Superstar Session: Key to Success in Post-iOS 14 Advertisements: SKAdNetwork Data Reconciliation

and our next up is a superstar session about the key to success in the post ios 14 world uh we've got uh carol from tengen joining us um carol can you hear me okay yes i hear you very well dave perfect well look thank you very much for joining us i'm gonna hand over to you for your your presentation um if folks in the audience would like to ask any questions here uh about uh tengen and about the subject in hand please drop them into the q a box and what i'll do carol is i'll come back in 20 minutes or so and uh help you to curate those but in the meantime i'll fade into the background and hand over to you perfect thanks a lot so let me share my screen very quickly and we can get started uh yes hi everyone i'm carol and i'm here today to talk to you about the key to success in a post ios 14

user acquisition world sql network data reconciliation i know it's a very long title but i promise you it's going to be interesting in a different way than the snapchat session you just watch just bear with me and the data lovers among you really love it yes a little bit about myself i'm a data scientist at heart and i'm the general manager of the berlin office of tanjin some of you may have heard of tanjin as a very affordable attribution for eaper casual this is true and we also offer other kind of services like putting all your marketing data in one place your cost data your ad revenue in-app purchase subscription and we calculate for you ltv-ri ross in addition to our product uh like our dashboard api and database product we also help our customers grow to the next level by providing them training that are custom made for their their size and their level of

growth and as mentioned we are affordable but even better than that we are actually free when you are very small and don't send us a lot of events and this was made specifically to accommodate indie developers and as you grow the pricing will grow with you as part of this training that we provide we realize that ios 14 can be very confusing to a lot of you so we made a few training specifically focusing on that for example conversion value modeling training and we are now preparing ios 15 training something that i will also mention a bit later in the talk with that being said uh let's get started this talk will be composed of three different parts i will start with what i call the basics which is basically what happened in the last six months i don't know in case we're in a coma or in a very isolated place and you know nothing about ios 14 i want to make sure everybody is on the same page then i will go into the data the fact so

what is the situation today and i will finish with some practical advice if you two is trying to do data reconciliation between the old way of attributing and sql network attribution so let's go last year uh apple announced that they will completely disrupt the mobile ad industry and uh enforce a much stronger user privacy they want to do that using this app tracking transparency policy that require that every single app on ios will have to ask for permission of the user to be tracked for any purpose that this tracking is serving so this was announced last year but only enforced end of april this year and you should probably all have seen this pop-up what we call the att pop-up that effectively has this permission should say one sentence about why do you want to track the user and then the user choose either allow you to track or not so since this announcement we heard a lot about

oh it's the depth of idfa and idfa is the apple created user id that is used widely in the advertising industry but actually this policy touch a bit more than just the idfa it's any kind of identification of a user anything that can make a company pinpoint who this user is and share this indicator with another company and then the other company can also easily identify the user so that could be any id created by a third party it can be uh you gathering enough information about the device that you can really clearly know who is this person behind the device and it can be emails all of these fall and there is policy where you need to ask permission first before tracking so since the announcement uh app developer publisher all have been striving and testing a few different things about how to increase the http opt-in rate meaning how can they suggest that their user uh accepts to be tracked and

that's fair game uh however you need to be careful that you stay within apple guideline which mean you can't say something like if you don't accept to be tracked your user experience will be terrible this is a bit like blackmailing and of course apple did not accept that so with this loss of user id how to do with marketing like marketing can not work as usual so apple also announced their own attribution uh what they are called sql network and this is their way to solve this problem like for all these users that will refuse to be tracked apple will provide attribution and this will work without the company the app developer or third party to know who this exact user is already all the biggest ui channel adopted this form of attribution and our partner of apple so this form of attribution would work with all your current uh channel i also want to take a step back and explain for those of you that are not familiar with attribution in a

