Making mobile ads suck less (and playing soccer) with Moloco VP Francesco Renzo
Why does every tiny engagement on a mobile ad on iOS seem to pop up an App Store overlay? Because as Moloco VP Francesco Renzo explains, the mobile adtech ecosystem is like a soccer game. (Or football, if you’re not North American.)
Bear with me on this.
It will actually make sense.
Mobile attribution: a global game of soccer
The mobile attribution games, Renzo says, are a global game of soccer. Ad networks are the teams. They’re trying to score goals, and they achieve that when they find a new player, user, or customer for their advertisers. Everyone used to know exactly where the goal was, thanks to the trackability of the iOS ad identifier, the IDFA. But since App Tracking Transparency made IDFAs scarce, it’s become much harder to tell which shots hit the target and get the right new users, and which miss by miles, to the sarcastic whistles of a scornful crowd. The result is that ad networks now shoot from everywhere on the soccer pitch: well outside the 18-yard box, their own side of the halfway line, the corner of the field, from their own goal line. Even, sometimes, from completely off the field.
Each of those shots is a triggered SKOverlay that fires off a “click” in SKAdNetwork.
Each is a chance to score a goal, win the game, add value for advertisers, and enhance the ad network’s ability to sell its services. So fire away, damn the torpedoes, charge of the light brigade, and bombard users with pop-up StoreKit-rendered invitations to please pretty please DOWNLOAD THIS APP NOW.
The only problem, as I chatted about in bad ads and the tragedy of the commons and as Mega Games CEO Adam Jaffe shared in a recent Growth Masterminds interview, it doesn’t make for a great user experience in publisher apps … the ones with the ads. It can make mobile ads suck.
Clicks and goals are good for ad networks, but a horrible user experience is bad for the ecosystem.
“There’s a bit of a disconnect between short-term incentives and long-term value or long-term ecosystem health,” Renzo says.
Plus there’s another disconnect for publishers: too many non-converting ads means they’ll run the risk of essentially demonetizing themselves.
Clicks, clicks, and more clicks in mobile ads
Renzo and I recently chatted over the problem in a Growth Masterminds podcast that I titled Making mobile ads suck less. Our non-sports goal: determine what’s causing the massive jump in CTR and concurrent poor user experience, and discuss how to fix it, how advertisers should deal with the situation, and what this all means for measurement in the era of privacy.
It starts with something that surprised me: there is no real definition of a “click” in SKAdNetwork.
“There’s this very interesting page on Apple’s documentation, which is about winning postbacks,” Renzo says. “And on SKAN, a click is equal to a view. Literally, the word click is not included in the winning attribution postback documentation.”
So adtech companies generate clicks via taps, missed X hits, attempts to play a playable.
And once one network starts, the others are forced to follow or their CTR looks bad, they take fewer shots on net, and they score fewer goals: all essentially suicidal business strategies.
It’s an arms race that hopefully won’t end up in mutually assured destruction for the ecosystem if users/players/customers start opting out.
But how do we fix it?
Smarter measurement models
Incrementality, media mix modeling, and all the other measurement methodologies that Singular is building as part of hybrid measurement is a big part of the solution, because that is immune to click spamming and other attribution shenanigans. Auditing your media sources, Renzo says, is another one, including asking the hard questions around what is a click, what are the tactics, when they fire clicks, and how they engage with attribution partners.
There’s also some promise in more granular click reporting, Renzo says.
“You can have a click that was explicitly on a call to action button. You can have a click that was perhaps an engagement with the playable component of the ad. Or you can have a click that perhaps is in the area where usually the X button is or an unqualified click that’s happening somewhere else that we cannot quite tell if it was indicative of installing intent or not. If advertisers had visibility on those types of click interactions, then they could attribute different value to those different click interactions. Just like now, for a click versus a view, you have a hierarchy of attribution.”
There’s much more, of course.
Watch the full video above, and subscribe to the Growth Masterminds podcast on your favorite podcasting platform.
