then probably the d ross there is I

don't know like 10% Like 30% or whatever exactly so 60% let's be more let's be more positive let's be more positive 60% yep exactly okay so so but how what you can expect based on the yeah how do you take those players like you said like I would say this is a high value player okay so how would I do that so what you would usually do is like create a schema in uh your MMP like singular and kind of say okay so uh so again high value users are equal kind of that if a user made more than that range of value then I would consider Highway user that's encoded kind of in the conversion value and then you see from this specific campaign okay I got the number of high value users and that's that means that the RW is 60% now you can try to make decision about this campaign is it e in your target do you expect that 60 is 60% a good value to me s pretty good for that uh for that range

then you can you know consider okay let's continue scanning this campaign looking good right but if you see that they not hitting the target you're barely seeing maybe you didn't see any high value user okay seems like something isn't working there right and we should probably kind of considered like other levers to optimize um so that's kind of the level of thinking I can imagine yeah okay so can I have a like different conversion schemas for different prospects then in the four 4.0 definitely yes how how does how does it work so yeah just to make things more complicated of course exactly actually so you actually have four different schemas if you think about it because you have okay the for the first postback you can get either a fine conversion value which is the type of conversion value that we have today the famous number between 0 and 63 so that's that's supported only in the first post pack so you basically you probably keep the same model that you have today for scan three that's probably going to be

the fine conversion value for scan four for the f post Bank uh but then you can also configure a course conversion value schema for the first BS back so for people AR less familiar with all the block post about scan 4 kind of the new type of conversion value that has more limited information not the famous number between Z and 63 but uh just basically a string just low medium or high just three options okay so basically if you don't past certain thresholds the Privacy thresholds basically to get the fin conversion value you might still be able to get the course conversion value so that's another model that you can potentially configure in the first postback so you need to think about okay so in the first two days of the user like understanding how to map these two numbers and also how to map this to low medium and High um so that's kind of the second schema that you need to configure and then for the second and third post BS you also need to configure two addition schemas what's considered a high medium low

value user after seven days and what's considered a high medium low user after 35 days which might be different mechanisms right you might say that if a user made a $10 purchase within seven days they're considered high but if they didn't make any other purchases by 335 maybe you'd say that they're a medium value user right so it might be different rules based on the different time windows that you have Okay so let's say let's say I have the same conversion schema let's let's call well let's let's use your Revenue schema so first first uh 63 bits it's all revenue brackets then for the first uh post pack I have another brackets for let's say low medium and high so I can go from I don't know let's call 64 to whatever 200 because now I know that in the next seven days from the day one players can spend anything between 60 and $200 and then from 7 Day to 30 days they

can spend another $500 so I can put all the brackets here from let's say 60 to 500 yeah okay that's definitely one possible schema that you can definitely configure um okay there's more there's like T you can easily spend like an hour in like converion optimizations but I'll give you like an example yeah if you think about it again in the course side usually there's a we see that there's huge effect of whether the user actually made any purchases versus they didn't make it out because obviously in gaming like most users don't make any purchases yeah right yeah so kind of a recommended practice that I would say is that like this is a big factor like it's more it affects more whether they even made a purchase or not rather than kind of what's the exact amount that they made yeah I would say probably as a default you would say that if the user didn't make any purchases assuming we were talking only about Revenue here then you're probably considered low and they'll probably be low like also it's

going to be pretty consistent after 7 day after 35 days low value users and that's probably most of the users that you're going to get right didn't make any purchases they're a low value user and then on the revenue ranges okay so how do you split like let's say the user made a purchase how do you decide if they're medium or high then that kind of the depends a bit on how you regard you know usually if I try to think about the the high is kind Reserve to the I don't the way users in a sense or like the really high value purchase by users and you can probably create like sort of like a distribution um kind of metric you see like you know how many purchases you usually get per user and kind of try to split it somewhere in which you regard okay if I get you know just like I'm just throwing a number but if you get like a very low number of high value users but they know you know that they're very high value yeah it means that you probably want to continue spending on that campaign so you need to kind of think about what's that number

that makes you confident that okay wow I'm getting some pretty high value users here it's worth like U um working more with that campaign it might be that $100 Revenue it might be like a $50 Revenue depends on the verdict that right sure sure sure of course that depends on the runner of the game as well okay so in that case like how that influences the the campaign optimization on the on on the u channel side because now I'm I'm used to see let's call let's let's talk about Facebook so I'm sending all the purchases uh from the day one into the Facebook with the with the value and then I'm pretty sure to see that uh as soon as I send as many events as possible the campaign gets more more events more data points and it kind of optimizes better but it's it's day one only so if I'm going to send the day seven or day 30 like how that's going to change the the whole well optimization

