$APP for CTV and Web? I'm skeptical

[intjudo]

$APP is now 17% of my portfolio so I’d like to examine their claim that they will be able to move into Web and CTV. The CEO makes it sound like some sort of slam-dunk, but is that really so?

There are a LOT of differences in the objectives, delivery and data of a campaign done on mobile vs. Web vs. CTV. $APP’s AI has been trained on mobile. Here are some concerns of mine:

#1: Availability of AI Training Data
Will it really be that easy to get training data for Web and CTV? Recall that $APP has its own stable of mobile apps that it can use as a source of training data and to use as a testing ground to tweak the AI algorithms. It has no such dataset for Web or CTV.

#2: Does collected data differ for Mobile vs. Web vs. CTV?
I think it’s unlikely that data collected and available for AI analysis is identical for Mobile vs. Web vs. CTV. For one thing, the type of content produced and delivered varies a lot (low budget vs. high-production-value), the placement of the ad (the place on the screen and what events drive delivery of the ad: banner-placement vs. popup etc.) etc etc

#3: Harder to know who EXACTLY the audience is for non-mobile devices
The great thing (mathematically speaking in terms of number of AI training variables to deal with) about an ad on an app on a cell phone is that you know EXACTLY who is getting the ad and EXACTLY what they are doing at that EXACT moment. The same is certainly not true for a CTV ad, where any number of people may or may not be in the room, may or may not be even watching the TV and may or may not be all that engaged even if they are watching

#4: Is $APP really equipped to handle different types of ad campaigns?
Let’s imagine Geico wants to place an animation of the Geico lizard in a CTV ad that appears in the bottom-right part of the screen during a football game. Their main target is the adults in the room, who they hope will switch over to Geico. But they don’t know who those adults are: for a particular CTV, it could be the people that live there, and/or their friends who came over to watch the game and/or their family.

Maybe Geico considers this a mixed campaign where they want the adults to buy insurance now, and they want the kids to be indoctrinated into the notion of Geico as a trustworthy brand to consider later in their life. Is $APP able to measure the effectiveness of this kind of mixed campaign as effectively as $TTD?

We’ve seen examples before where a CEO waxing euphoric over a company’s prospects caused a runup that didn’t turn out to be supported by the actual story, so just trying to think ahead a bit.

Your thoughts please?

[stocknovice]

I wouldn’t say I’m skeptical as much as their forays into areas other than mobile are unproven. I was encouraged by the TTD CEO’s comments about advertising tailwinds, but APP can only tap into those as quickly as it can ramp in other adjacencies.

I was impressed by APP’s story and quarter but figured there was no harm in waiting for TTD’s print before doing anything. The stock’s run from $60B to $100B in a blink since. I’ve never been afraid of high valuations but have to question that large of a surge. From that perspective, RDDT might be more interesting as an ad play at $25B.

Not saying APP can’t do it, but some of it will be dependent on how many existing customers they can cross or upsell versus new customers in new verticals which will need to be won for the first time. Anyone have any insight into how multi- or omni-channel APP’s existing customer base might be?

[IamM]

Good thoughts @intjudo.

I think it will take some time for them to prove us right or wrong. Getting into CTV and Web with promising performance is not going to happen overnight, but their AXON tech could help them to achieve better result in that endeavor. Of course they have to fine-tune it, change some data/levers but their Machine Learning engineers have experience and should not be that hard to adjust AXON.

As you mentioned the data that AXON will work with is extremely important and they have to figure out how to gather, combine and take out the best of it. After all, they might come with some brilliant idea like TTD’s UID2, which will help them get great results. Of course they might fail, but we can’t know that.

APP is 11.7% for me.

[Buffjan2]

Thanks you intjudo for starting this great thread. With a high flyer like APP it is so important to understand the company and what can go wrong.