And … here’s a full transcript of our conversation …
Making mobile ads suck less, with Moloco’s Francesco Renzo
How can we make mobile ads suck less? Hello and welcome to Growth Masterminds. My name is John Koetsier.
I recently published a massive post on Singular’s website. I was a little pissed off, not so much at mobile ads per se, but kind of at the infrastructure and architecture within which they’re delivered. The fake clicks when you try and open the playable, the imaginary clicks. when the ad just randomly opens the App Store, the non-functional X button or the intentionally tiny, small little X button jammed into a corner.
Is all this fixable or is this the end state of mobile advertising?
To chat, we’ve got Francesco Renzo. Beside his super Italian sounding name, he has nine years of experience at Google helping app developers win. He has an MBA, he had a stint at McKinsey and is now VP go-to-market at Moloco. Which is, as you probably know, a performance advertising platform that relies heavily on AI and machine learning and did very well, by the way, in our recent Singular ROI index.
Thank you, John. Great to be here. Thank you for having me.
Hey, super pumped to have you. Let’s start at the beginning.
What sucks about mobile ads right now?
Oh look, just like you were hinting earlier, you know, there’s a lot to be liked about mobile ads.
You know, they power the internet, they power a lot of the free content games and apps that billions of users worldwide use. You know, sometimes I have this conversation with friends who blame me for all the ads that they see online. And I’m like, look, that’s probably the biggest wealth transfer in history. You know, there are rich high rollers who watch ads that are basically funding … free content for a bunch of other people around the world who would not be able to afford those services if those were paid.
So there’s a lot to be liked about mobile ads.
There’s also a lot that sucks these days about mobile ads. I thought that your blog post from a couple of weeks ago, if I remember correctly, really hit the nail on the head. The user experience is quite terrible and there’s a number of interesting incentives in the ecosystem that have led us where we are. If you think about what deterministic mobile ads looked like a few years back before ATT, there was a precision in how those ads were targeted, and every player up and on the stack could tell exactly which users had seen a certain ad and could clearly attribute the outcomes that that ad drove for that particular user.
With limited ad tracking both on Android and iOS and then ATT on iOS in particular, that whole predicament went out the window.
And so, you know, there are two things that I obsess about. And you know, one is mobile ads and the other one is soccer or football if you’re European like me.
And you know, one metaphor I like to think about, it’s a little bit like, you know, the attribution games that are played on mobile ads are a little bit like a soccer game. And the teams are the networks who are serving those ads. And scoring a goal is finding a user who will deliver good results. And everybody knows where the goal is, what the goal posts are. And so the players, the network to serve those ads, are very precise at shooting on goal.
And limited tracking a little bit got rid of the net, got rid of the goal. And you know, it’s hard to tell which of your shots in that general direction are hitting the right users.
And so what’s happened is that people, people as in the networks who serve ads, try to take as many shots as possible towards that end line of the pitch in the hopes that some of those are going to turn out to be good users, right? If the referee in this silly metaphor is the advertiser, the advertiser is just going to see: Hey, how many balls eventually did end on goal.
And the team that wins the game gets more budget from the advertising. So it’s quite unfortunate, but there are also some really powerful incentives that got us where we are.
So it’s Apple’s fault, ATT is causing … 🙂
That’s probably not the correct interpretation of what you said, but it’s one possible.
Look, we have the challenge here and we’re gonna have the challenge on Android as well. As Privacy Sandbox rolls out probably early to mid or late 2024 – we’ll see when that happens – there’s a little bit more targeting capability in that than we’ll have on iOS. Although SKAN 5 might borrow some things from Privacy Sandbox, who knows?
Before we talk about how to fix it, who can fix it, and how advertisers survive in this game of football, what’s the risk factor here?
I mean, all these networks are trying to score a goal. Thousands of balls are being pumped towards the net. The goalie is like ugh …
What’s the risk factor? How does this impact the ecosystem, which is the people, right? The players, the customers, the users of these apps?
