ABOUT EVA: Linkedin | Pocket Worlds |

ABOUT ROCKETSHIP HQ: Website | LinkedIn  | Twitter | YouTube


KEY HIGHLIGHTS

🧗‍♂️ Factors that contributed to scaling TikTok.

⚖️ The testing process: structure, geos & metrics.

🦋 Working with ad sets that included both old and new creatives.

⏳ Accounting for data delays on iOS.

🌔 Dealing with saturation on TikTok.

🧵 Why IPMs are tracked.

💼 Working with actual users for UGCs instead of actors/influencers

FULL TRANSCRIPT BELOW

Shamanth:

I’m very excited to welcome Eva Juretić to the Mobile User Acquisition show. Eva, welcome to the show.

Eva: 

Thank you. It’s great to be here.

Shamanth: 

I’ve been very impressed with how sharp you are and how on top of things you are. I’m excited to talk to you because you’re among the very few folks I know that have scaled TikTok massively and very significantly. I would love to dive into more details. 

What are some of the factors that contributed to scaling TikTok massively, and what’s the kind of scale you guys have attained?

Eva: 

We’re talking about scaling the game Highrise, which is a virtual world like Metaverse, where people can express themselves, build this world, own their avatars and experiences for them and their friends. This game has a community and everything revolves around the community. You can join the crew or meet new people, participate in events, and build anything you want.

So Highrise as an app and game is trying to be a platform that provides these tools for users to build a world. We tested a bunch of media and social media shown to work the best for Highrise because of the community part and the social part of the app.

So for TikTok to work for us, we conducted a lot of creative testing by using different approaches, concepts, and variations of creatives that we had success with on other platforms. But as TikTok does require a bit of a different approach, we began creating content that would be fun, informative or valuable for users. 

We started posting on our TikTok page and account instead of using video ads, which we tested a lot. Once we created that, we used it on Spark Ads. User-generated content or UGC-type of ads really helped us scale on TikTok, especially on iOS, which had been a struggle earlier.

Shamanth: 

It sounds like you guys leaned into UGCs using Spark ads, using content that your audience was already making and a lot of these contributed. 

Can you help me understand how you guys test creatives? What does the testing structure look like? What geos do you test in? What metrics do you optimize for? How do you account for iOS data delays and iOS data issues?

Eva: 

We started with tier 1 geos, mainly the US as it’s our biggest and most profitable market. With TikTok scaling and being this big social media in the US, we thought to start there. This structure changed and adapted over time. 

On TikTok, you have campaigns, ad sets and then ads. Ad sets are limited to 20 ads and we had a lot to test in the beginning, with frequent ad rotations.
Our main key was what works – what resonates with people on TikTok and our audience. So for that reason, we had in the beginning our business as usual ad set and then a separate ad set with creators to test and gather data.
When we saw something performing well, we would add it to our business as usual ad set. As we figured out what worked over time, we were left with one key ad set, which would have our top performers competing with the new creatives and that showed to work well for us. 

This was maybe risky to test in the beginning. All the suggestions and advice coming from our managers and just in general on advertising on TikTok were to do it separately. We’ve tested both and saw this work well and simplify things overall as well.

One thing to note, it’s really important to take into consideration the rules and guidelines of TikTok which can vary by region or country. So it’s important to follow them and work with the customer success team to avoid ad flags or account suspension. We did have more than a few of them.

These included promoting anything related to guns, cigarettes etc. We had one situation for Halloween when we were filming ghosts and it was flagged for promoting smoking because it showed smoke. So there were false flags like this. We really take those into account when creating ads.

Shamanth Rao: 

I know folks who had trouble because of alcohol-related rules. 

I would love to just follow up on some of the mechanics of testing that you described. You said you eventually moved to a setup where your business-as-usual assets have new creatives plus older ones. Why did you guys move to that setup?

I also ask, because that sort of setup just doesn’t work on Meta because Meta privileges older ads so much. I know TikTok has a very similar setup, although I don’t think favoring older ads is so extreme on TikTok. 