very simple way what it is so attribution is when you span your marketing budget on several channels for example facebook snapchat google and you need a attributer to tell each install which channel should be credited for and then it allows you to measure the performance of each of your channel and campaign you choose a mobile measurement partner mmp and integrate their sdk in your app and this way the mmp sdk will track every single event every single install at least and send it to their server and on their server they will also track let's say click an impression from your ui channel so that they can see which channel send the last click before install which is a very basic way to do attribution and reward uh this channel attribute the install to this channel so in this slide for example we say facebook window install and is informed so bear in mind this is a signification that just puts everybody on the same page this is how it worked before ios 14.5

with apple new attribution scat network it will be slightly different you can still work with mmp for example tangent and heavier sdk and the sdk will still know every single install there will also catch any conversion value which is a new indicator that apple put in place because with the new policy where you cannot see a very specific user ltv and rri cannot be calculated anymore so instead apple provide these conversion values so you can still define that a user that will reach level 2 and play 10 games have a higher conversion value than a user but only reached level 1 and played two games [Music] so the sdk the mmp will still catch this info that will be sent from your app to the apple server and then apple will take care of comparing clicks impression and decide who is the channel who in the impression let's say here it's facebook again and then from this ui channel they will share this information to your mmp

and here uh there's a little bit of a problem as this is a new process the industry is just starting to put things in place and some channels were not ready in the past few weeks some decided to share it through api rather than real-time callback some share of a real-time callback but decided that they don't want to share certain information uh like country or publisher source that i will also mention later so here there's already a lot of work for your mlp to kind of receive and standardize and make sense of this different kind of data that they're receiving for you so that's all the theory like what we know and we're prepared to happen now what has been happening since the enforcement of ios 14.5 um ios 14.5 just to reiterate is when this http was made mandatory and was launched in late april this year and when we looked at our own data last week of all our ios users 68 of them

were on one of this newer version of ios so ios 14.5 or 6 or later but that's really few of them um so that means not yet all your user can fall on this new attribution and this new way of doing things what was interesting is that it was launched end of april and the adoption rate of a u.s user was fairly slow and it's only around june when their apple pushed ios 14.6 that we see a increase in the adoption rate until it reached 68 and what we notice as well is that the advertiser on ios kind of slowed down their span as the adoption increase and that's kind of expected because people have to deal with new data set a new way to do their ua marketing strategy so a lot of the advertisers decided to slow down their span observe how things are see their operation can keep working the way it is and some of them even decided to shift their budget to android during this time another interesting

number was of all these users who are on this new newer os of ios only 24 actually accept it to be tracked so that's a minority of them and think about it as only two users out of 10 can still be attributed the old way and everybody else would not have idea anymore and can only be attributed through sql network so knowing this information in the background in the back of your head how would you go about looking at your old attribution data and the sql network attribution data which as of today happened both in parallel so i decided to take a very theoretical situation let's say you are used to have a hundred new install every day and you're working with nine ua channels and plus organic of course then and you have ten new install on all of these channels so that's a very simplified picture of what

a real situation could be as of today if you look at your all attribution data and considering that probably you have probabilistic attribution enabled which means we can keep doing the attribution using click and impression the same way we were doing it for limited tracking user in the past um you will see that okay probably your your attribution numbers are still pretty good for most of your channel but some channel you won't have any install because the self attributed network like the big one like facebook apple google snapchat do not support click and impression i have never supported this way so there cannot we cannot use probiotic attribution on this one and their install will go into organic uh that's not the only issue the only issue is that also down the line there's a lot of talk that probabilistic attribution may stop and may also be cracked down upon from apple there's

a bit of debate on is this going to be tolerated or accepted for a long time and if in a world where this was not possible as of today your data would look like this where only the user that are accepted to be tracked to simplify let's say two out of your 10 user per channel will be attributed properly and everything else will go to your organic so that's not great and it looks at how it would be for your eska network attribution first of all you should know and that's the first challenge but as can network do not in form of organic install it's only for your paid install and as of today as i mentioned earlier only 68 of the users are on this new os and can be tracked through excalibur so for each of your channel you will only have this shelf user that can be attributed of course down the line in a few weeks or a few months when all your ios user will be on a newer version of ios then you will go back and in theory be