on the campaign side yeah are you already talking about that with the with the channels oh yeah oh yeah we're talking with man there's a lot of conversation there because it's a really problem right yeah they're like they're also trying to like figure out how how are they going to haveit on their optimization but I think again the good news is that it's not like there's some nuances or the timers and the things that have changed but they're still going to get the same first post back as they used to get so yeah in terms like the current optimization it seems like it's not going to remove things things are only going to get better because now besides the optimization by the first postback suddenly they can get more information that they on of course that kind of reaffirms if they're on the right direction or not right so if I had to guess and I think every media channel their own kind of

strategies for optimization they're trying to everyone is saying we have to test it at scale we have to see how the data looks like and all that there's tons of Sil questions on how it's going to work at scale but I think the common kind of way of thinking is we can keep for now still optimized by the first postback as exists today yeah but now we're going to get a bit more signal on you know later on on how did we truly perform and everyone has like their own kind of you know prediction algorithms trying to understand from the initial signal whether they're going to perform or not now we kind of have like this um validation right after seven days or after 35 days if you're really right if it is if it really looks like it's uh you got it right and I think that's probably going to be like the first phase still optimizing by the first signals but getting a bit more information to see if you're right or wrong uh and from there we'll see how they they can innovate and kind of update their optimization accordingly yeah I'm looking forward to that it took only like almost alternative for the U

channels to actually adapt what's what's happening on the Iowa side so it's going to be very very interesting but I'm looking forward I'm always very optimistic about uh any any change in terms of the iOS uh but how how can developers actually prepare for for this like any anything um for those that you know updated the SDK but should they do anything on their end anything else any checklist or anything that you know we go through it's like okay so I need to do this this this and that and then I'm fine I'm well prepared yeah um so there is a checklist so let's uh think about it for a second so first of all yes you need to update the the SDK that's number that's the most important thing and you don't have to do it like literally like today like shout everyone need to make sure it's part of your plan to prepare right so it's usually those release cycles and making sure that it's somewhere in the backlog of your Dev team uh so that's

the most important thing so once any UA Channel says oh we have scan for support then you're you can test it out fin yeah good to go that's number one exactly the second thing that you can already kind of prepare things is think about this is more planning so think about how do you want to leverage both the course conversion values again how do you define the low medium and high users and the multiple post packs so how would you define those users for each one of those time windows it's as if you need to create three new schemas you already had to think about one schema now you need to prepare find what do you think are the next kind of three other schemas now I get a lot of questions from like Advertiser like what should be like the best uh core schema for me I have no idea and you know it's obviously something that we've put a lot of think thought around and basically our product offers kind of out of the box schema so we can use your keep your same scan 3 Model that maybe you've iterated and got

to something that you feel really confident about and kind of click a button or like you know upgrade to scan 4 and we'll have uh our suggestion for course conversion values for our three models so you don't need to you know like obsess about like what's the right schema you have something already available that's kind of align with your scan three model so for example had a revenue uh model then most likely you want to keep a revenue model for the other course values for example right so with kind of these practices in mind so uh again if you kind of want to be more I don't like you want to think about it more you want to plan more for it you can think about that that question if you want to customize that uh the schema but if you prefer that you know to start with something and then see from there you can just choose the out of the box models so that's kind of the I guess the second part of like the checklist plan your conver your schemas basically yeah then the third part again if you want to

be like the first mover like early adopters is you can actually speak with uh your media uh Partners right just ask what's their how how they're doing with the progress towards SC for now we speak with all the partners each one are in their own kind of process some of them have better programs already that some of our customers are using and testing side by side others are taking more slow SL so it really depends on the media partner but you know if you're working with someone you're spending a lot there you can ask them when are dep planning to be ready with scan 4 maybe they will tell you you know what we actually already have a better program we'll be happy to test it out out yeah yeah exactly so that's kind of I guess the initial checklist that I would uh okay okay okay okay now it's it's going to be a lot of a lot of fun uh I guess yeah well as you said at the beginning I guess it kind of steel is like three months out uh until like 80%

of the the users is actually having the the 16.1 um iOS there is some time but I guess the thing is like I guess you're never going to be like 100% prepared for what's goingon to happen you're right and honestly I I tried like so many conversion schemas before um it's just like I was trying to be the early adopter and it it took a long time until I figured it out so now I'm I'm in that like wait and see kind of mode uh but it's already I guess uh time to start looking into things right yeah I know I think it's a definitely a I think it's a very valid strategy to kind of wait and see but uh it depends on the you know again many are trying to use that as an opportunity to get an advantage early on and kind of create the expertise nobody really knows how it's going to work like the scan for so

