Lifetime value for mobile apps

[Music] [Music] okay good afternoon everyone yes my name is Michael level as I've already mentioned and I've worked with a company called Paulin VC okay sure yep that's great yes okay yeah I don't know how many of you know Poland we see but Paulinus is FinTech company who enables and game companies to grow fast by giving their the access today earned

revenue in weekly basis instead of waiting waiting 35 to 40 days that money coming on from the platforms we also do user acquisition support and part of that user acquisition support is is LTV modeling hence today we are little bit talking on on the LTV LTV is a measure of expected total earnings for what you would get from a user on its lifetime within your app or a game so it gonna give you a gun of monetizing some potential of your game and it's a key metric to use I want to understand if you if you want to gonna do any paid acquisition in scale because you need to really have the customer lifetime value to access your your user acquisition cost your your CPI CPA and when when you have that model done right then you are

ROI positive and you you can really scale up your business but anyway before we go on on the panel I would like to also hear a little bit your background so how many of you are working indirectly on on user acquisition within your work just gonna hand some okay good good do we have any any product managers also yes anyone you know developing the app developers okay so mainly you are from the UA it's a school so then we can shift our answers accordingly okay but let's go and sit down okay I'm not sitting down but yeah so welcome and if you could just quickly state what is already shown over their owner power but you know who you are where you come from and company and hi Michael is my name I work with small

giant games our most recent game is called empires and puzzles it's a great match three puzzle RPG game available for free on the App Store etc I work as a user acquisition manager so solely using LTV for various functions including how I can actually by users and know when they're making money for me I what's the price that I can pay for users so that's my LTVs morning good afternoon everyone my name is Peter fedora I'm a founder of a patient we help mobile companies both startups in apps and games to grow their user base and revenues through five areas mobile marketing strategy data analytics creative services user acquisition and growth management today we work for example for Luca or CI games hello everybody my name is Matthias Petula I'm product manager from two dreams

basically I'm in charge of the portfolio of the games we are the greatest of the driver heads family of games and I joined the team three months ago and I'm really happy to be here talking about something that difficult as LDV yeah my name is Regina I'm the product manager for Amara's first game quantum siege we soft-launched enough made this year and heading for hard launch interesting times that can say yeah I joined a matter one year ago roughly and yeah happy to answer all your questions on your TV and but it has to do with product management okay thank you okay let's let's start this little bit talking about you know overall metrics routine and analytics on daily basis okay from the Regina you have now as you said okay you have a game in a soft launch how do you gonna work on weekly on daily level with with the game metrics and and how how how do

you do it so on a daily basis we look at the user suspects it gives insights on new users from our main interest in channels also our activity and progression metrics checking on a daily basis what now in saffron's is extremely interesting for us is the retention rates especially the retention rates also sanity check on a daily basis on them and of course revenue but on a weekly basis we work together closely with our UA experts to see what we can improve on our campaigns also what how we can improve our marketing creatives and on a weekly basis we also check like currency balancing on hard and soft currency with to be signed to him so that's a domain regular metrics views and of course our

TV but that's more passes I would say okay Peter I mean you you're coming from the agency point of view so how do you work with beat the gun of this analytics and the matrix within the agency you know this is basically the same what Regina mentioned we divide ApS into two areas first one is so-called top of the funnel which means everything what happens before these are opens the app so CTR of its conversion rate in the store of course cost per install course person action and then KP is related to the product itself so retention onboarding hope it will progress to the first couple of screens up to some very specific metrics which are like specific for that game or game genre what we try to do is to look at trends we quite often look at daily cohorts especially when there is some change in

the campaign setup or in the game itself some new mechanic or change of the product itself then we can understand if that particular change is improving the performance or not so this would be my advice to look how it evolves day by day and make some conclusions from it ok then ok on those metrics aunty as you have already had the game out further from the string so how do you kind of what would be the con your advice for the develop when they can click on a hit on the gas and start scales so what what are you looking in the metric side and and how do you interpret then that you know that hey now this is okay we can can push the middle right I think the most important thing is consistency and seeing that you have a very clear that you have the best possible idea of what your product actually is and where it's

actually performing so let's say you have day one day seven retentions for example that are very sustainable and there's no major spikes in in no matter sort of where your your user intake might differ and will differ day to day but at least so that players actually you stay there and then of course monetization is another thing that you might want to keep an eye on whatever that metric then is that you've decided to go for but basically when you get to a point where you can have certain amount of consistency and you rely on your modeling beat from pollen or be it internally built when you have a very consistent track record then you'll know that okay now is the time that I want to start testing whether or not this will actually sustain so I think whatever your main indicators are that they're consistent is the most important thing okay yeah mattias do dreams is more on

in the arcade side so how do you how do you do you see the things differently in in your genre and in minimal strategy games or how do you kind of know basically the main challenge of the kind of games we we create it's try to identify the proper segment where we are getting the best results and based on the behavior of that segment start building the strategy in terms of features but basically we try to keep an eye on which is the segment that is actually bringing the most interesting amount of revenue and around that star building new features or making changes to the gameplay or to the type of all personal that always aiming to improve their lifetime value okay yeah okay let's also talk a little bit about user acquisition in the UA it's all the time changing as as you al over so how is

