Meta Ads in 2025: Expert tips from Marcus Burke

Scaling Meta ads is a bit like playing a video game: every time you think you get it and understand the level, it gets harder, and it feels like the rules change. There may be hacks and shortcuts that some people swear by, but the true gamers know they won’t help you win in the end. 

Not only that, but everyone has a very strong different opinion on succeeding in the game. Take how my partner and I each play Hogwarts Legacy. He waits for the dark wizard or mythical creature to attack him, uses Protego (a shield), and then attacks back. Meanwhile, I shoot off endless spells combined with an occasional dodge—take a guess which of us has been diagnosed with ADHD.

We’ve both completed the game, and we both stubbornly think our way is better. Meta is the same: people have completely different ways of structuring and approaching both campaigns and the creative. They are adamant that their way is the right way despite them being completely different.

Combining this with the changing rules makes Meta a tricky game to understand. In the 10+ years I’ve been running Meta ads, I’ve learned to be selective in which Meta gamers I listen to and to always take an open, willing-to-experiment approach to ads. Before we move on from this example, let it be known that I tried my partner’s shield-first approach in Hogwarts Legacy, and yes, it worked better in some cases. 

The same goes for the advice you’re about to hear. All I ask is that you listen, experiment, and see what works for you. There is no one-size-fits-all approach to Meta ads, but I want to provide you with the updated rules of play for the game of Meta to understand better what is and isn’t changing on the platform. I’ve tailored this advice to be specifically for subscription apps. 

To provide this, I’ve interviewed someone who I know isn’t the type to try to hack the game but is considerate of how different players prefer to play and focusing on an overall better understanding of the game of Meta. 

Drum roll… I’m speaking to Marcus Burke! As an independent Meta Ads consultant, Marcus helps subscription apps unlock Meta as their primary growth channel. He has 13 years of experience scaling Paid Social for apps and a background in both gaming and consumer tech (Forge of Empires, Blinkist, Tandem…). You’ll also find him on LinkedIn sharing practical advice on Meta Ads, web-to-app, optimizing paywalls, and improving user onboarding. 

I’ve been following Marcus for the last few years, and his content has always stood out in this field. When it came to this subject, I knew he’d have excellent input on the state of Meta ads in 2025 and what subscription ads need to know.

We discussed handling creatives, dealing with less data, and much more. Here’s Marcus’ expert advice on scaling Meta for your brand in 2025.

What are the biggest changes you’ve seen in Meta ads for subscription apps in 2025?

Let’s immediately dive into the big question on everyone’s mind. What’s changing with Meta ads, namely for subscription apps, this year? With the way that the internet, and specifically Meta, evolves, this could be a great deal.

Marcus clarifies that while there is a lot going on, most of the changes are a continuity of trends that started in 2024 and the increasing importance of certain areas.

1. Using AI for creatives

AI is definitely at the top of everyone’s mind, especially for ad creation. In Marcus’ opinion, brands that know how to do that are at an advantage over the rest.

Now, with the millions of AI tools popping up at the moment (almost all of them with AI in their names of course), I had to ask Marcus for his favorites in creating ads:

• For static ads – Midjourney AI and Flux AI
• For animating them – Runway AI and Kling AI
• For creating videos from text – Veo 2 from Google looks promising, and Luma Dream Machine is particularly powerful
• Arcads AI and Poolday AI for Talking Head UGC

Marcus has also been experimenting with working with an AI Creative – someone specialized in prompts to create ads. He commented that seeing what they can create with just a few prompts has been amazing. He’s curious whether you can have a high-quality video produced by the end of the year in a few sentences. To summarize, 2024 felt like the rise of the prompt engineer, and now 2025 may be the rise of the prompt creative. 

2. How to track ads success 

The rise of Aggregated Event Management (AEM) has been steady since 2024. In 2025 most subscription apps have switched to Metas on-platform probabilistic measurement framework and he isn’t seeing SKAdNetwork (SKAN) being used much anymore. What’s the cause of this shift? 

Well, Marcus attributes the quick adoption to (1) strong performance for most advertisers, (2) reduced pain of using it compared to SKAN (no privacy thresholds or postback delay) and (3) Meta making it the default without telling anyone as they love to do.  Yet AEM is a bit of a black box which makes it harder to troubleshoot, as well as harder to understand why it tracks in a certain way—a bit like trying to understand all these acronyms!