So what is the thinking behind putting new ads along with the older ads in your business-as-usual ad set?

Eva: 

That was my starting point as well, being familiar with Meta and how we conduct A/B testing or just testing creatives in general. We just gave it a go and our rule was, if we have enough data to make conclusions, we’ll just continue doing that. So if we have a minimum of 5k impressions and a couple of installs, we have some sort of data.

Initially, we set higher benchmarks, but through continuous testing, we would leave creatives to gather more data. But nothing really changed. 

If it wasn’t performing, if the click-through rate was low in the beginning, if you add more impressions, it won’t improve.

So sometimes TikTok has to give it a check because we would see spend on ads that aren’t necessarily top performers. These ads can also be new ads so it wasn’t a matter of ignoring them just because they were recently created. We don’t see this happening on TikTok.

Shamanth: 

So you’re looking at even early signs with 5000 impressions, what is the click-through rate and if it’s bad, you shut it off; if it’s good, it sounds like you are able to compare older and newer odds side by side using that setup.

I know you mentioned iOS, there is still the SKAN data delay, if you add a new ad, that takes like three, or four days to get enough data. 

So is there anything you do to account for those delays? 

Eva: 

What we do now is we primarily look at Android data separately, draw some conclusions, and then go ahead into iOS campaigns individually and try to pick up the trends. If something is working on Android, it’s working on iOS as well. 

We focus on creative metrics. We do look at conversion rates, and IPM over time, considering if something has been running for a while IPM can go down. Creativity tends to saturate much faster on TikTok. We do look at IPM but also day one retention and then at the end of the day CPI, because our goal is to reduce costs of install.

Shamanth: 

You also said that saturation is much faster on TikTok. How do you look at this saturation? 

How do you evaluate that it is actually happening and how does it compare to Meta or other channels in terms of saturation?

Eva: 

IPM is installs per 1000 impressions. We are tracking that and how it changes or behaves over time. We graph it out for creatives. We can see how the new ads are compared to already existing ones. We can see over time, how that is looking separately per ad or altogether.

We do these quick rotations weekly. After every optimization, there is an increase in performance, both IPMs and lower CPI, it’s very consistent. So we keep on doing that.

Shamanth: 

So if I understand correctly, you see an ad that was spending a lot and used to perform very well, starts to decline its IPM and you pause that so the other ads will start to spend, therefore your IPM goes up. 

Did I understand the process correctly?

Eva: 

We do not pause it unless we have a creative that’s outperforming it.

Shamanth: 

So there’s an ad that’s getting a lot of spend, but performance is bad. Then you have the best creative, performance is great, but it’s actually not getting spend. 

So you killed the higher spending one.

Eva: 

We actually have one ad that’s been running, in two weeks, for almost a year and has consistently been a top performer or among the top three. But on TikTok, I haven’t experienced something that I find on Facebook or Meta.

So in phase one, if you have a top performer, it’s very hard to overthrow a top performer. TikTok, I don’t find that to be the case as much. We do have one ad that has performed really well for a long time. But still, some of the new ads come to first place as performers, but they don’t necessarily stay there as long. So there’s a rotation.

Shamanth: 

Certainly. I’ve certainly seen that as well. There’s a lot more flux within TikTok. On Facebook oftentimes we see the ad that wins is the ad that everybody hates.

Eva: 

On Facebook, we have this one ad that is difficult to get rid of, just because it’s not very related to gameplay necessarily. We have better ones that actually show gameplay. We want to overthrow that one and we are working hard.

Shamanth: 

You did mention a big part of the success has been with creative and very specifically that you guys use actual users and footage, not like UGC, from actors or influencers. Can you tell us about the creative strategy?

Eva: 

After testing a bunch of stuff, we clearly realized that UGC-type of ads work well for us, even though we were the ones creating that. These ads involved recording your phone with another phone or device where you’re clicking, creating an avatar, and showcasing gameplay. We thought as UGC works, we will let people actually create that TikTok content. So basically, we tested with a couple of micro-influencers.

