Targeting isn't being used enough in gaming | Ofir Pasternak from Persona.ly

foreign [Music] from personally how are you doing today Ray how are you good we're at the end of the show for the first day anyway halfway through so uh surprisingly energetic considering you guys have been dead downstairs in the floor what's it been like down there crazy a lot of audience a lot of people that want to hear more about personally and we are really happy to be here excellent that's what we we very much like to hear now if people haven't encountered personally tell them what the company is all about uh so we are DSP uh it means that we are exposed to a lot of auctions at auctions that we have the ability to bid on um we do it with a lot of machine learning involved in the process from predicting uh who is going to click who is going to install and who eventually will make a purchase and make sure it will be a big one so we have the ability to Target the actual what we call whales that will actually span and will

generate campaign with high Ross and that's basically the goal for Sally we do it with a lot of data our models look at more like around 60 dimensions of data including age and gender predictions that we we have we build everything in-house the whole tax tag is built in-house extremely scalable we listen to around 3 million uh auctions per second so extremely scalable and working with the big guys here and we're here to find more clients excellent now I'm asking everybody this were there with changes to the gdpr rule regulations a few years ago and then Apple's ATT thing recently as well you talk about data and the amount of data and the types you can collect have been changed in some areas how have you adapted to that how has it impacted the way that you work yeah so since ADT style is basically iOS 14.5 uh so yeah the the signal is not as it used to be uh so with less signals you go with the things that they do have and try to

capitalize on those um and those are like attributes like device model which you still have the context itself so we utilize a lot of context features we have models to understand the context and we need the app description and then we understand how close it is to the app that we are actually promoting and that's how we can use context in our models and compensate on the loss of signal basically right so it's making the most of the data that you do have yeah so we capitalize on what we do have of course it's also changed the prices in the market so prices are actually lower these days because like we lost signals all our all the other dsps and players in the market lost also the signals so they bid lower right so yeah it's kind of a lot of opportunities there as well where are you finding the most valuable UA is coming from these days with everything changing as quickly as it does in the market you can't just rely on one model for forever because constantly having to adapt so at the minute what you find is working for your clients

um so yeah especially so what they do these days is a mix of you was like definition and retargeting which is like that's the low hanging fruit right so you know people are paid in your app or or used it so to return them is much more uh much like it's less expensive than acquire a new users so that's something that we really recommend on clients to start uh doing um and we also do it with in-house with platform that we have that can segment the audience based on characteristics like when was the last time to use the app if they paid how much they paid and so on uh so we're targeting is essential these days and it's really not adopted enough in gaming so we're targeting is practice that came from e-commerce but it can be easily be used um in gaming and sadly it's not being used enough so really working on educate educated clients basically and try to make them strongly targeting because it's really easy to do uh and essentially those campaigns Drive the highest laws and very early the campaign life cycle I was just speaking to

someone who was talking about they worked with subscription and live style apps and moving that towards gaming in a similar way that it's a different kind of service but a lot of the same techniques will work and apply to gaming right in terms of UA and the rest of it so that's interesting to hear now when it comes to well I don't like to say mistakes but I'm going to say mistakes when the clients are approaching you and saying hey look we've spent a whole load on UA and it's not it's not worked for us we've dropped 50 Grand here or there what are we doing wrong what is the most common thing that you see clients maybe doing not quite right and they come to you and you can say ah we recommend you do this instead yeah so there are like few things that we see that really repeat themselves so basically also like if it's not if you don't see a trend when you start a campaign after I don't know 10 or 20K you have to ask the tough question and understand where is it going um and not just let it run because you trust someone or because yeah so we need to ask questions all the time we understand what's going on behind the

hoods it's really important to understand because a lot of companies these days they're saying okay we have machine learning it's a black box we can't really and explain it but in reality it's explainable right so it's really important to ask the question before you get to the 50k um and and yeah yeah and another thing is due diligence like there are the companies you work with understand if it's really DSP or if someone claimed to be DSP check how many employees they have on LinkedIn if they have like data science team and so on so your mobile due diligence before you start a campaign and understand uh what's going on and another thing that's really important for a success working with the DSP is like sharing the data and beta spend as possible with your kpis and be clear with your kpis like being here at least stick with your KVA because sometimes we get um we had a campaign two two years ago on Facebook on a very small scale and this is your kpi but this is not realistic so we need to get some average kpi that we can actually hit and from

that point it prove it so you are into sort of sharing this stuff with the users yeah avoid the Black Box thing is there a danger any point that you kind of overwhelm the the end user with data they don't really know how to interpret how do you share the relevant information and then give them the tools with wish to make the informed decision so they're like very common standards on how to do it so it's it's not uh it's not housing though so yeah I could do it so what about 2023 you guys are obviously looking forward are there any initiatives you're rolling out or things you're looking to do are you expanding what's the plan so Google is about to launch their uh privacy sandbox which is similar to ADT so we're trying to find a way to how to manage that and work with context on Android as well uh we also explained to other uh exciting territories uh in Saudi Asia that's really uh this is really a strategic move for the company ah man yeah that's it looking to to continue to grow foreign

[Music]