Superstar Session: How We're Preparing our Customers for iOS 14 Attribution Switch

next up though we have our superstar session from our kind sponsors tenjin so uh here's our speaker next so uh mathaus is going to be talking all about how they're preparing their customers for ios 14 attribution switch which if i'm not mistaken ios 14 is coming today i think the it is but i think like like the specific um feature will only be launched later right well yeah so i suppose you're gonna fill everyone in on there how they can work with that and everything yeah how are you today though before we get started great morning great coffee uh it's a sunny day here in berlin

oh fantastic so yeah so what is it about half 10 there yeah and 10 30 yeah yeah so nice time nice time so yeah fantastic yeah we we're all looking forward to this uh obviously you know our superstar session uh for the tracker track sponsors so thank you very much for that you know without you guys we can't do this so um contribute to the community and it's so lovely for you guys to bring the community together in this time so we really appreciate that oh thank you yeah i mean it is nice to be able to continue doing these things and yeah just keep the engagement keep the creativity and yeah keep people together at what what is proving to be a difficult time but yeah i'll before i leave you going i just want to say like i did with alina's talk to everyone in the audience i i think you know this because you're very engaged but uh drop your questions into the q a and and i'll be coming to them at the end and yeah we've got the chat function as well but for now i will leave you to it like last time i'll wait until you've got the screen share going just in case there's any issues there it is perfect and uh yeah i will

be back for the q a at the end looking forward to this hello everyone i hope everyone sees my screen um excited to do this just to do this and we have uh we'll be talking about how we are preparing our clients uh for the attribution switch and i think the way how i'm going to do it is um i'll do a quick elevator pitch why you should be uh listening to it and and why this is available and and i'll be talking about how we have club helped our clients so far and then i think this will be the highlight i'll do a training which we've done with some of our more advanced clients on conversion value modeling i think this will be the highlight but i also be talking about i think some of the upcoming trainings this week to apply in future so first of all let me try to win you over and and and make you understand why it's maybe available to listen in i think you know like when when the news about scott networks broke

especially in hyper casual i think it people were shocked and and then they went from this curve of denial anger bargaining and uh and depression read like a lot of pundits said like hyper casual won't work after i was 14 and i think what's happened since especially for people who are a little bit more pragmatical and practical is they went to a little bit more to problem-solving and this is where i will try to take you there right um matthew saunders from from network uh in his kickoff panel like a session on um earlier in this conference you know he went to his take was like look at the history this happened to the industry a couple times i think that's a great take and matthew is totally right but i think what i'm trying to do is to make it a little bit more practical and give you a very very specific task in this very hopefully practical session where we will create like a conversion value table and then

i will create also some conversion metrics awesome so let's do this the value of this of course will be once i think you've done problem solving and i think merit from rubygames he if you go to linkedin he published this um like like very version earlier just this like i think two weeks ago um once you've done that i think you go from pretty much like guessing to actually seeing things and we were at the back a little bit at the training and um and i'll be talking a little bit not too much about what what use some of you guys saw probably on that linkedin post um but we will do it pretty much like a version of this which is much much simpler awesome so let's do into let's talk quickly

about tangent and what we're doing with like like a quick one minute elevator pitch like what we're offering to clients is tools training at a very very good pricing and we have a whole set of tools um attribution has always been only one piece of that and we offer i think people a lot of training and special type of training because we have a unique product which is our data warehouse which allows us to help our clients on code level um we're particularly good with hyper casuals and index this is some of our clients some of them known in the industries and this is how for example like how our clients receive training and what you'll see here is like that things like like the to-do list which i'll show are available on the free tier but like some of the other trainings like if you really want to talk and engage with us uh like everyone i will be presenting later are available on

higher tiers so let's do this uh i was 14 to do listen so this is something we sent i think a month before facebook announced um their take on this and we gave clients like like please look a couple of notes please do this please think about that um i think that the key point out of this is i think the point around apple's ab tracking transparency pop-up um there were some questions from our clients whether it's something which should be part of our sdk or whether it should be something built and the case for why they should be doing is is that we see that clients want to have their own logic on how this is going right so like uh say games wants it very differently than some of our clients and i think something people wanted to control and it was unclear how old people would go it um when the biggest i think questions like like like i think value

which we drove was we made our clients talk to ad networks this point number two right and we didn't not only tell you like just talk them but we gave them like very very specific tasks and here's like an example of such tasks and we tell them like go to facebook we think this is how your metrics will be changing let's drill down a little bit deeper what this means so you see for example the column with scab network installs and our argument was like we believe given what we understood at the time is that creative and level and counter level data will not be available anymore and we wanted our clients under to think through what it means to the workflows and talk to the ad network pretty much every main channel whether that's true or not the key part what you see is which means for which is important for our clients is the column with ldvs

as you see a lot of ltv data would not be available anymore and the wording not available means not deterministically available it's not like a kpi which everyone will agree on it will maybe be a kpi which people will be calculating with different specific methods um if you want to download this to-do list it's um this is a big qr code um um it's it's available for free um um i think it's like our marketing shout out to roman and ebony uh who did this did a very good job in sharing this with our clients um so let's drill down deeper to maybe my my my favorite part of this talk which will be like an actual demo um so when we off of training so when we train clients on this so