With AEM bringing back realtime reporting on creative level and AI on the rise, Marcus sees an impact on volume and way of creating creatives. Before ATT we came from an era where big advertisers created large volumes of ads, testing tiny iterations and letting the algorithm find the perfect combo. SKAN didn’t allow for this. Its restrictions in terms of attribution pushed for a more thoughtful and hypothesis-driven approach. 

AEM and AI have the potential to revert this trend. Creative velocity is already a buzzword on Linkedin. However, Marcus does hope we land at a more in-between spot so that we can be more considerate of the creative we want to test. He aims to have the best of both worlds: being able to test more while staying thoughtful about creatives—only time will tell if this is possible!

3. The lifespan of creatives

This made me curious about the lifespan of creatives. I’ve seen subscription apps, even when using AI, struggle to keep up with the volumes that platforms require. 

Would the speed of creative creation and the required increase in creative volume mean that content dies quicker than ever? I fear that we’ll go from expecting a few good pieces of content per week, to needing multiple per day—as demonstrated by creatives on TikTok right now. Yeah, I’m panicking a little right now.

Fortunately, Marcus was more optimistic than me. Regarding the lifespan, Marcus doesn’t feel like it is getting shorter.

Ad fatigue is to be expected with scale, but he said it also depends on the strength of the ad to begin with. Meta has always been competitive, but it is harder than before. I thought that was a really great point. If we want our creative to last in 2025, we need to be more conscious of creating great content in the first place.

So realistically speaking, lifespan may shorten a bit in 2025, but not to a huge extent if you continue to invest in quality creatives.

4. Goodbye classic UGC

Which brought us to the type of content that’s going to stand out this year. Marcus and I shared the same feeling that it’s been User Generated Content (UGC) for way too many years now.  With all the AI changes, what new content types will be on the rise?

Marcus made a great point in that what we refer to as UGC was always quite strange. We basically say UGC, but in 90% of the cases, it’s just talking head content. This was never natural to begin with, and never a real case of user generated. So, while we’re unlikely to say goodbye to Talking Head UGC completely, we can finally say hello to new types of UGC.

One format that has been working well for Marcus are text wall ads. A text wall ad is basically a complete story written out as a static or video.The background can be used to create context, hook users with weird imagery or even be blank. It goes against many of the classic creative (visual) rules as you have tiny text on top and lots of it. If we went back in time to the start of social media, a brand team would be ripping out their hair at the sight of it. Yet it’s what users posts organically (UGC) and that’s why it works.

Examples of text wall ads

That’s what we’ve seen perform surprisingly well in 2024 — content that actually looks like anyone could create it. It simply feels more organic and authentic. Anyone can create a text wall ad, the video editing skill needed is minimal, and if it works for your app, you can easily iterate lots of texts and different stories. Win-win.

Now let’s apply this more specifically to subscription apps, as that’s what we’re really here for. When we look at ads for subscription apps, Marcus made a great point that they tend to be more ‘meta’ in general because you can’t just show the physical product, so what works well does differ from the classic rules. For example, another type of content that Marcus sees gaining popularity are infographics. A great example of this are habit tracking apps, which love to employ infographics that show your productivity and motivation as a result. Again, this has a very organic feel. 

Meta Infagrahic ads by The Fabulous

They come in various shapes and forms which tend to give away some level of value directly in the ad, making users hungry for more. When paired with a web-to-app quiz funnel where users can jump straight into exploring more of that content, they become a powerful tool to diversifying your ad account away from just short-form video on Reels placed based on Marcus’ experience.

More examples of infographic ads

Another trend we’re seeing on organic social are faceless AI accounts. Maybe you’ve seen the posts on Instagram or Youtube, “Make $10K a month with a faceless account.”

Marcus expects this to trend to increasingly impact ad content. As user behaviour shifts and gets used to consuming this content, marketers pick up on it especially as it allows us to quickly grab some B roll and add AI voiceovers. Again, cheaper and easier content to create, yes! Either way, we need to shift away from the classic talking head format which has been overused in the last five years.

So, it’s safe to say that UGC is here to stay, but in different formats.

Honorable mention: many apps starting using messaging in 2024 which compares them directly to an app that’s already established. Like “Spotify but for…” or “Duolingo but for…”. Marcus has seen interesting results when experimenting with this approach yet overall is skeptical how far you wanna be pushing a messaging that solely depends on another product.

