Paul Muller from adjust on Mobile Ad Analytics – Tracking Conversions And Cost-Per-Install


Paul Muller is the CTO and a co-founder of adjust which offers one of the leading app analytics and tracking and intelligence services.  He is responsible for building and scaling adeven’s technologies. Previously, Paul was the CTO and Managing Director for two companies, rapidrabbit and Müller & Wulff GmbH.  Paul spoke at App Promotion Summit in London recently and covered the topic of mobile ad tracking and how to capture conversion and cost per install costs using analytics technology.  He focused on one of the issues developers frequently mention when it comes to tracking – the importance of a stable SDK that won’t crash your app and covered the following areas:

  • How To Track Your Campaigns – Best Practices, Hints And Tips
  • Alternatives To Cookies– Fingerprinting And Server-Side Tracking
  • Optimizing For ARPU And Active Users

We have the “TECHNOLOGY DEMONSTRATION: MOBILE AD TRACKING” talk recorded in high quality audio.

You can check out his presentation “TECHNOLOGY DEMONSTRATION: MOBILE AD TRACKING” below:

And we also have transcription of the talk below:

Paul H. Muller: So first off, hi. So why am I standing here? Why did I build this Ad Even Company. I was an ad publisher for five years. In January of 2009, we started ad publishing and we’ve basically been through the whole story with the iTunes store and the Google Play store. I published around 25 apps. We generated a ballpark of figure of 10 Million downloads and we spend around half a million of euros on marketing; and at some point we figured that we kind of need to understand where all of this money is going and what it is doing. And out of this need, the idea of Ad Even basically was born, and a couple of years later, we found a VC to take us and went to bankers and there we are. So that’s me. Ad Even has [inaudible 01:13] because the BBC article about the zombies, that’s from us. They’re also on CNBC, Huffington Post, all kinds of [inaudible 01:21] kind of big news sites. So what we do is download tracking, and combines with analytics, and combined with market research, so it’s business intelligence for apps. So what app download tracking is for was also lastly was introduced a lot here, so you should track, track everyone has set that, right?

So the point is if you want to track your downloads, the most important thing is, and that’s kind of a very unique standing from us, is that you need a third party independent company to do that, because if the guy that is selling you the downloads is tracking your downloads and tracking your users, you have two problems. First of all, the whole conflict of interest there, and then again, if you choose just to buy downloads from some other guy, you’ve seen kind of a lot of companies that want to sell you media downloads, users, and everything, and if you want to change those, your data will be stuck with one of them and maybe not comparable with the other one. So you need an independent party that tracks all of that for you and then you just pay for the tracking and not for delivering the users but for getting you the data that you need make an informed decision.

So when you’re looking for a tracking download company, what exactly are the kinds of the things that you have to look for? The first on is of course transparency, because you don’t want to fly blind. And I mean true transparency that you own the data that they track. It’s your data and you get to keep it. You can try as many sources as you like, and you can try to compare them from the same dashboard and you can see what they are doing no matter where you’re buying your dashboards. And so you always know exactly what’s happening, and you have a company whose own interest is to track the most precise data; not just to tell you something that their users that they actually sold you are actually good. Transparency means also to understand what the company is doing, because technically, we’ll come to that in a second, it means that you have to put a [inaudible 03:17] into your app, and I don’t know if you’ve asked your developers to do that before, but usually that is not met with cheers. And transparency means that you have the technology that how the tracking works is understandable to you, that you know what it actually does, and that you can understand what it does with that.

A result of that as well is the whole privacy issue. You are all aware that the European privacy laws are coming along and adjusting a bit to the kind of German standard that we already adhere to. And what it means is that if you have a download tracking into your app, you need to care about if it’s privacy compliant or not, because it is not going to be this company who does the tracking that gets the problem when there is an issue with the privacy. It is you, because it is your responsibility. So if you put anything into your app that hurts the privacy or kind of violates the privacy laws, it is pretty much your lawsuit and not theirs. So if you put in tracking, make sure that it’s privacy compliant.