first thing we make do is we make them watch this video from kevin bravo from alexia on conversion values manage two pathways so let's assume um like somebody um has been uh somebody has clicked on and like user has clicked on an ad then we see an install and then we 24 hours happen and then between 0 and 24 hours next hour like apple will tell you which ad network this this install came from so anywhere between 24 hours to 48 hours we'll see where where where the install had come from in the second piece in the second example there's an install then there is 24 hours and then something happens let's say a user goes to level in and this is like an um valuable event which is important to you and you set a conversion value for it if that

conversion value is being hit then the timer starts again for 24 hours and so forth and so forth for example if if if you if um if the user is hit level 63 after seven days only then between 24 hours and 40 24 and 40 hours you will know where where the installs come from so what i'm saying is here you have to take decisions on how early you want information or versus let's say how um how um how deep that information needs to be so once you know know these two concepts like like what is the timer and to some degree uh conversion value let's go turn deeper and see like a conversion value and how you could do some table and some modeling um there is pretty much 64 conversion variable values available for you technically 0 to 63

and if you think about it in a spreadsheet form it's just numbers one two three four five six seven and so forth and all they all have a binary representation and so you can again and if you think in spreadsheets you can just shut them down this way and what you would want to do is you want to map that to player actions or meaning for events to your business model and in our view in hypercasuals these are ad formats watch banners interstitial rewarded videos if you watch uh the example from kevin bravo and alex here where he talks about more about levels and erp so these will be very different for the type of game you you have um we then make our clients do some analysis on their immediate distribution within 24 hours on on some of these meaningful events uh

this is an example from attention more popular d training where you see within days how many you how many banners interstitial android videos usually see so on average you see in this case like around 18 banners two and a half instructional and maybe one or two rewarded videos and that's a great thing to know because like this that's the next step you should try to think about what could be good outliers for these meaningful events which you would then attribute the conversion value to in this case of the example which you did we went from one banner to 20 banners one interstitial eight interstitial and one video to eight row or doing videos um and you go then this and essentially say like 1 2 10 20 banners and you attribute to these meaningful events these conversion values um this type of

view has become quite popular i think it also gives you a good understanding of what a controversial value is now that you have that the question is how can you generate metrics out of it um what i'm going to show is is a very very simple model um to show you some basic mechanics and the idea is that you get like your eureka around how that conversion value becomes a metric so in this example we go straight for a simple d0 ltb prediction where a banner is um 0.1 cents uh and strangers 20 cents or a one rewarded video is 50 cents and there's no separation around country there's no time effects and if you if this is your assumption then you could essentially uh take that in in information and do a simple multiplication

and say 10 banners times zero one cents is one cents and then you have like a d zero ltv value or in the case of like a second conversion value 10 which we said we will know that for example after some pretty much between 48 hours to 70 to 20 hours that for 10 banners one rewarded video we would have gotten a d0 ltv of 51 cents yeah i think this um this is like really really i think awesome to know because it will it shows you actually that for most hyper casuals where they optimize for like ltv day free ltv or this is not too far away from what they're doing anyway and and i believe what you'll see in this demo is that the hyper cache model will be alive and kicking uh and i think this is

very very different to i think what we're would if you follow the news what the constructors find people meet say or some of the other industry bloggers we deeply believe from what the type of modeling our clients run is that hyper casual will be alive and well kicking this example from ruby games of course is a little bit more more complicated and we think what you what i showed you is a good starting point and our clients you should go from there and iterate um on it a little bit more what you see here is that there is like a country element like for us and um and also like the values of of the different ads have been calculated very differently and we and this is certainly a much better model than the one i showed you but as i said the one i showed you this is just like a starting point awesome um so let's have an outlook here right like i was 14

scat network changes have been postponed so the question is like why should you be doing this now we think if you do this it really will make you understand how your workflows will change how your business will be impacted and we strongly suggest to do this kind of exercises already now ahead of his skype network changes um and the second step on it is like as you think through the uh changes is to think about uh ltv prediction in in the entire huge age which is coming uh we think like agalift and and and paul boren where especially um has made good comments like you can find his post on on ios 14 ltv prediction and and to get like a little bit more good framing for this but we think you definitely think that without deterministic attribution um app advertisers will need to leverage like probabilistic methods how to get

there and um and i think this will be a very very exciting um time uh which is coming we see some we have some clients who speculate that and we see that this is some of the questions they ask and i think this will be questions which a lot of people in the industry will be asking but one question is how long do you want to collect data um for what we see is that in hyper casual i think people will be earning more let's just do 24 hours because like i want the timer to start as quickly as possible so i get the data as quickly as possible um so that's uh that's our sense of it um and if this would mean for is14 for ltv prediction that this data will be the starting point for the prediction the second point question maybe is what revenue data granularity do you want to choose and we have early again indications that