if yeah so if kind of if you feel that you want to kind of take the leap and yes you probably need to reiterate and kind of try to learn it and make mistakes for sure but you you kind of gain maybe like a six months advantage over everyone else and be able to use that in your compaign strategy which is pretty good yeah that's pretty of course and I still see some some bigger and smaller companies they don't even they don't even run any iOS UA because well it's hard we can't even evaluate yeah sure of course thank you very much please don't do that at least it's less competition on iOS exactly it's really painful for me to hear yeah these cases I hear yeah we just gave up on iOS you're not gonna spend there but yeah it's just living room for the others to uh you know to uh take what that's the thing like those those iOS whales and payers they just they didn't disappear they are still there of course it's kind kind of hard to to get them to to play your game but scan scan 4 could

definitely help could definitely help agree uh so you mentioned at the beginning that you might be able to see some ad set level data some creative level data any anything more more than maybe because it's still pretty unsure that if you can so if if you know more please share yeah really curious okay sounds good yeah yeah okay so uh let's try to break it down so it's around the deeper granularity and it's associated with one of the also very interesting features of scan 4 that apple called The Source identifier so the cool thing is you know in older versions uh scan had kind of to represent where the user came from you can encode like the network name so we know that us came from Facebook as an example and a pain ad which was just a number between one and 100 so basically

two-digit number um and that made things difficult for the network side because it's not a lot of room to optimize understand okay how basically you know how how the segments basically of the users and how granular they are so it led to you know many channels saying you know we can run up to eight campaigns per account for example and they can't run more than more than that in scan campaigns or that they could only show campaign level statistics maybe with some modeling they could show some you know adset level uh data so it really depends um but it basically because you only have 100 options like think even about like country level data if you're run a global campaign like everywhere for a global audience it's let's say try to split it by country in the it is it's it's pretty tough to match with only 100 values and so it's a problem for the network so

now with scan 4 kind of apple redesigned it and called it the source identifier and they basically said again based on kind of the Backward Compatible design they're saying in the source identifier you always get first of all two digits which you you would always have so it's kind of like equivalent to the campaign Ed you had before and uh you have the option to get an additional two digits depending on volume so if you run at a certain scale you get more than compaign granularity you get also another digit and and then another digit so total of four different digits it's like 10,000 different combinations of where the user came from so that gives a lot of flexibility for the ad Network to decide what they want to know about where the user came from they came from a specific adset really from a specific ad maybe from specific videos that you're running you have 10,000 different options that's a lot right that's pretty good news um why is

the maybe thing here why is it with the cavette again the good news bad news so good news I have up to 10,000 options you can map it to different creatives different an maybe different placements great bad news is it depends on your volume by default you're only going to get that two digits which is the same as you have today only if you get to enough daily installs and that's still a question mark how many uh you need then you exposed with more grity so you need to pass the Privacy threshold which is now going to be named as what crowd anonymity or whatever is is that uh yeah another kind of new terminal yeah another complex City into the the whole iOS okay so that's why maybe okay so all right now I that's the maybe so they've redesigned no more privacy specials now only Crow limity tiers and the more tiers you pass the more data that you get that's why like

my personal guess is probably only the large scale advertisers would be able to get more granularity because they're running at scale and you know if they need more than just like a campaign level obviously everyone it's more than a campaign level but they would have enough volume to get more details such Asti creative or placement or whatever exactly exactly okay okay no no no that makes that makes total sense okay yeah thanks yeah great now now it all clicks actually okay well I think we could we could spend another like 50 100 200 minutes whatever on scan 4.0 but as soon as we hit the adoption rate and I have more uh campaigns running on scan 4.0 I'm definitely W you back and we were going to discuss like what's what's happening in let's say in couple of months so y for sure thank thank you very much for coming this was really good and I kind of got what I what I wanted from the let's say actionable

point of view so I know like what should I do and I hope our listeners know as well so uh yeah thank you very much for listening Eran if our listeners want to connect contact you can you give us some uh contact details if they have questions yeah so um you know you can contact me at LinkedIn you know Ron fredman from singular feel free to uh connect with me I mean also on all the slack Channel Community slack channel is there like you know NDM and such we have also a mobile attribution privacy slack Channel happy to connect there uh yeah hit on Twitter on LinkedIn or even send me an email iring girl.net and I was happy to connect and chat for hours about scan you know happy to talk nice this was amazing thank you very much for coming and uh our listeners thank you very much for listening please share um the podcast with industry friends uh dogs and cats they love SK 4.0 uh jokes as

side know thank you very much uh see you next time cheers bye-bye [Music]