Comparison-style ads help communicate positioning fast, but overreliance can reduce perceived originality.

5. Placement is king

Let’s talk about another major aspect of Meta ads (and Marcus’ passion topic): placements. Reels placement continues to dominate with both Instagram and Facebook Reels driving and ever-increasing amount of impressions in Marcus’ accounts. There’s actually been interesing news from Meta, considering to make the product into a standalone app to take on TikTok.

However, large levels of spend especially on Instagram Reels present a risk for subscription apps because IG Reels tend to drive younger, lower intent, less qualified traffic. When optimizing for StartTrial, CPA often looks great to the advertiser as well as the algorithm. However trial-to-paid rates tend to lack behind placements like Facebook or Instagram Feed which deliver to older audiences. Any placement can be a valuable and profitable addition to your traffic composition yet you can evaluate them against the same cost per trial goal.  It’s truly a case of ‘you get what you paid for’!

We know what you’re thinking: why not just optimize for the subscription instead of the trial? Well,the challenge with focusing on subscription as conversion is that it takes toolong. Meta only allows for a 7 day post click window if you use AEM and really you want your conversion event to happen within 24 hours after install so that Meta quickly gathers signal to base their targeting decsion on.

So instead of moving your Conversion Event too deep into the funnel Marcus advises brands to gather first party user data at the beginning of the journey and triangulate this insight with  Meta’s breakdowns. E.g. by asking users for their age in onboarding and using the same brackets Meta does, you get a better understanding of trial-to-paid rate by age groups and can estimate profitability. Same goes for channel / placement, gender and users goals..  You’ll be surprised how often Facebook is more profitable than Instagram and how much better older age groups with high acquisition costs perform.

Marcus’ current favourite placement is Facebook Reels which tend to deliver to an older audience yet have low CPMs. Now once you find out that a specific audience from a specific placement is performing best for you, you can check which creatives have been responsible for delivering that traffic and make more of them. E.g. statics naturally appear more on the Facebook feed, To gain control over the delivery of certain ad types, which Marcus knows to perform long-term, he runs them in separate ad sets by creative type.

When you do your research, you’ll really understand what placements are most succesful for your audience. You need to see which placements are bringing quality trialists rather than cheap trialists and adjust your content accordingly. Definitely check out Marcus’ article improving trial conversion rates. 

Placements are potentially going to get even more interesting in 2025 as it’s likely Threads will be launching ads to a broader set of advertisers, as they are trialing now. This is closer to Twitter ads, more text-focused, more news-focused. So Marcus expects Threads will be an interesting new format to test for certain audiences.

Marcus has found that people really underestimate placement match for products. Advertisers optimizing for Purchase like D2C / Ecom get away with it because Metas algorithm has better visibility on the funnel yet apps can’t ignore the  difference in trial quality. He had this aha moment while working at Blinkist. Over time they applied a strategy using 2 completely separate ad accounts for their App Promotion and Web Conversion campaigns with the web campaigns delivering 80% of their impressions on Facebook Feed bringing in a much older audience much more willing to pay. To do so they designed ads that looked nothing like ads but new articles which are native to the Facebook Feed. These would redirect to an actual article (like this one), soft-selling the product and redirecting to the App Store like this one.

If a subscription app struggles with scaling Meta ads in 2025, what’s the first thing they should troubleshoot?

After listening eagerly to a pro gamer explain all the trends with Meta, I couldn’t help but ask one more question. One that I’m sure most of us Meta gamers have struggled with at one point or another: being completely and utterly stuck on a level. That situation when, despite our best efforts, we just can’t level up. This is usually when we’ll turn to cheatsheets and online forums, or in this case, to Marcus.

I believe ceilings are impacted by trends, and what’s needed to level up does change per situation. So here’s Marcus’ advice for whatever point you’re at.

Scaling up from 0-20K monthly budget

Level one is early stage spending (think up to 20K per month). Marcus claims that in this early phase,the most important thing is the technical setup. 

Double check Meta numbers with your in-app attribution survey and how many trials are coming in there versus Meta. You want to look at the match rate and see comparable or even more trials attributed in Meta than the survey because the survey is inaccurate. For example, some people might answer App Store if they eventually searched for it there, even if what led them to search for it was seeing a Reel on Facebook. 