your gun of user acquisition strategy currently it's more kind of the CPI that you know that people looking in cost per install is little bit kind of you know going away it's more on C cost products and cost per event do you see this one it kind of happening in your in your work as well Peter yes of course this is a thing which works very well for our clients of course it's much more expensive when you go for a CPA it could be like three times more the CPI for the same title and register same target group but also we see the growth of the profit about like 30 percent for example so it's definitely something you should explore probably not in the early stage of the product and marketing testing in the software but once you go out for a global launch this can help you to attract really high quality users and

the magic is in how you define this cost per action what it should be and we'll probably discuss it later on them so those events are important so any any kind of practical tips on what events to field over there and also what tools do you use basically what is your analytic stack what you use anything on that Regina yeah yeah now for matter we haven't deeper at the moment but from our previous companies we always used to purchase as event that was really crucial but also the price extremely high so for casino games for example I was working on it was around $90 for the CPA former I would say we should look more into we have a PvP mode and we know that people who play for example 5 PvP battles or the first day they have extremely high lady tensions

on day 30 it's around 17% already so we were looking into that and I think I hope in the future we can define and go for the CPA on debt metrics or yes let's also talk about a little bit on on the LTV yes as it's there it's there it's the title over here so how do you see the LTV over all the methodology on condom measuring the LTV how it's calculated and so do you example do it for the you know to model it out for the 90 days 108 days or how do you how do you do it auntie well for us for for gaming there's basically there's a few different routes you can go you can either try to model it based on your monetization monetization is sort of scaling up according to where however long the game goes on or you can do it

based on based on the retention curve of the users and 50 or more decision to that we have a pretty simple approach in terms of where selecting various different cohorts depending on be the daily or country-based or always based and then we measure the entirety of revenue that's come in for certain cohort divided my team on three get basically revenue per install and then just fit that to the to the modulation curve and that then just backs up into a certain LTV at some point in time and that's combining and then on top of that we then put ad revenue in addition to to our in-app revenue which is the majority of course for us being a strategy game but the Adele TV itself I think deserves another panel or something because it's way beyond my control at least to understand where where that's coming from and that's a big challenge that a few companies have been able to really solve

but those that do that have been very special with it okay so you haven't mentioned the K factor which is you know the simple o TV model is a lifetime multiplied by average revenue per daily active user but it becomes more interesting when you start discussing what's the variety how what's the K factor how you count it and because we work with many teams I've seen very different approaches to this one group of companies don't consider variety at all in their LTV modeling I'm not sure if it's your case because you haven't mentioned it then there are some companies which consider everything else then paid install as a variety so they put everything in one basket the third approach which I personally prefer is that I try to measure the organic boost on the top of my paid acquisition efforts so if I buy one user then

attribute let's say 20% more which is correlated to the install so only variety coming from paid efforts and the most sophisticated companies such as Miniclip for example they are able to analyze the data so well that they can say this is a key factor for featuring by app or Google this is coming from in-game invites etc but this requires most of data and very good data analysis team to be able to go to this granule level to me it's enough to count what comes from the patents by platform country and also channel because this might differ and that's kind of like conservative approach for small teams it works very well yeah well we are a little behind in terms of LTV calculation because of the nature of our games and and also because they the main

challenges our games mainly monetize through ads the calculation of a RP you sometimes is not quite accurate to make a projection so now we are working on gathering all the information possible from ad networks him from the game to have a clear understanding of which is the ARP you as a main element in calculation of the LTV and also one variable that it's important for us to have a clear projects you know clear understanding of which is LTV is a turn rate it's one element that it's really important to have a at least an idea of which will be that projection yep okay how do you how do you actually do the LTV measurement I mean any kind of tips or tools what you use but you've basically you know advise people to use

any not not stating the obvious in the room but hunch we take all the data we can I get the moment we started quite simply with just a simple retention curve so that's using a the weighted sums of small day one day seven day 14 day 30 retention and we use that at beginning so last week I started to use also the famous Eric spreadsheet okay it's really easy to use you just the first day for you retentions of them they've wanted to vary retention and then copy pasted in the spreadsheet so you get a nice integral after retention function and I would recommend using this it's pretty easy to use you know for Mesa there I think that spreadsheet is where everything is started matheus what tools do you use we are in the process right now okay okay

trying to use the current tools basically you we use a lot of commonalities but basically in this moment we are working with a new that data scientist join our team and we are making a lot of changes into the games to gather that information and to have our own sources of data to have a clear measurement for the beginning I believe that Excel is a good starting point and then if you have someone it's a bit more capable of doing some API connections than Regina mentioned with Google Spreadsheets that's the second level because you can import data automatically for example from Apps flyer or add just so it is can be calculated automatically and you can see some LT Vickers without manual work and then of course it's a