But that was overall a fail. We think it’s because those influencers weren’t actual users of Highrise. So unlike puzzle games or any game with levels, there’s progression, that might make sense, and work with influencers. 

But Highrise, at its core, is a community. So a Highrise user will know these internal little jokes. They can’t really fake it.
So our next idea based on our failure with influencers was to reach out to our users and see if there are people that create TikToks on their own. We had an in-game contest/challenge. Users reached out to us, they would post-TikTok videos. We tested and worked with a couple, some went well, some not so much. But, one user whose TikTok performed really well contacted us, because they saw a win-win.
So we worked with them, they sent us an email and came up with a proposal. They were getting high reach and impressions for their content and their profile that they wouldn’t otherwise get. So that worked really well. We started finding more people that we can work with like this. For us, native UGC content worked really well for bringing in new users.

Shamanth: 

Yeah and like you said, because it is a community, the fact that they are actual users makes a big difference. And because you’re putting money behind this on Spark Ads, they get new views and new followers. That is a big incentive for any creator. They want to be influencers if they aren’t already.

Out of curiosity, you’re getting this footage from creators, is this typically in a shape that is ready to use as an ad? Or does this require edits from your designers? I ask because my team works with creators and oftentimes we do go through some rounds of edits, because I think the designers can make a big difference. 

Eva: 

We do not edit our user’s content at all. We want them to do whatever they’re already doing.  But we do provide feedback, mainly technical suggestions such as using commercials and music that can be used as Spark Ads. We also provide general guidelines like avoiding content related to guns, cigarettes, or smoke and telling them to record on their phone and show and explain what they are doing. 

Maybe if we tested something in-house, creative-wise, we would just tell them to show their face. But it’s up to them as we want their content to remain theirs.
So there are no edits on our end. 

Primarily why we think this works is because they are actual Highrise users, they know all the good things and the struggles. The content they do is very relatable and it’s not fake. They’re incentivized to improve it because if they run it for a longer time, they will get more impressions. So if they ask for feedback, we tell them what they may do.

Shamanth: 

That feedback oftentimes can point them to improve the quality of their output. Is there any quantitative feedback you guys give? Or is it more like best practices?

Eva: 

It’s more of best practices. On their end, they see the impressions and engagement they get, the comments and the likes on their videos. So they also have a little community there and they talk to the TikTok users. For us, if it’s well-performing, the video keeps running. So they see more impressions, more engagement, and more comments. So it’s more of general feedback. 

We do share some numbers, but they’re just not really interested because they don’t necessarily mean anything to them. They’re more interested in getting a bunch of comments and likes on TikTok as it motivates them.

Shamanth: 

And that’s the kind of quantitative feedback which matters to them. Excellent. We are coming up on time, but this has been incredibly fascinating. 

This is perhaps a good place for us to wrap. But before we do that, can you tell folks how they can find out more about you and everything you do?

Eva: 

So I’m working with Pockets Worlds, which is a developer of Highrise. I personally have a little consultancy myself. So I think a good place is LinkedIn, or email me on social@pocketworlds.com. 

If people want to reach out to me personally, I think LinkedIn is a good one or eva@clickbite.hr

Shamanth: 

Excellent. We’ll link to all of that in the show notes. But for now, Eva, thank you so much for being a guest.

Eva Juretić: 

Thank you for having me. This was really fun.

A REQUEST BEFORE YOU GO

I have a very important favor to ask, which as those of you who know me know I don’t do often. If you get any pleasure or inspiration from this episode, could you PLEASE leave a review on your favorite podcasting platform – be it iTunes, Overcast, Spotify or wherever you get your podcast fix. This podcast is very much a labor of love – and each episode takes many many hours to put together. When you write a review, it will not only be a great deal of encouragement to us, but it will also support getting the word out about the Mobile User Acquisition Show.

Constructive criticism and suggestions for improvement are welcome, whether on podcasting platforms – or by email to shamanth at rocketshiphq.com. We read all reviews & I want to make this podcast better.

Thank you – and I look forward to seeing you with the next episode!