I wouldn’t say it’s Apple’s fault. I would say, you know, there are some very legitimate arguments for, you know, limiting ad tracking.
The problem is that the ecosystem is a bit more complex than that. And so there have been some perhaps unintended consequences. And, you know, you’re asking the right question.
If ultimately the biggest damage gets to the user experience, then the next natural question is … what does that do to the rest of the ecosystem? And what can the rest of the ecosystem do?
I think the challenge is that, you know, you were calling it a tragedy of the commons in your blog post. I think it’s correct. It’s accurate.
Another way to qualify it is also, it’s a bit of a disconnect between short-term incentives and long-term value or long-term ecosystem health. And the problem is that the short-term incentives for the players in the industry are all very stacked towards these kinds of outcomes.
And if you start from the ad tech companies in between, the publishers and the advertisers with the users on the other side, those ad tech companies benefit from winning that football game, right? And they have to contend also with the fact that measurement, which is critical of course for performance marketing, is now in this interesting state where we have the SKAN framework that Apple has given on iOS, and then there is all the different attribution approaches that are SKAN-friendly but somewhat distinct that the mobile measurement partners put in place, whether it’s probabilistic attribution or whatever other labels different MMPs are using.
In this environment, if you are on SKAN, there’s this very interesting page on Apple’s documentation, which is about winning postbacks. And on SKAN, a click is equal to a view. Literally, the word click is not included in the winning attribution postback documentation.
And then there is this concept of a fidelity score that applies to SKAN, where essentially, in addition to who delivers the last click, you get a higher fidelity score if, for example, you render a store kit ad, whether it’s SK Overlay or AutoStore.
So you get an advantage.
So now you see a little bit of the advantage of SKAN measurement from creating all those overlays or those additional App Store experiences as part of your ad experience. And then if you are measuring using MMPs, the challenge for those app tech companies in between is that the StoreKit-rendered components of the ads are purely controlled by Apple.
And if a user clicks on one of those because they were engaged by the ad, they liked being sent or being exposed to something that’s similar to an App Store experience, But the ad tech company that served that ad does not know. The trackability of that click is entirely lost to the ad tech company. And so for an advertiser who uses a mobile measurement partner, how do you know if that ad was helpful or not?
Well, the ad tech companies have addressed that problem by resorting to a number of tactics that get more clicks fired or more attribution signals passed on to the mobile measurement partners. And everything that you were pointing out earlier, whether it’s the very small X button or the multiple store exposures, or the inadvertent click that miraculously brings you to the App Store, the playable engagement, which is marked as a real click. All of those ad interactions, some more engaging and positive, and some a little bit more damaging … all of that ends up being a click fired to an MMP.
So … for the ad tech platforms in between, the incentives are very powerful to move in that direction. If one ad tech platform doesn’t do that, they will be at a huge disadvantage versus their competitors who are moving in that direction. And then you have the incentives on the publisher side, right? The publishers are on the receiving end of that bad ad experience. And if they were to totally step away from it, the algorithms that power adserving penalize their inventory.
Their ads would appear less performant, and therefore they would lose volume. Now you will say, well, but I know for a fact that when I have a terrible ad experience, I hate that game, and I don’t play it anymore.
The challenge is that, as I was mentioning earlier, it’s a trade-off between a very real short-term loss of traffic versus a fairly uncertain and unmeasured, I haven’t seen any meta study on that, loss of customer loyalty and increased churn.
So for the publisher also, the incentives are really stacked around playing along these rules.
And then there’s the advertisers, who of course, they have to drive profitable growth. They look at their cost per install. They look at their return on advertiser spend. And all these techniques eventually help drive down those performance metrics. Also because a lot of these very aggressive attribution tactics end up catching some organic installs, right? Users will be installed anyway. If you were to clean everything up with a magic wand tomorrow, probably CPIs would go up and return on advertiser spend would go down. And so cleaning things up is not only hard because it’s a systematic issue that no advertiser can individually address. And also it’s not like my performance review as a marketer is damaged by this, if anything, it’s helped by this.
