How to get the most value for each ad impression | Geeshan Willink from Nefta

[Music] we are now joined by gishan willink from nefta how are you doing today yeah doing really great thanks uh really enjoying the show so far and great to be here back in London it's been three and a half years since I've been in London and uh it's good to be back I understand you did a talk earlier on in the show how did that go that's right I did talk yesterday um monetizing first parting dates in advertising um it was really good really good audience at a good time of the day um people were taking pictures it was uh it was very engaged reallying that's what we like to hear well if people haven't encountered nefta before give them just a rundown of what it is you folks do yeah maybe I can start with how many problems game developers are facing nowadays um which is why nefta came to be because we want to make it easier for game developers in such a hard environment you got so many

barriers nowadays uh facing game developers of course all of the Privacy changes China regulation hyper competitive environment the funding situation so yeah it's just so many issues uh facing game developers and a big component of that is the Privacy changes from Apple um and so what we are doing is we are trying to bring back the 18 to 50% reduction in iOS Revenue from ATT opt outs um and the way we do this is through leveraging first party data we created a standardization in tracking how users interact with game but in a privacy compliant manner so no pii no IP address um no linking data no Universal identifiers models are built specifically for Publishers so everything in order to be future proofing our our privacy uh being uh

compliant from a privacy perspective um and so basically leveraging that first party data to improve ad revenue for iOS off outs that's um that's an overview it's incredibly tricky these days because as you say you mentioned the China issue and what's happened with their you know limiting daily rewards and that sort of stuff just recently but we've had four years of people really clamping down on what data can be collected and taken so how do you get a stronger result out of fewer sigar how do you still get what you need from this reduced Mar you talked about first party go further it's a great question um and of course the industry has been um leveraging all of this private information about individuals and why wouldn't you when you've when you can make so much more money using that data um what we are doing is collecting a lot of data about how users play games and this is progression uh what do they receive what

do they get um and the idea is to collect so much data to help the machine learning algorithms differentiate the value of the user you know do they complete lots of levels are they repeat Spenders are they focused on inventory um allowing the machine learning models to differentiate that value and correlate that value with bidding price so for advertising uh the bids represent the value of the user that sees it but only based on the in-game behavior and nothing related to their identity so kind of anonymous but there's still information in there yes that is obus to the machine learning well it's been a tough couple of years full stop we've seen a lot of people with within the games industry anyway being laid off and struggling in difficult times and VC spend this down so fewer games are getting made to begin with are you seeing a reflection of this in the the people that you work with are you getting different kinds of requests or being asked for different kinds of help

met so I mean I'm going to speak a little bit about the ad industry that's where we are well focused um because I think a lot of that impact has come from the Privacy changes which means Revenue reduction which means uh reduced UA it's harder to acquire quality and quantity of players um so and from an ad Network perspective um there's a lot to disrupt um in the upcoming year which can really benefit uh game developers a lot of game developers are talking about how how they have no transparency into ad networks um am I really earning the most from the ad Network partner that I have um and so we are opening up the attri uh mediation inside apps um to enable app developers to to stack up ad networks

directly in app which gives them transparency into um which ad network is earning them what for this segment of users as well as ensuring you get the most value for that ad impression and you're not relying on an ad Network to prioritize their own private deals in the background where they earn 50% margin um so these are ways which we are helping game developers now with the iOS opt outs but now going forward with with mediation to um bring back that lost revenue and transparency into the ad Network side so I kind of veered off your question a little bit but um I wanted to highlight the fact that there are so many problems and we are solving revenue from ads being a key problem there sure it can be tricky as well because a lot of people see the the way the ml works for ad networks as a bit of a black box it is quite opaque and it's

tribut to you just have to trust the information that comes out the end but I I think a lot of people will often ask for more data but at what point do you think it's like dumping a whole bunch of metrics on on the user and saying this is all this information is itself not necessarily very useful unless you know how to interpret it and what to do with it so is it a case of like less information but more useful information rather than just like a a throwing a ton of metric to people yeah and this is it's an interesting point this is where um iteration comes with what we call Dynamic models and we have a system in place where we train models on a daily basis you can pick and choose which features go into models and you can then evaluate um which features have the highest importance when predicting the click-through rate and so through iterating in this process you select the best features from all of

the in-game behavior to produce the best models and this is in a an iterative process that data scientists and machine learning experts they uh they do that's their job um a lot of the talk was surrounding AI last year and we talked to a lot of companies like yourselves who've been dealing with ML for a long time and there was never that kind of public awareness I think of AI as there is now people are familiar with generative because that's what everyone's been playing with but they're suddenly realizing that ml is also Ai and they're going oh are you one of the bad AI people what of one of the good ones do you do you feel that the changes and the the bringing to light of this kind of uh this tool creates problems or has it been useful and is there anything from the new developments that you feel would be useful for what you do um I think firstly if you're a machine learning engineer you're in the right industry because AI is not going

away anytime soon uh I think yes generative AI has taken the spotlight um to all consumers with chap gbt um we and I have been doing machine learning data pipelines big data and so on for for many years now um and that's that's just getting better and better I think um with more attention comes more money being pushed into the space so I'm excited to see the evolution of how smart these algorithms can get as a tool for leveraging performance in our industry you know um we can earn more money for game developers by improving our algorithms um continuously and that's the real end benefit to uh game developers for our um R case right if you know what I mean I do and you've come to the show today and say you did

the talk earlier on as well for from your perspective what's the value of coming to like inperson events like this what what do you do are you checking out the competition are you making new contacts so we've spent the last year um building this ad Network ad networks are tremendously difficult to build there are so many moving parts and we're now at that launch phase where we've acquired some really good early adopters and now we're scaling that attaction excuse me uh Beyond um and so for us it's about getting the name out there meeting old friends from the industry talking to to new partners um just getting it out there because it takes a lot of time to build that trust that uh game developers have uh to rely on us um so we're also going to quite a few events uh this year it's going to be a real busy year for us excellent well we hope to see you at some them in future best of luck with

the launcher thanks so much thanks for talking to [Music] me