On the CTV front, APP purchased a company named Wurl in April of 2022 in order to get access to the CTV market. While the e-commerce advertising seems to be organic the CTV vertical was not.

https://investors.applovin.com/news/news-details/2022/AppLovin-Completes-Acquisition-of-Wurl-to-Extend-Reach-into-Connected-TV-Market/default.aspx

[intjudo]

@Buffjan2 thanks for the reminder; when I posted this I neglected to mention Wurl and I’ve been remiss in not doing any real research on Wurl.

Has anyone dived into Wurl and if so please share if you have thoughts about how useful Wurl’s data may (or may not) be for training AI to add CTV to their auction platform.

[wpr101]

I am not too familiar with how the web pilot they have going is actually implemented in terms of training, and I do not think the company has revealed that. What AppLovin has mentioned on the previous quarter was that the web pilot was going extraordinarily well. They said this quarter the following,

If you check even Twitter today, there’s tons of noise from e-commerce brands saying, in our pilot, they’re seeing as much scale and a strong ROAS as they’re seeing anywhere in the world today on their user acquisition buys.

So when you have that kind of performance, you’re delivering an automated return on ad spend model approach to a very large fragmented category, like e-commerce, that desperately needs another marketing channel after having so many promises in the past that haven’t panned out. It’s something that’s probably going to get a lot of noise and be very attractive to the other side very quickly.

The one thing I’ll say in terms of partners, we haven’t disclosed any partners, but I referenced this a couple of seconds ago. If you search Twitter, you’re going to start seeing more and more noise because the customers are seeing a lot of success, and we’re not going to be able to stop social media from talking about the successes that they’re having on our platform. So that’s the way you can really decipher which customers, how are they performing and then what else is out there.

Early data has exceeded our expectations, with the advertisers in the pilot seeing substantial returns, often surpassing those from other media channels and, in many cases, experiencing nearly 100% incrementality from our traffic.

So e-commerce is still in pilot, as we touched on last quarter. I’ll say, before jumping into impact this quarter, it’s the – it’s a super compelling product. Our team has done an amazing job building it. In all my years, it’s the best product I’ve ever seen released by us, fastest growing, but it’s still in pilot.

10 years from now, we think every advertiser that has a transactional model, whether it’s collecting an e-mail address for a newsletter or a local pizza shop or an e-commerce brand or a gambling brand or a gaming brand or anything across any category, can buy on our platform and do it at scale. So there’s no limitation to the power of the math and the technology that we’ve written.

But the only limitation, the only reason why we referenced it as e-commerce today is the go-to-market. We do want to be thoughtful about how we penetrate new categories. But even inside e-commerce, we’re not approaching it as we’re only going to go to fashion and then we’re going to go to beauty. We’re approaching it as we’re going to go to the entirety of e-commerce, and then we’re going to go to the entirety of the rest of the categories of advertising that matter. And then hopefully, we’re going to go to the whole world of transactional businesses over time.

---

AppLovin clarified multiple times the long term goal of 20-30% is for gaming only, and almost all the results this quarter are from gaming alone.

The step-up this quarter was entirely on the gaming side, which, obviously, is the vast majority of our advertising business today. We’ve talked for multiple quarters now about how there’s a long runway in gaming.

Like we’ve said, we’re confident in the 20% to 30% in gaming alone. This is sort of additive. I also just said it’s the fastest-growing product I’ve ever seen. So it looks really good. Now if we were scaled in e-commerce, we’d have seasonality in the business.

We don’t today really have seasonality in the business. When you’re in pilot, you’re talking about tens to hundreds total shops. And so we’re really just early stages here, but the performance is really strong.

---

With regards to AppLovin having trained on their own game data before, they used to get asked that question by analysts often. They always replied that having their own games provides no additional advantage for them in terms of training, it works the same regardless.

I am viewing the web pilot more as upside for the company than anything else. The free to play gaming category is still growing and this quarter provided 635M GAAP EBITDA and 434M GAAP net income this quarter, which is a magnitude greater than any other software company of their size.

I do not really see how we come away reading that last earnings transcript leaving more skeptical. If anything it sounds like there is tremendous buzz around the product already which users are saying is working better than anything on the market for web advertising.