able to get all of your install for each of your channel using sql network so let's say you accept the situation that not everybody gets is on ios 14.5 or above and want to do a reconciliation on the channel level like just channel a so you look at your you expect 10 install knowing that what you usually get but you saw that you only received three sql network installed so the first reason why this can happen is there's a time difference like in the old way of attributing we know exactly when the install happened like the day and the time then the minute and the second while with a attribution from apple uh there purposely added a timer that can be up to 24 to 48 hours after the user finished all their conversion value path let's say in your app so but you cannot pinpoint that escape back you receive from apple come from this one user you got in

france on monday so they wanted to prevent that still to protect user privacy um so for all your install on monday you need to look at esca network from tuesday wednesday to kind of find them without being able to know exactly which one will be the monday's install so that's a bummer but let's say okay you move past it past this you accept that that's a difficult problem to solve and you want to ignore it at first um you still think yeah receiving only free sql network installed for my monday user but still seems like lower considering this time discrepancy and you might be right uh depending on the channel you're running your campaign on some of them don't have sql network attributed attribution enabled for all of their campaign there's only a few campaign type where they enable it by default um and that's different channel to channel so talk to your ua channel you can double check that with them but it's possible that you're running a few campaign but don't have sk network

attribution enabled and therefore you're not receiving the sk a network install for them then let's say you want to drill down lower than the channel level and look at i don't know publisher source or country uh you might see even less sk a network installed that's what you expect particularly when you just start your campaign apple also have what they're called the privacy threshold where if you only got one install from a certain source or country uh then they will not send you a scan network pushback yet they will wait that there's a certain number of installs so that when they send them to you you cannot easily pinpoint who exactly was the user that this postback correspond to also a certain channel as i mentioned earlier um sends the sk a network install from apple in a different way so don't include countries um also don't want to include publishers so sometimes depending on the

channel you might see that these granularities are missing so that's a lot of challenge to overcome uh but here is what would be my advice and uh some guideline of how you would still go about doing this reconciliation first how to solve the organic problem well that could be quite simple but you should always keep a tool or a place where you can monitor your overall new install even if you don't know where the install is coming from and know exactly every day how many new installs you get then when you're confident that you are receiving all the sk network install for your campaigns you could always do this simple formula total install versus sql network install and that should be your organic second advice uh when you compare day day by day let's say your old attribution data to sql network data you there might be a bigger discrepancy even if you take a week of data or a month of data because

in a week of data that might be only the last two day of the weeks but you wouldn't have a complete you you wouldn't have received all the sk network uh installed so the longer the time frame the more you have days where you did receive the complete data set from sql network advice number three uh doctor you talk to your ua channel and uh ask them like uh how many campaigns do you have that don't have sql network enabled of course the data scientist in me would tell you just enable all of them stop the campaign that don't have scheduled work but i can hear some of the ua manager here uh saying well but this kind of campaign maybe don't perform as well or the cpi is much higher i don't want to do that and that's also fair however i would advise you to plan ahead and that at some point you pull the plug on this kind of campaign and just accept the new way of attribution and try to have all your campaign with sql network attribution enabled but when can be a good way to do that

that's my advice number four keep monitoring your ios user base and which os version they're using like for us it was 68 for you it might be more it might be less but when you see that like almost 100 or percent of your ios user now use a newer version of ios 14 maybe now then it could be a good time for you to be more bullish and only adopt a sql network campaign and really get more familiar with sql network data bonus advice uh wait for ios 15. so a few weeks ago apple made a pretty exciting announcement but in ios 15 things will be done a little bit differently so remember how i explained attribution earlier on the apple sql network well they added this pathway where your mmp or yourself as an advertiser don't need to rely on the channel to bring back the sk installed to you apple can also send that directly to you so that solves the problem of um for example not having certain of the