In Europe, that is a big topic. Some of our biggest clients are really, really, really keen on this kind of privacy thing. We have a constant privacy audit for his kind of stuff, so you should care too. Even if you think like now maybe that isn’t a big topic, but you don’t want to be landing in the news with a “This app spies on you” saying. That’s not cool and that’s the kind of PR that you do not like for your app. And one of the most important part, of course, is comparable data. I’ll get to that later what exactly I mean about that, but it means that you have data that you can base your decisions on and actually make the right decision, because there is a lot of data out there. There is a lot of people tracking stuff and maybe you have three or four companies where you buy downloads, and they all kind of misplace something, some RPU, some lifetime value, some KPIs, but most of the time, they aren’t comparable to each other which makes it really tricky to compare them and to make a decision.

So let’s start from the beginning. If you want to do download tracking, you need to know the [inaudible 05:17] you have, and that’s a fact. And 90% of the time, you end up asking your developer to put in a compiled STK, and yes, you may have tried to do that, and usually you should run the developer ever asking him, because it will crash the app. It will. Trust me. I’ve done that five years, and it will crash the app almost certainly. And it will crash when you deploy it to the store but never when you test it, because a compile STK is a black box that you have no insight to, and you just strap onto your app, throw it in the store and hope that it works; which it doesn’t, because most of the time, it is extremely poorly made. Because no one cares about the STK, it’s not their core asset. It’s not what they do. It’s just the thing that they need to sell you downloads. So they’ve got this STK for a really good price on, it’s compiled. Who is ever really going to notice that that doesn’t work? And if you throw in three, four, five, six of those, because you want to buy from different sources download, they start haggling with each other, because some brilliant developer had the idea to call a certain class something just [Ad], and if a true STK tries to do that, yes, it doesn’t work. And you can’t do anything about it, and so if it doesn’t work, you’re basically sitting there with the information.

There is something wrong within that black box and there is nothing that you can do about it. So this leads to some extreme frustrations to a great adversity in the market for developers to put STK’s. And I said, “Look, we are going to do this whole download tracking thing, there is one condition. We care about this STK stuff, because it is really important. It’s like the most important thing. I mean not for the business people that sells it, or the people that want to see the data, but for the guy that actually has to put it in. It has to be nice. So what’s the answer?” The answer is pretty much letting your pants. It’s an open source STK, which means you get the recipe and not a cake. That means the developer that puts in the STK can see exactly what it’s doing and how it’s doing. It can’t hide any trade secrets in there, but you can also not add any bad code in there. If you fuck up in that, the developer is immediately going to tell you that this company is an imbecile and you shouldn’t use them, because their technology is bad. So if you’re doing an open source STK, you have to make sure that it’s actually good and that’s the hard part, making a good STK, and we’ve gone through a very long process to actually make that happen. To have a STK that all of our customers say, “Guys, this is one of the best STKs I’ve ever seen. It was one of the smoothest integration processes I have ever seen, and it doesn’t really interfere with other STKs”, because it’s not built into it the same way. It basically gets baked into your app and then becomes an integral part, and so it can’t just crash randomly under certain circumstances, because if there is an error, you can actually see where the error is. So the bottom line is if you want to put in a tracking STK, and you don’t want to fight with your developer, get one with an open source STK like for example, ours.

Also, a really important point for you is if you want a tracking STK, it saves you all of the other STKs. So that means you want to have a STK that can get rid of you for the Facebook STK that you made, and just only have in for the mobile install ads, the [inaudible 08:39] STK, and I don’t know, like we replace around 65 different STKs that you don’t have to put into your app because of us, and trust me, that will make your developer happy. And coming back to really one core point that we can see specifically with biggest clients being the foremost issue, we had audits from like 12 law firms so far from our big clients that made sure they were privacy compliant. The problem with a complied STK also is I can put anything in there, you can’t see what it does. I can hide any kind of suspicious activity in there, and so that is technically not compliant with any privacy thing, because it can’t be audited. Our STK can be audited or any open source ticket for that matter can be audited, and that is actually why you should care about us. I mean I’m really stressing this point, you’re thinking oh privacy, I’ve heard about this. It becomes really a more and more important issue, and it’s one just one of the boxes that you kind of need to take off when you put something in your app to make sure that you are on the safe side, because you want the data and we know that you need that data, you should track, track, track, you’ve heard that, and you should also try to play fair while doing that.

