Brutally Honest by Matej Lancaric

Share Brutally Honest by Matej Lancaric

🤺 Quote of the week

“only starting and already making more than 95% of the apps in the world” - Elias, WhatsApp message

Wordle Review

🧞 This Week @ Playablemaker

If going for core gameplay, do core gameplay

“No creative stealing, just pure gameplay? I keep saying that using only playable with core gameplay is not the correct approach. But if you have to, do it right and mimic the game as closely as possible.”

Get a 20% discount for 3 months with code: DANCINGUNICORN

Try now!

👾 THE AI (R)EVOLUTION

The AI revolution is here, and it’s transforming the way we approach mobile games marketing.

From creating scroll-stopping ad creatives to optimizing user acquisition strategies, AI is the game-changer we can’t ignore

🥦 Before you go

Ask Matej AI a question. It’s actually quite good!

We produced one more episode at 2.5 Gamers. Check it out now:

We started a new segment. Give it a go:

🍿 Fun starts in two & a half gamers Slack! Join now

We are opening up a Slack channel for everybody. It is the most inclusive slack for the gaming industry. Female, male, non-binary, juniors, mid-level, seniors, vendors, developers = EVERYBODY! We are waiting for you ♥️

Feel free to join here!

🧙 Thanks for reading this week’s newsletter!

Please do me a favor and share this with someone in your network who you think would benefit from the insights and brutal honesty. It would mean a lot to me!

Take care, everyone! See you next Monday!

👇 Reach out for feedback, questions, or say hi!

Give me some love (or hate), & Subscribe everywhere! ♥️

Playable Ads Trends in Mobile Games & Apps - March 2026

Matej Lancaric — Mon, 13 Apr 2026 20:18:50 GMT

As the UA landscape evolves, so do the creative trends that shape how games communicate with their audiences. With the rise of Applovin and increased spend on SDK networks, I’ve decided to bring you a monthly trends review of top-performing playable ads across different genres.

For those who read only summaries, here are the playable trends in March in a nutshell:

How apps are using playable ads and why you should pay attention
- I talk a lot about how playable ads are not only for gaming, but also this month. In the Top 10 new creatives on Applovin, 6 were not from gaming. So I decided to have a look at how different apps are utilizing this format. We will discuss 2 major use cases and what you should copy, even if you are in mobile gaming.

March: Apps are using playable ads, and why you should pay attention

Playable ads aren’t just for mobile games anymore, and that’s not just a fluke. In this month’s Top 10 creatives on AppLovin, six of them came from non-gaming categories.

🦄 Brutally honest AMA - April 2026

Matej Lancaric — Sat, 11 Apr 2026 21:16:58 GMT

Welcome to the new AMA (Ask Me Anything) section. I am going to gather a round of questions and then release a video for ONLY paid subscribers with all the answers.

Will be doing this monthly, so buckle up! The questionnaire is below. Go wild.

😲I generated 100 Ad variants for the price of a coffee. Here's the exact AI Workflow

Matej Lancaric — Fri, 10 Apr 2026 08:03:39 GMT

This guide walks through every step of a reusable Layer.ai workflow that takes game characters, transforms them into cute animals, animates them using multiple video generation models, and stitches everything together into a finished ad video.

The workflow is designed to be game-agnostic. Once it is set up, swapping in a new client’s game characters and end card is all that is needed to produce a completely new video. The creative concept, animal references, environment prompts, and animation style all remain the same. Only the inputs change.

Before we get into the how, here is the number that matters most: running this workflow end-to-end produces roughly 100 ad variants - three characters, three video models, multiple stitch combinations, across formats - at a total compute cost of somewhere between $2 and $5. That is not a typo.

Traditional video production for a single polished ad creative runs anywhere from $500 to $5,000, depending on the studio. This workflow does not replace the creative director. It replaces the production line.

The full pipeline covers six production stages:

Preparing reference images and character inputs
Transforming characters into cute animals
Upscaling the transformed images before animation
Animating each scene across three video models in parallel (Kling, Veo, Seedance, or whatever you want)
Stitching scenes together per model and then into a final master video
Adding an end card and exporting for the ad network

Each section below explains what to do, why it matters, and what to watch out for. Prompts, node settings, and reuse instructions are included throughout.

or it can look like this if you play around with it

… up to you!

Step 1: Prepare your reference images

Before the workflow can run, you need two distinct sets of images loaded into six input nodes on the left side of the layer.ai graph. Getting these right is the single most important thing you can do for output quality. Everything downstream - the animal style, the scene composition, the animation behavior - flows from what you put in here.

Whatever you have in your own game assets folder…

Animal reference images (3 nodes)

The first three Workflow Input nodes are for animal style references. These are example images of cute animals in the visual style you want your output to match. They are not stock photos of real animals.

They should be illustrations, renders, or concept art that shows the specific aesthetic you are targeting - soft 3D, kawaii, match-3 art style, big eyes, rounded shapes, pastel shading.