[intjudo]

@wpr101
Yeah, I think you’re right.
I’m just getting too wonky on speculation about the tech, like I do

That said, at a technical level, it sounds to me like expanding to Web is going to be a lot easier vs. CTV.

And we have a way to verify his claims about Web (…search X for rave reviews about it) vs. all we have for CTV and beyond are the CEO’s assertions.

If the scope of what he is saying is be be believed, $APP will soon go from a “uni-channel” to a “multi-channel” Ad-Tech company and will eventually be an omnichannel “Marketing-Tech” company competing directly against current heavyweights. Sounds like 5-10 years of revenue expansion to me

[wpr101]

@intjudo

That said, at a technical level, it sounds to me like expanding to Web is going to be a lot easier vs. CTV.

And we have a way to verify his claims about Web (…search X for rave reviews about it) vs. all we have for CTV and beyond are the CEO’s assertions.

That is a good callout as it does seem like the CTV side of expansion has been put on hold while prioritizing web. Two quarters are ago they were talking more about CTV and web together, and the last two quarters the discussion and pilot just mention web. Here’s what was said two quarters ago (May 8) by the CEO on the earnings call,

So Wurl, to us, it just added eyeballs. It’s added supply. CTV channel versus the mobile app ecosystem, we touched on. We have over 1 billion daily actives in the mobile app ecosystem. We’ve got a lot of access to eyeballs, but 5 hours a day of TV watching is inaccessible if we don’t get to an SSP that sits in connected TV. So that was the idea with Wurl is bringing a lot of supply online. Now it’s our job to go monetize it.

To create growth, the systems have to improve, which we’ve shown can drive a lot of growth, and we have to go access more eyeballs. And that’s the goal around Wurl. We do think in connected TV advertising, as we get into e-commerce and prove an efficient model for shops to advertise on our platform that will extend very naturally to the e-commerce, to the CTV landscape because the shopping ad could be very beneficial for consumers in that media.

---

That seems to imply the e-commerce pilot works well with CTV shopping, because they will already be in e-commerce and maybe have some advantage there, connecting e-commerce shops to CTV buyers.

It seems normal to me that AppLovin would first try to get into either web or CTV one at a time rather than do both at once. If they get a big foothold in web e-commerce it is already going to be a game changer for this business, and I think that is what the market started to realize in between these last two earnings reports.

---

At least one of the personalities on Twitter who is sharing details on the pilot is Zach Stuck, his description is, “Founder of Homestead, a growth marketing agency for 7-9 figure eCommerce stores. Currently building Easy Street Brands.”

His concluding statement on the web pilot, “If you are spending $500k+ monthly on Meta ads you should highly consider running ads on this platform”

---

It is interesting to me that spending above 500k on the platform would be even more beneficial than below that level. It seems to indicate performance or ROAS gets better as a customer scales up.

The ROAS numbers shown below I believe are very strong. My understanding is that return on ad spend, of 2.33x means that for every $1 spent on AppLovin’s platform they get back $2.33 worth of sales themselves. Maybe someone on the board knows how this metric works a bit better though.

[wpr101]

Had a few more follow up thoughts on the tweets shared above,

The conclusion about META ads implies that META may be AppLovin’s biggest competitor in the web e-commerce ad space. It sounds like these ROAS numbers on AppLovin’s system is outperforming META. The ad market for META is enormous, and probably why AppLovin’s CEO is so excited about this new product.

Here’s what the CEO said two quarters ago about their customers spending,

One, again, I don’t want to build out a sales force. And if you’re selling someone to buy a brand-advertised spent dollar, you have to really convince them that, that dollar is well spent. There’s no data that backs it up. The attribution is murky at best. And so you have to have a salesperson that convinces the other side, that $1 spent was well spent.

Our model is the advertisers spend $1 and everything is measurable. It’s all closed loop. It’s real-time reported. And they know if they spend $1 and they made more than $1, they’re buying as an arbitrage marketer, and there’s not a whole lot of selling to do in the middle there.