So up and down the ad tech stack, there’s very powerful incentives that led us to this status quo and make it a challenging uphill battle.
You supplied a soccer or football metaphor. I’ll supply a hockey metaphor – ice hockey, because I’m Canadian and I play, and it’s the playoffs right now as we’re recording this – and there’s a saying in the playoffs, especially overtime after the traditional three periods and teams are still tied, the score is tied.
There is a saying that “there’s no bad shot on net.” And that is apparently. the ethos in the ad tech ecosystem right now.
Okay, so you’ve outlined why it’s happening. You’ve outlined what are the factors that are causing this. We talked briefly about the risk factors, which is maybe somebody churns from an app, that’s possible. Maybe somebody just doesn’t go for those rewarded ads anymore. There’s a rewarded ad unit, they just don’t do that anymore. Maybe they turn off mobile a little bit and turn on to something else or who knows what. Lots of things there.
So let’s talk a little bit about the fix. If we’re going to fix this at an ecosystem level, how can that be done? I know that somebody – there was some discussion on LinkedIn – and somebody I believe formerly from AdColony was saying: hey, we need a count of how many times an ad unit has kicked off SK Overlay or an App Store View. And that will tell us if it kicks it off 10 times in a session, let’s say, it’s a little suspicious.
When we see click-through rates that are really high, it’s a little suspicious. So we need some extra functionality in SKAdNetwork to tell us more about ad quality and people who are basically manipulating that.
Is that a solution? Is there another solution?
Yeah, it’s a great question. I would say even before going there, I would say it has to be fixed. It has to be fixed because it is becoming increasingly apparent that the costs outweigh the benefits in the long run.
For publishers, there is exactly the point you’re making, users being churned away.
I would say also for advertisers, there’s a hidden cost which cannot be … ignored much longer, right? The advertisers who essentially end up rewarding the media sources that are most aggressive at this game are probably misallocating their budget a little bit, and they’re probably paying for users who will come in anyway.
And so what is the fix?
I think that there is a lot of talk in the industry around complementing bottom-up attribution. what the industry has been doing up until now with more top-down measurement approaches. So, incrementality measurement, marketing mix modeling, more coarse and long-term aggregated measures of performance for different media channels have to become an element of sophisticated marketers’ toolkits.
And then, I think that there are some short-term … things that advertisers can do or marketers can do right away. Some perhaps medium term levers that the white knights in the ecosystem can experiment with. And then there’s perhaps some longer term things that I would envision and hope the ecosystem will consider.
I think in the short term, the first thing is try to educate yourself and your management team on what is happening in the industry. It’s a little bit tricky, right? Because from a marketer, media buyer perspective, you have to go look at what are perhaps some less incremental dollars that you’ve spent with some channels. You need to have the conversation … if I think about the marketing organizations that these decisions are made in, you have to basically go to your boss and say: hey, look, up until now, in the last couple of years, we may have not spent our money as efficiently as possible. And we did a bit of an audit of our marketing channels. And this is what we’ve seen in terms of the tactics they use for attribution.
So there is a little bit of that awareness that needs to come up on the marketing side. Auditing your media sources, understanding how they’re doing attribution and what kind of click firing behaviors they’re driving is the first short-term thing that I think must happen.
Second, there is an opportunity to get a little bit smarter on testing, right? Whether it’s incrementality and lift testing, or whether it’s about considering spend consolidation for all the open internet programmatic inventory. If you think about it, inventory access used to be a huge competitive advantage back in the day.
Now the vast majority of the in-app inventory is open to real-time bidding and fairly accessible to all the demand players or SDK networks that you’re working with. And you know if you are to look at one player only, right, for a one app in one market, and compare to what that performance looks like relative to your current media mix. And what do those two test cells look like relative to maybe another test cell where you’re not spending at all, right? What kind of incrementality do you see when you are eliminating these competitions and when you’re eliminating these overexposure?