All three reference images should share a consistent visual language. If one reference has harsh cel shading and another has soft watercolor coloring, the IP-Adapter will receive conflicting style signals and the output will be inconsistent. Pick three images that feel like they belong to the same art direction.

Use PNG or JPG, ideally square or portrait crop
All three should share the same lighting, proportions, and colour palette
Avoid references with complex busy backgrounds — the character should be the clear subject
The cuter and more stylised the references, the more the transformation will lean into that direction

Tip: finding good references
Search for match-3 game character art, mobile game UI characters, or kawaii animal illustration packs. Pinterest, ArtStation, and Behance are good sources. If you are producing for a specific client with an established art style, extract characters directly from their game assets.

Game character inputs (3 nodes)

The next three Workflow Input nodes take the actual in-game characters from the client whose game you are making the ad for. These are the characters that will be transformed. Each one maps to one output scene in the final video.

Character image quality matters here. If you feed in a small, low-resolution screenshot, the transformation and upscale nodes will have much less to work with. Where possible, use official game art at native resolution, character portrait images from the game’s marketing materials, or clean in-game screenshots cropped tightly around the character.

Each character should be clearly visible and occupy most of the frame
Clean or transparent backgrounds produce better transformations than busy scene backgrounds
Characters should be visually distinct from each other — different silhouettes, different colour schemes
Three characters produce a three-scene video. You can add more nodes for longer videos.

Reuse mechanic
To produce the same video for a different game, the only thing you change is these three character input nodes. Load in the new game’s characters and run the workflow. Every other setting stays the same. This is the core efficiency of the pipeline.

Step 2: Transform characters into cute animals

Each game character image passes through two image-to-image transformation nodes. The first one, called Transform to Cute Animal, takes the character and reshapes it into an animal using the style information from your reference images. The second one, called Transform the Environment, places that animal into a suitable scene or background.

The animal transformation prompt

The prompt for the Transform to Cute Animal node needs to do three things: describe the animal category, reinforce the style, and establish the mood. Here is a solid starting point that you can adapt:

A cute fluffy animal character, big shiny eyes, rounded soft body, pastel colors, kawaii style, smooth 3D render, warm studio lighting, simple clean background, expressive face, match-3 mobile game art style

The phrase “match-3 mobile game art style” at the end anchors the output in a genre the model understands. If the your game has a different art style, replace this with a description of that style - “cartoon RPG style”, “flat 2D illustration”, “isometric puzzle game art”.

The negative prompt should exclude anything that conflicts with the cute animal direction:

Realistic, photographic, dark, scary, aggressive, ugly, low quality, blurry, distorted, human face, text, watermark, original game art style

Or you go pro mode:

The environment transformation prompt

After the animal is generated, the second node places it in a scene. You have two strategies here. The first is to fix the environment so that all three characters appear in a consistent setting - a kitchen, a forest clearing, a candy world. This makes the video feel like a coherent world. The second strategy is to let the AI assign each character an environment that suits their personality, which often produces more interesting individual scenes but less visual consistency across the video.

For your first run, a fixed environment is recommended because it is easier to QA and easier to iterate on. Here is an example for a kitchen setting:

Cute animal character in a cosy cartoon kitchen, colourful tiles, wooden counters, warm sunlight through the window, pastel colour palette, match-3 game environment, cheerful and playful atmosphere

Once you have run the workflow once and are happy with the style, you can add a variable to this prompt - for example, swapping the room type based on the character class - to create more variety across iterations without losing stylistic consistency. Or you go pro mode and tweak the enviro prompt as you wish:

Step 3: Upscale before animating

After each character has been transformed and placed in an environment, the resulting image passes through an upscale node before it reaches the video generation models. This step is represented by the three amber/gold nodes in the middle of the graph, one per character.

The reason upscaling happens here rather than at the end is that the video generation models - Kling, Veo, and Seedance or whatever you want to do - use the first frame of the video as their primary anchor for animation quality. A higher-resolution, crisper first frame gives these models more detail to work with when generating motion. The difference in output quality between a 512px input and a 1024px input is significant, particularly in fine details such as fur texture, eye reflections, and the edges of small objects in the scene.

Thanks for reading Brutally Honest by Matej Lancaric! This post is public so feel free to share it.

The upscale node runs per character, so you have three of them in the graph. Each one outputs its result to all three video models simultaneously, which is why the graph branches out dramatically to the right of the upscale nodes.

Target output resolution: 1024x1024 or 1024x1792 for 9:16 vertical video
Check the upscaled output visually before the workflow continues - if there are artifacts, tiling patterns, or over-sharpened edges, address them here rather than after the video is generated
The upscaled image should look like a final, production-ready illustration before it enters the animation stage

Quality gate
Treat the upscale output as a checkpoint. Pause here and review all three upscaled images. If any of them have composition issues, style inconsistencies, or visual artifacts, regenerate those characters before proceeding. It is much faster to fix a static image at this stage than to regenerate a video after the fact.