the san installed or having some data being hidden like publisher source or country now all the data you receive will be the same there's other problems that will arise but i will be talking about this in the next slide because of course i would be damn if i don't advise you to use tangent one other problem that will remain with ios 15 is that the sk a network plus back only include the campaign id of sql network what they call sql network campaign id which is not something you can easily map to your ua dashboard campaign like you cannot easily know which uploading campaign or facebook campaign it's matching too this is still a black box but with tangent we will solve this solution and map this for you we are also probably building a solution called attribution modeling very different from probabilistic attribution using scat network information we will bring back uh information like ri and

ltv on the user level so that you can have all the granularity you're used to as of today this solution will be delivered in three phase we already finished phase one we are now in phase two so that if you are a engine user very soon you'll see in your dashboard this new calculated roi and ltv using sql network and the last phase would be for our most advanced user to also have the data behind the hood of for each single user what probability did we calculate it but they came from this campaign of this other company so a quick recap you can take a screenshot of this this slide if you are interested today very hard to do the data reconciliation you're trying to compare bananas with oranges let's say but i hope with this advice you will still try it and stay on top of your data set so that you are prepared when the data become cleaner and cleaner to monitor your operation using sk network data you know how to calculate organic take a longer time frame but what you're used to if you usually compare a hot day

today just like weekly cohort talk to your ua channel make sure at some point all of your campaign have sql network enabled and monitor your ios user base be patient but be prepared also if you have any question if you have been trying to do this data reconciliation yourself we want to hear from you uh this is the email where you can ask any question about tanjin and for me you can just replace support by carroll tangent.com and i would be happy to hear from you thank you fantastic carol thank you very much round of applause on behalf of uh all the audience there i know that uh at some point soon we'll be able to do this all in person but until then you'll have to imagine the applause coming from from everyone really interesting uh presentation there about a hot topic that we know we've you know we've covered many times on the pocketgamer.biz website and the the people are it's been one of the hot topics of this year right the um the the change in the way that the ads work on apple devices so that was really great and really granular information

there you know you you explained a whole bunch of stuff there that that was uh that kind of demystifies some of those processes for for uh for me and i'm sure for others so we've got a few questions there from from people i'm gonna just take a few minutes we've got we've got some uh minutes before the panel join us so i'm gonna grab these questions thank you very much to our attendees who have dropped these in here and i'll put them to you if that's all right carol um so uh the first person says here on behalf of indie developers um uh they ask how many adjustments and what impact will it have on monetization and what does 10 the tengen solution absorb part of this impact so the this question also apologizes if they they miss this i know you went into a lot of detail but i basically i i guess they're asking um as an indie developer what impact uh what what impact does tengen help solve basically yes that's a very good question and a

very uh general concern like the monetization side of the impact of ios 14. for us in tanjin we are not the one helping you get ads or pushing your ads so what we bring is visibility like you can monitor what's going on and if effectively your ad monetization is dropping and you can drill down into lower level so you can maybe pinpoint is it only dropping on the country level or for certain ad placements this is where we have we would help and particularly also for e per casual what is important is does your monetization is still generating enough revenue to cover your cpi like this is the whole business model of hypercultural using ads so this is something where you want to be monitoring and be confident that you have the right information to make sure you still have enough roi got it thank you and this next question really i think the person is asking for your opinion on this um apple of course made these changes to ios uh the the removal of the idfa and so on and

um it comes as uh part of apple's drive for privacy i guess so the question has asked here though is the ios starts really about privacy or do you think it's more about apple having control over the data okay so this would be very much my personal opinion um i think there's a bit of buff uh maybe 99 about user privacy i mean this is also a good selling point for apple to sell more more iphone and computer either can promise that your data is more secure and we know that the apple consumer love that that's the big selling point of their marketing uh and the other one i heard a lot of like complete theory that oh yes they're gonna gather all this data and have all this user data and use it for themselves who knows i'm not really sure i mean these companies are under scrutiny in the us that's not entirely in their interest to do that okay that's interesting that's good thank you for your opinion on that um again another another good question