Of course, there are trade-offs there, right? Because you want to keep your media mix diversified.
But even just to see what different media partners are doing and measure the relative incrementality, spend consolidation tests are one lever that we’re seeing some of our biggest partners consider. And then look, the question is, these are patches. This helps illuminate and size the problem, but then how does the ecosystem clean up going forward?
One possible outcome is that it doesn’t. And so the savvy marketers are going to essentially rely more on top-down measurement, incrementality, marketing, mixed modeling, lift tests, and really reduce how they look at bottom-up measurement, and just use that for very tactical and fairly minor media management decisions.
And then another, you know, I’ve been toying around with this idea for quite some time. I’m wondering if the industry needs to look at more granular click reporting. So imagine if advertisers demanded ad tech companies provided and the mobile measurement partners enabled more visibility on the type of ad interaction that any given user had.
Now all of a sudden a click is not a click anymore. You can have a click that was explicitly on a call to action button. You can have a click that was perhaps an engagement with the playable component of the ad. Or you can have a click that perhaps is in the area where usually the X button is or an unqualified click that’s happening somewhere else that we cannot quite tell if it was indicative of installing intent or not.
If advertisers had visibility on those types of click interactions, then they could attribute different value to those different click interactions. Just like now, for a click versus a view, you have a hierarchy of attribution. You have a different attribution window. You could envision a similar hierarchy of attribution for different click types. And if you do that, all of the sudden you have media budgets from advertisers who will go to media sources who can drive better interactions.
The experimentation from a buyer perspective, from a marketer perspective would, you know, right now it’s hard to do a lot of experimentation given that it’s hard to do deterministic user level attribution, but all of that analytical horsepower could go to experimenting on different click interactions.
I keep thinking of creative testing, creative experimentation.
I was at Google in 2017 when there was this big shift from manual campaigns to universal app campaigns, right? The first big machine learning driven automated product in app promo. And you know, my team and I built this entire capability around creative advisory, because creative was one of the big levers, right? And that’s been the name of the game, right? For quite some time.
And then ATT and SKAN come along and all of a sudden creative testing is so much harder. And it’s made harder by the fact that when you’re looking at click-through rates, you can’t quite trust them because a click can be recorded for all sorts of different reasons that are not truly reflective of the real quality of the creative assets itself.
But in this hypothetical new world that I am envisioning where you have different click types now you can start experimenting with different creatives based on the type of click they drive, the type of user engagement they drive.
I think if you start with that, then the rest of the ecosystem would need to adjust, right? Because then the media would follow the better interaction and the incentives to have these less valuable ones would die out a little bit.
Meta comment for a second. It’s such a pleasure to chat with somebody who is an expert, been in the space for a long time, understands it deeply, and can correlate a lot of these different things together. Super, super excellent.
So much in what you just said. I’m gonna hit on a couple things and then maybe introduce our last topic.
I had a conversation just recently and just published it actually on Growth Masterminds with Adam Jaffe. He’s CEO of MegaGames, he’s been at Jam City, he’s been at multiple game studios and VP of Growth and UA roles and stuff like that and does some cool stuff.
Now we’re talking about this problem.
He says, you know, as an ad network, I could be honest and come to you and say, you know what? CPIs, we report them, you know, $2, $3, $4. It’s actually $35.
There’s a lot of clicks that you get there and a lot of users that come in that don’t do much, everything like that. But if I report to you $35, you’re going to say, are you kidding me? Forget that. Not happening ever. My VP finance, well, all this stuff.
But he says, so that’s a challenging conversation. So that’s interesting as well.
We can talk about quality of click and what is a quality click? A click on a call to action. That would be very interesting compared to a random miss the X button click. And so I hope that something like that comes out. We’ll see. That would be interesting.
I also liked your comment about spend consolidation in the era of programmatic. Of course you wanna try a lot of different things and you always need to be testing. I think Zynga is always testing two ad networks. Just as a matter of course, they’re always testing different, different places, seeing what’s going on.