Now, Metas AEM, when set up perfectly, can attribute a bit better. You don’t want to see 40% or 50% of trials missing. While you could easily factor that share in when evaluating performance, the algorithm can’t. It will have less data to base its decisions on, and 50% trials missing on account level could mean 100% missing on one ad and 0% on another, distorting the reality. 

Thanks to the black box of AEM we discussed earlier, it really is a matter of trial and error,testing your technical setup: using eventauto logging, manually logging them, your MMP and and other third parties involved in logging (e.g. RevenueCat) can and will all make a difference when it comes to how efficiently AEM displays the reality. 

Marcus also made a great point that people often look at Metas Event Manager, seeing events flow in and thing everything is all right. However, just because you send that data doesn’t mean Meta is able to attribute it. They just know that an event has occurred but not necessarily Who triggered it. He has seen cases where only 10% of trials were being attributed to Meta correctly. Tracking for subscription apps is tricky, so this needs to be the initial focus.

Scaling up from 20-100K monthly budget

Once this is dialed in, it’s a matter of creative, where you need to do some big swing testing and avoid getting stuck with small iterations. Marcus recommends you try different creative formats, styles and types of messaging to find out what resonates with which audience and combine that with the lower funnel data as described earlier. Too often, people start with a tiny A/B test, e.g., only changing the headlines on a bunch of static ads. While statics are impactful for some brands,  only run in certain placements. If that placement wasn’t a fit to begin with, your test results are irrelevant. So, start with your audience and competitor research, map out a number of wildly different content types that could be relevant to your ap,p and start testing there.

Now, Marcus explains, these things are consistently important as you scale but become absolutely crucial in getting to that 100K per month spend. 

Scaling up from 100-500K monthly budget

Once you are at 100K in monthly account spend, diversification becomes more and more critical. It’s not about running your single best ad only and scaling that in 1 campaign. By evaluating performance on audience and placement level, you’ll learn how to maximize your reach and addressable audience. You also want to start using all that traffic and data flowing in to improve the funnel, focusing on onboarding and pricing experiments for a new brand. For most apps, their onboarding isn’t ideal, and there are a lot of wins to be made. I appreciate that he focuses not just on the Meta side but also on the infrastructure on the app side to support it. 

As you continue to scale  it’s about keeping blended costs down by ensuring your organics scale with Meta. Referral and sharing loops can be a part of that strategy but ASO definitely needs to be. Growing install volumes will enable you to rank better on the store if you’re able to covert that traffic into ratings and have done your keyword optimization. Creative assets on the store should be optimized based on the learnings from your Meta creative testing and Marcus is a big believer that your app icon matters as much (if not more) than your screenshots. 

Web-to-app becomes interesting around this scale, if not earlier, as it also allows you to run Sales campaigns on Meta which can help you target completely new audiences. Marcus sees Web-to-app beingespecially interesting for reaching older users. As shown in the Blinkist example, creating a funnel that’s not using the App Store as the main destination allows you to tailor to certain placements and create more engaging user journeys. Charging on web comes with a number of benefits like lower payment fees, flexibility in terms of SKU creation and upsells…

Whether you want to start with web or app campaigns depends on your product on skillset. Marcus wrote a great article about this and highlights that eventually, you’ll need both flows in order to get your account to $500k and beyond

Overall, it feels like when struggling to scale in 2025, it’s crucial to ensure your technical setup is strong and that you aren’t boxing yourself too soon in terms of creative type.

Meta ads for subscription apps in 2025 

I couldn’t believe how much we covered in just an hour’s conversation, but Marcus honestly amazed me. His analytical and thoughtful approach was precisely the cool head you could expect from a pro-Meta gamer. I didn’t ask him if he’s ever played Hogwarts Legacy, but I have a feeling he would take my partner’s systematic approach rather than my rapid spell-shooting.

It was interesting to see where AI is helping to make Meta easier, such as for creatives, but it still isn’t at a level where it can focus on longer-term quality sales. However, in fairness, the last year of privacy and tracking changes has made that more challenging.

My number one takeaway was that context matching matters for subscription apps. We need to be more critical of where each type of ads will be displayed rather than just splitting creatives by angles. We need to test doing it by content type to boost performance. I don’t know about you, but I feel both reassured and inspired for the future of Meta ads in 2025!