natural way to progress to integrate these into your BI tools and have it in a more complex setup but for the beginning what was mentioned before I think it's a good start and it's the easiest way how to figure out what works best for you without the small costs yes so for a small giant we've been there where where do you just have the Excel and you just try to run everything as smoothly as possible based on that but then after making a few games that with only that that's the background came to the realization that we also have to build something like really really capable so so the guys were able to build basically their own analytics engine we know within a year and and after that it's it's just made my life as a marketer so much easier because I can I can pretty reliably get all the data that I need so so if you have the time and resources please build your own

thing because it's totally we're gonna back out for you okay excellent pretty good thanks no I think we have also some time left so we would kind of do any questions from the audience okay all right Frederick look under here so my question is x-ray actually related to LTV and also the other side of the coin so CPI or CPA as you were talking about so if the event in when when you're doing cost per action kind of campaigns if the event is that the person needs to spend then essentially what you're talking about is cost per paying user not CPA right so the cost per paying user the flip side of the coin is no longer LTV as we would normally call it but it's the LTV of spenders because they have different retention profiles right so how have you

guys gone about that do you have spender retention profiles separately and then create the LTVs for spenders only or do you still use the kind of generic retention rates when calculating LTVs anyone yeah you put a spin a bit of a tricky question but thanks for that so CPI is definitely a world of its own and and a lot more ad networks are now bringing aboard CPI best models and those are sort of trickier for advertisers also to figure out what to do with so the valuation dilemma is exactly like what you've said that do we should the the more specific you go the further way you get from sort of broad enough cohort type thinking that it actually ends up being worth your time because you can spend a lot of time

trying to model very specific people's behavior and then you're missing out on the forest's basically of what you're really trying to do with you a which is to increase your audiences so I'm not saying you shouldn't do CPA based things and you I'm not saying you shouldn't do very deep funnel of things you just need to consider that what how well is your time spent actually figuring out these things maybe well I tell one more thing which is about the definition of the cost per action what action should be purchasers are one of the goals you can use for Facebook or Adwords but it might happen that your game monetize is later not the first day maybe in a week maybe in a month in that case it's necessary to figure out some event which happens on the day one and it happens in let's say volume to ten to twenty five percent of users performed at event on the first day and then you can optimize all that Facebook and adverse optimize campaigns very quickly

without waiting for the purchase in a long term and then you become somewhere in the middle of the cost per install and the cost per pair which you have mentioned because you are paying for some prediction which in some time will generate the payment oh yeah I've got he's got a very small height there but I'm going to still still find you thank you could you give some examples for kind of campaign manager rules of thumb for when you're buying users at scale how many days you kind of give a cohort before you estimate it and what kind of rules of thumb do you use is it just a single analytics event it is a combination of events or

is it the the ROI or is it something different [Applause] [Laughter] very quickly a great question thanks for that we always look at several KPIs because it might happen it you're acquiring group of users which for example will pay later on and if you look only to payments you excluded them so if there is a good retention for example or good some custom post install event let's say level free reached etc then you keep it running for a couple of more days to make the final decision about you always look at more things and I think that were so the expert value comes that you are able to think deeper than some algorithm and give it some time or stop it even sooner than some agree and do it so I'm gonna assume that

the of course there's several types of campaigns either there's something that you want to have results right away and then just do a burst or something like that and get results immediately and then scram what I'd like to think about campaigns is always like sustained always on type UI campaigns and for that at least what I've seen as a sort of rule of thumb is that is to give a campaign at least seven preferably fourteen days at the certain scale to give you the sort of idea that the consistency of said campaign is like good enough that you're buying let's say from Network X and they're saying okay what I want attention seems to consistently be this and my day three-inch and also seems to be consistent through this and then you can back out revenue based on that and then the more sort of day seven also gives you enough time to to give your model enough time no matter what it's based on that that you can also have a more defiant idea of what this what this sort

of user source is actually giving you and then after around seven days or amateur seven days you can then figure out should be bit higher should be bit lower she just post completely or or or what should be done Tammis want a question over here yep thank you a manual attack chocolate I have a question from all panelists we talked about LTVs I would like to kind of compare performance so of course we were disclosing you know kind of internal figures but maybe comparing with a base of 100 for like a Navarrete organic users in the US market how you LTV comparing various jails across Europe compared to us and across different channel of acquisition compared to

organic like user acquisition and Facebook AdWords video ads etc compared to index of 100 we're like over below what kind of performance for channels and for geography so for me that's a fairly sort of broad broad question but for me the at the end of the day as a marketer I approach it that I don't really sort of consider the LTV as my main goal the strategy that we have as small giant is the return on adspend is the main goal and the LTV is kind of the the vessel with which I know how much money I can actually spend and then at various points in time be 30 days 60 days 90 days etc I need to be able to hit that goal so whether or not the source has a high LTV that's great that means that I can bid higher get more of the same so in that sense it's more for me a ways to an end and the end being to

grow the game as much as possible and still a way to make money on it I have to agree we basically all net profit in Indonesia at the CPI could be 50 Cent's and LTV 75 and it's good and in the u.s. it could be $5 for install and OTV $8 and still it works for as well so we don't look at the LTV as a number but at the profit the question is how much profit do you need how do you calculate your company costs if you include your you know wages etc so some companies look for or recouping their advertising spend only and some want to get back 150 percent of their investment that's a completely different discussion [Music]

[Music]