But when you think about programmatic and you think, hey, there’s the river of users. And like you said, most of them are accessible to most of the major programmatic players. How many fishing poles do you need in that river, right? Do you need fishing poles from 15 different companies in the same river? You might just be clouding up your marketing measurement. Who knows?
And that’s kind of where I want to end actually, marketing measurement. We’ve touched on it in a couple of instances.
As I look at marketing measurement of the future, we talked about the phrase everything, everywhere, all at once. And I mentioned the phrase that we used at Singular, which is Hybrid Measurement. You want to get all the data that you possibly can. And you want to use certain bits of data in certain scenarios for certain things.
So if you can get your click-through data, great. But grains of salt, all that stuff. But you can get your top-level, impression-level data. You want to get that as well. You want to understand that, what’s going on, what’s happening there.
You want your first-party data. You know what’s happening. That’s the most reputable data.
You also do want your SKAN data eventually, your Privacy Sandbox data. You know it’s last click, you know it’s limited, you know there’s a bigger, wider user story there, but you know as well it’s deterministic even if it is aggregated, so that’s got some value.
And you also want this big probabilistic picture with triple M or incrementality studies where you can say, hey, if I throw an extra $100,000 a month there, what changes? If I take that $50K a month away, what changes? And you need to understand that. You need to understand whether that’s monthly or whether that’s quarterly, or whether you can do that even with some magic pixie dust, fairy dust, a little more frequently, great.
So I think that’s kind of the future of measurement, and it does entail, and I think we talked about this as well, a little bit of the abandonment of the illusion of certainty.
We’ve never had certainty. Just because somebody clicked on something on the web in 2001, that’s not the entirety of the user journey. And everybody who understands marketing or people knows that. There’s so much more going on. And the same is true for high-value users, players, customers, and mobile apps. So I think that there’s a mind shift. And you mentioned that. Eric Seufert’s talking about that as well.
Talk about measurement as you see it.
Yeah, look, I think you hit on a number of points that I wholeheartedly agree with.
You mentioned Zynga. We work with a lot of very sophisticated partners who are well on their way on this journey and are combining first-party data, top-down measurement, and bottom-up attribution metrics to come up with as accurate … a vision as possible of what is actually happening to their media dollars and where the next marginal dollar should be deployed.
So I hold hope that this industry has a lot of really impressive minds that are thinking about these and they are thinking about what to do to operate in this new environment and how to win in the new environment. Because again, while the short-term incentives are stacked against a clean up of the ecosystem, it is also true that when you have really smart people, there will be someone who will say: Hey, you know what, I’m going to sacrifice some short-term metrics. I will actually bring my CFO along. I will explain that we have to rethink. How do we look at our marketing budget? We have to rethink how we look at our performance metrics? We have to reconsider what is truly coming from paid versus what is organic.
And then, you know, all our first party data, our understanding of the user value curves is the other side of the equation that we have to match with our marketing investment.
And, you know, those are the advertisers, marketers, mobile companies that are going to win in the long run, because guess what? They’re going to make a better use of their budget and they’re going to do a better job at getting the most bang from the buck. And look, one thing that I see those savvy marketers doing and one thing that, investing heavily is, you know, trying to get transparency and control from your media partners. While I agree with you that we’re probably past the illusion of certainty that characterized the deterministic attribution era, it is also true that marketers with big pockets should demand from their channels the ability to control.
Right? You know, when are you rendering a StoreKit element? And what kind of attribution behavior are you driving? You may or may not be able to control it fully, depending on the kind of partner you’re working on. But you should be aware. You should get a report. You should get some transparency.
And then that becomes part of the picture that you’re painting with this 360 degree measurement approach that you were describing, which I fully agree that that has to be the future.
Excellent. Well, all the good things must end and so must this conversation. It’s been a real pleasure. Thank you so much for your time.
Thank you so much for having me, John. It was a lot of fun.