If you have someone spend $1 and earn $2, they will spend that $1 as many times a day as you will spend it on their behalf. And so all of our algorithms, our entire system is predicated on that concept. And what’s powerful about that concept is when we can create lifts in our business, as you’ve seen over the last year, and I touched on the incredible flow-through of this business that we’ve consistently set as a theme in our business model, we don’t have to go convince advertisers to spend more. They will automatically spend more. So our constraint is just how many dollars can our systems accurately, on their behalf, place in the universe.

---

Taking a look at this company Hollow mentioned in the tweets, they spend $12,000 a day with the web pilot. This roughly translates into about $24,000 of revenue for Hollow, or a 2x ROAS.

AppLovin discussed the “incredible flow through” of their algorithm. We could estimate based on current numbers of that $12,000 spend, AppLovin gets about $9,000 of GAAP EBITDA, and about $6,000 of GAAP net income per day.

Since the company Hollow is getting an immediate return, they want to spend as much as possible, and will likely spend 365 days of the year if possible. That would mean this single company is generating $6,000 * 365 = 2.19M of net income for AppLovin in a year at the current run rate.

[100invested]

Thanks everyone for a great discussion so far. Here are my thoughts after doing a deep dive on APP in the last couple of days. While I am excited about the company, here are my thoughts and reservations:

1. Revenue growth has been predicated on research and improvements in algorithms, much like UPST. At some point, just like how TSLA is seeing smaller improvements in their FSD due to fewer interventions per mile driven, the improvement generated by AI will plateau. Axon 2.0 launched Q2 2023, which rapidly increased revenue growth, but this is now tapering off.

• Q2 2024: -3%
• Q3 2023: 21%
• Q4 2023: 36%
• Q1 2024: 48%
• Q2 2024: 44%
• Q3 2024: 39%
• Q4 2024: if 5% beat over guidance, 39%

They are still showing 10-11% sequential growth which annualizes to 46-52%, so to me, this is does not portend a fast revenue slowdown.

2. Platform is still limited to mobile gaming apps, which according Clark Lampen (BTIG analysis, Q2 2024 earnings call) is 20-30B/year. APP revenue is at 1B, so they have 5% of their TAM.

3. Mobile gaming apps which is growing at 7%. They mentioned non-gaming apps, but I can’t find any reference to examples of non-gaming apps that have ads. When I think about the apps I use - yelp, google maps, and banking apps, I do not see opportunities for ads. If anyone sees this differently please let me know.

4. Transition to CTV is…ambiguous. Wurl only has BBC, Bloomberg, A&E Networks, and Roku (in comparison, TTD has Netflix, DIsney, Fox, MLB). I don’t think they can replicate their exact model of immediate ROAS like mobile gaming because people don’t want to leave the show to buy something. I think it will be harder for them to demonstrate ROAS on the connected TV platform.

5. Foray in to e-commerce might cannibalize current gaming ads. The platform is the same. There can ony be so many ads per game before the eyeballs get tired. Perhaps they can charge more for e-commerce ads, but I don’t see how this is additive.

Therefore, I still own TTD because connected TV is a much larger opportunity. I am cautiously optimistic about APP and am still deciding if I should start a small position.

I just started looking into APP and with non-advertising background, may not have all the facts right. Please feel free to correct me if I’m wrong!

[wpr101]

Here is some of the incredible feedback that continues getting posted on Twitter on AppLovin’s e-commerce pilot. This is cherry picking a few posts, but there are lots and lots of posts like this,

• massive demand to get into Beta
• the FOMO about AppLovin is gonna skyrocket after this weekend… wait for the screenshots (posted December 1)
• we’ve been on AppLovin for a week and will cross six figures of spend today, every day we nearly double the budget and the return holds
• wouldn’t doubt we spend more then our Meta budget by the end of Q4
• real biz operators get it - much of wall st is still asleep on the name APP… not much longer
• Yes its better results than Meta so far