And now coming to like the most important point of this tracking stuff is like what you track. You can track all kinds of numbers, and you can have all kinds of colorful dashboards that display all kinds of colorful numbers and that do look good. The problem is that most dashboards display not really any sensible information. That’s tough to say, but it is, because displaying sensible information is really hard. I’ll show you in a second in the demo, but, what it means that if you look at the user for example, it doesn’t make any sense to compare campaigns just by wrong numbers that have been run in different time frames. You need to normalize a user onto a single time frame that you can compare them. Say a campaign is two months older and user has spent on average like two euros, and it’s a one month old campaign that they have spent an average one euro. And you would say, “Whoa, those with two euros are better users. I’m going to buy more of those”. That’s wrong. They’ve been around longer. They had more time to spend money on your app. And so to figure if they are actually comparable, you have to kind of say how much do they spend after 30 days or how much do they spend after seven days and stuff like that. Always have KPIs that are comparable, and the best way is to do a cohort analysis to find some kind of common criteria, usually some kind of time frame after the [inaudible 11:09] that you can kind of cohort the users together with.

The point also being with the comparable data, there are not too many more one measurement partners. Facebook, I think, is right now like nine or maybe ten. And if you want to run Facebook campaigns, and you want to see the Facebook data in the same dashboard that you buy like your [inaudible 11:29], as you buy any kind of AdWords or whatever, you need a company that’s also a Facebook mobile measurement one, because otherwise, you have your Facebook dashboard with those numbers in there and then you have your other dashboard, and you don’t want that. You just want one dashboard where go in and where you see all of your information. You need to make a decision for your marketing. So I think that’s enough talking on pretty slides. I’m going to show you how it actually looks in real life.

So there we are. I’ve never worked with kids and live demos. So okay, what we’re look at is the dashboard of our example, our demo, access, and I’m going to jump right in. I’m just giving you a real quick overview. We only have a couple of minutes. So you should basically see in your dashboard, the kind of raw information that you need, like how much [inaudible 12:18], how much clicks did I get, what kind of users did I get out of this, and what kind of revenue did I make. These are the kinds of basics that you need to understand based on where they are coming from. Here they are different sources, and you can see what each kind of source is contributing to your revenue, and you can select time frames.

Let’s look at All Times because it usually has a little bit of more information going on. And you can see that they make different kinds of revenue from different kinds of sources, and this is kind of your quick overview to see actually what’s going on today and how much revenue did I make, and where does it come from. But, to actually compare this information, you need a cohort analysis like this. I hope it’s visible, but what you see here is that each individual source has an assigned cohort, like seven days, thirty days, and lifetime, and has certain values. Like for example, the RPU, and this is actually real data. It’s not fake data that we made up. This is an app that we stripped of all its names to show you how real data would look like, and you can see that they’re normal organic user gives them $.24 revenue in seven day. So this is kind of the baseline that they’re looking for. When they’re buying traffic, like how are the users stacking up against that. And when they are buying, for example, this is their internal, this is some kind of barter deal with themselves, they make $.60. So they know that these kinds of users are extremely engaged compared to the organic average, and they can see that other sources are performing less than that.

Okay, so there are all kinds of cohorts for longer different time frames where you can definitely see that overtime, the users may actually develop a little bit different, so some of your users that end up spending a lot of money in the beginning may not end up spending money in the end. But what is important for you that you need be able to see these numbers lined up next to other and say, okay, now that these numbers are there, they are comparable. They have been cohort and analyzed for fixed time frames no matter when these campaigns run. So the data that you saw were campaigns that were running apart three or four months, and yet you are able to say, “Okay, these kinds of people make more revenue, they make less revenue”, and it’s an actual thing that you can base your decision on. Not like, “Oh, this is no longer around so I can compare stuff like this”. I would love to show you, but the dashboard goes on there, for example, showing you ratings, rankings, and reviews within the same dashboard, so you don’t need to log in and log out, open you [inaudible 14:59] and jump around, but in the same dashboard see [marketing] campaign is doing to your rankings and you can see what it’s doing to your rankings, and you can see what it’s doing to your ratings and your reviews. What are the people actually writing about you, and if you want to give this a try, it’s pretty easy. You just go to , sign up today, everyone gets a free trial so you can just play around with it and see how it works for you. Ask your developer to put in the STK and see what he says, and trust me, you’ll be surprised by the unusual response, and then you can actually start tracking and see where your traffic is coming from and where your good users are coming from. So then, thanks.