here forgive me if it's detailed so forgive me if i as i read it out so as you mentioned there are a lot of challenges and trade-offs with the time-based conversion tracking model in the s-cad network and determining what to actually measure is not so easy anymore what will be the long-term impact of that for smaller developers is it or is it just a case of having an adjustment phase uh definitely there's some adjustment phase that people reduce what they're tracking like in the past they'll have a lot of custom events and track after the install maybe forever like what the user is doing right now with apple that they are limited to track only what happened after the first 24 hours just because they want to have information of which channel won the install fast uh so this already we saw most of the casual studio do reduce and only track in the first 24 hours after the install and then i think for indies they will learn and they will be educated more and more into what kind of

products uh analytics events are important and can help predict the valuable user and i think what will happen for them is that more tools because we know that in this necessary have a resource to hire data scientists team and gather a lot of data themselves so i think a lot of tools will go into the space to fill this gap and try to sell to indie developer knowing that all these bigger studios started as indies and bits right no that's great thank you actually just on the point that you made a moment ago i'm going to skip to this question here someone's asked what changes in kpis would you advise developers to do to get the most out of the s-cad network uh kvr well definitely conversion value that was a kpi that didn't exist at all before that everybody had to get used to so if you are getting used to and you set them up properly you might be on top of the game in terms of being able to optimize your campaign and that can help you a few slot in the app store and then later on learn to translate this conversion value into

predicted lte great thank you someone here's asked will sparse signals overly favor advertisers with the most budgets and is there a workaround to that um with the most budget um i guess these people will always have a bit of advantage but still if you if you just bid in the dark at some point you you lower your margin so i don't think on the long term these people will be advantaged really the people who would be the most advantage are the ones that understand the sql network data and can use it for their operation and optimize their company properly okay okay um oh here's a question how should we act to protect ourselves from fake data on the new ios system uh if we know that post-install conversion data can be faked um yes so i'm not the

best person to answer this question that's not my expertise but apple will have a system where you can't send these escapes back you receive for them or you think you receive from them but you suspect that it might be fake for them to verify it and of course your mlp can also do that for you okay and just while we wait for the uh the next panel to come along i've got a few more questions if you're happy to answer them i'll i'll bring them to you so how are local changes in legislation for instance the copper in in the us uh the age-appropriate design code in the uk which is similar um how do they impact the way that developers and publishers can gather and use data yeah uh that was maybe the first wave with gdpr of a developer becoming more and more conscious of what their collector's data um i think like apple policy now is broad enough that there's a big overlap with most of these local businesses you just need to stay on top and see if any of this local pro policy suddenly is more aggressive and

you need to to hide even more but really i don't see what more you would need to hide than what apple already recommends right uh let's go let's let's just take one more here while we where we go um how would you advise us to deal with the issues in the s-cad network specifically the distributed postback data and limited granularity uh yes that's many things basically try to collect as much data as you have and start today to do this exercise of trying to reconcile it and of course as i mentioned today it's going to be really hard because there's so many differences in bob dataset but by just monitoring it you should over time see that there's an improvement and the data quality get better uh and do that at different granularities like don't start at a very low granularity because you'll get desperate and want to give up immediately start on the channel level uh like i recommended and just also keep an eye on the country and

publish resource level but at some point in a few weeks you should see the channel level data already looking really good and then have hope that more valuable data will do better and with ios 15 even more all of this granularity should be covered that's perfect thank you well we'll leave it there thank you to everyone for all their questions that's great and uh carol thank you for taking the time to answer them uh brilliant there if people i want to follow up i guess they can uh again you put some details on the screen there but they can find you on the discord where they meet to match and and so on so carol why hi thank you very much uh round of applause from all the audience that was great really really insightful and deep dive into their into uh into that hot topic so thank you very much thank you thank you gamer you