more granular lrd data like more popularity will become more important um then it gets even more interesting right as people will be building for more models they will start using external data sources and so there will be daily uh headings or weekly endings we've seen some very fascinating months happening here already and as you start automated which impact will that have for your data how are harding warehousing how will your etls look when you fetch the data and how will you calculate for metrics i think this is really exciting and yeah like one last show that for what we're doing we of course think about that and and in our data warehouse product datavolt and we will be definitely offering uh scat network data in it and training like the one you've seen here before awesome so this is pretty much it from me yo uh just me on twitter if you want to

talk to me uh not even on discord but also later i hope you guys enjoyed this session fantastic thank you so much yeah for such a fascinating and interesting talk and especially one that's so uh topical right now uh yeah we've got a few questions so we'll see how many we can get through uh before the time is up so uh but as you said you'll be on uh in the discord and uh in the meet to match and everything after and obviously you just posted up some ways that people can contact you but uh first up um sy has asked uh can we do targeting with display ads in mobile web in gif gif or html formats in ios 14. are they too dependent on idfa um the first question is like i'm not as i don't know as much about the mobile web yeah uh but i think like my my intuitive answer is none

scott network changes don't affect like the type of ads you run it and and it affects your knowledge whether it is the user who who installed your game has come from facebook or let's say google um so the change is not on the facebook's or google's ability to run any ads it's much more around whether you know where this came from and how valuable that install was yeah okay great um next up uh someone has asked um why is it so important to learn more about ios 14 will things really change that much um i've been i'm i've i've you know like i've been in mobile since games were still happening on bob and i think i've seen like five big changes uh um and i think this is another very very big one i think though many businesses gaming

businesses will not be viable anymore um i think for everyone who has let's say a casino game or a long-term subscription game um things will get much much harder because like you will not get very good data on what happens to users after some time and you will be have much harder time to calculate metrics like this so um this is a biggie yeah um well i mean i suppose this leads into this question or rather it might be the same answer for this question but uh someone asked when it comes to the new ad rules for ios is it something developers should worry about the most or is it more on the side of the ad providers who should be worried um i think everyone it really affects everyone's workflows and we we found like that the most like we're all nerds right so we found giving these spreadsheets trainings uh the most available way how we can

educate clients instead of reading posts or or anything on social media because then they will they can they are able to see it for themselves what it means for their business and we and we don't want you to react with fear we you know we're all nerds you've built your games do a quick demo and then you will start be able to start get answers what this means for it for for you yeah um so sy is back again and anna's asked um how is it possible to track number of ads watched does tenjin do that in the back end um we've been doing this with uh so far and will continue to do this and on android i think with uh we with with the arrival of skype networks will start doing this slightly differently right so what you've seen is in my demo is

that if you as a user set a certain conversion value then you will be and you care about like ads revenue then you will be able hopefully to find a way how to get this information and then calculate metrics for this for you fantastic uh just a few more and while we've still got time as someone's put how is ir ilrd possible in ios 14 um we um that's a very good question um we we've seen i think everyone's talking to each other and and in some ways they truly don't know uh um what we what we're seeing is is that uh that there will be still very granular um data available but like maybe identify a lot of the identifiers for some of the data will be wrong um

i think it's a very very fluid um way however people exchange if you like even like when facebook announced something like every couple of days i think we go post what's doing going being a change i think background in the conversations with people and i think this will continue to change um right like that's why i use the term when i claim my thing my thesis with granular data that early indications are we right my ask is watchful space um don't don't assume that what i said is will be true yeah and uh one more we've got uh when is the best time to start adopting an skad network or scab network that's a great question because in a way you know we all talk about like uh you know like like a 10 year olds about sex we have never seen it but we all excited about it and in a way viscad network data are not flowing yet right like some people have seen it in

in some test devices and and and and come and speculate about some things in some ways i think like you know you should implement it when when these changes go live right like um um and i think there will be a lot of terminal well at the time that's hence our suggestions like don't launch games when that happens yeah i think would would we strongly suggest is that you do demos like this and think through what these changes will mean for your workflows on your teams what what will it mean for your ua managers what does it mean for your company um [Music] we want you to do demos as quickly as possible yeah and uh the last thing it's not really a question but someone has um put um someone um sort of queries why tengen has no integration with traffic card so i thought if we adapt that into a question is that something that

you're looking at or is there inherent problems for why it's not possible or is it just not something that's demanded by the audience um i i think we support so many third-party integrations um um i think if we want to add partner it usually takes only two three weeks yeah so it's it's it's like like our if our clients demand for it when it usually becomes available yeah um oh and actually we've got two minutes and someone's thrown one in but this will be the last one so um i'm just warning the audience uh can um tenjin track retention at campaign uh creative level uh we can track it at uh campaign level um i think the big news will be dead with ios 14 it won't be no one will be able to track it on on on creative level anymore yeah yeah this will be one one huge way how like your creative testing uh

how you should you decide which creators work the work rules will be very heavily diverse disrupted right yeah well fantastic thank you so much and like i say i know some people this is people are still trying to ask questions um and you'll be you know there's there's you know you shared ways to get in touch and you'll obviously be in the the meet to match and in the discord and everything else like that so you know there's loads of ways to continue the conversation with you guys um so yeah thank you very much thank you thank you bye you