Step 4: Animate each scene across three models (if you want)

This is where the workflow earns its complexity. Each upscaled image fans out to three video generation models running in parallel: Kling, Veo, and Seedance. With three characters, this produces nine video clips in total - three per character, one from each model. The parallel approach means you get multiple creative variants without additional manual work, and you can choose the best-performing clip per character after reviewing the results.

Writing the animation prompt

The animation prompt specifies the action the animal should perform in each video. Unlike the image-generation prompts from the previous steps, this prompt must describe motion. Static descriptions produce stiff results. Action-oriented language produces better animation.

The prompt should describe the animal, its action, and the overall mood. Keep it focused and specific:

A cute fluffy animal character bounces happily, tail wagging, ears flopping, big eyes blinking, looking at the camera with a playful expression, smooth cartoon animation, cheerful and energetic

If you want different actions per character - one bouncing, one waving, one spinning - you can customize the prompt per animation node. This adds variety to the final stitched video and makes it feel less repetitive.

Clip settings

Each video clip should be between three and five seconds long. Shorter clips feel abrupt when stitched together. Longer clips give the animation room to develop but increase generation time. For a standard three-character video with an end card, a total runtime of around fifteen to twenty seconds is appropriate for most ad networks.

Clip length: 3–5 seconds per character
Frame rate: 24fps for a cinematic feel, 30fps for a more digital/game-adjacent look
Aspect ratio: 9:16 for TikTok/Meta Reels, 1:1 for square placements, 16:9 for YouTube pre-roll

Comparing the three models

Each model has different strengths and will produce noticeably different results on the same input. The right choice depends on the specific character and scene.

Efficiency tip
Once you have run this workflow several times and have a sense of which model consistently performs best for a given art style, you can disable the other two animation branches on subsequent runs. This significantly reduces generation time without sacrificing quality once you know what works.

Step 5: Stitch scenes together

After the nine animation clips are generated, the workflow passes them through a two-stage stitch process. The first stage uses three per-model stitch nodes — Kling Stitch, Veo Stitch, and Seedance Stitch — to concatenate the three character scenes within each model. This produces one reel per model. The second stage uses a master Stitch Videos node to combine all three model reels into a single output.

The reason the workflow stitches per model first rather than mixing clips immediately is that it gives you three complete model variants as intermediate outputs. This is useful for creative review — you can watch the Kling reel, the Veo reel, and the Seedance reel as standalone videos and pick the one you want to deliver, or pick individual scenes from different reels.

Scene order

The order in which the three characters appear in the video matters for the rhythm of the ad. A common structure is to open with the most visually striking character (the one whose transformation is most dramatic or whose animal form is most appealing), build with a second character that has a different colour scheme or action, and close with a character that sets up the end card well.

The stitch nodes take clips in the order you connect them. If you want to change the order, rearrange the connections in the graph rather than renaming nodes. This is faster and keeps the graph readable.

Transition type

The default stitch is a hard cut between scenes. This works well for fast-paced ad formats like TikTok and AppLovin, where cuts feel energetic.

If you want a crossfade between scenes, add a transition duration setting in the stitch node configuration - typically 0.3 to 0.5 seconds is enough to smooth the edit without slowing the pace.

Avoid long fades. They make the video feel slow and eat into your total runtime, which matters when you are working toward a fifteen-second target.

Hard cut: best for energetic, fast-paced formats (AppLovin, TikTok, Meta)
Crossfade (0.3–0.5s): better for slightly longer formats or premium placements
Do not use fade-to-black between scenes — it feels like the video is ending prematurely

Review checkpoint
Before passing the stitched video to the end card stage, watch it in full at least once. Check that the transitions are clean, the clip order feels right, and there are no unexpected freezes or artefacts at the clip boundaries. These are much easier to fix now than after the end card is composited.

Step 6: Add end card and export (optional)

The end card is the missing node in the current workflow graph. The workflow goes directly from Stitch Videos to Save Outputs without compositing any branding, CTA, or store badge. For a finished deliverable ad, the end card is essential - it is what tells the viewer what to download and where to find it. If you really want it!

What the end card should contain

A standard mobile game ad end card typically includes the game logo or wordmark, a store badge (App Store, Google Play, or both), a CTA line (usually “Play now” or “Download free”), and optionally a brief animated UI moment from the game itself. For this workflow, the simplest approach is a pre-rendered static end card video — a three-to-five second clip that you produce once per client and reuse across all video variants.

Adding the end card to the workflow

In Layer.ai, add a new Workflow Input node for the end card video file. Connect its output to the Stitch Videos node as the final clip in the sequence. The stitch node will append it after the last character scene. This means the end card is automatically included in all three model variants without any additional work.

Create the end card as a standalone video file: 3–5 seconds, same resolution and frame rate as the main video
Save it as client_endcard.mp4 in a dedicated client assets folder
Add a Workflow Input node to the graph and load the end card file
Connect the end card output to the Stitch Videos node as the last input
Run the workflow and verify that the end card appears cleanly after the final character scene

Reuse mechanic
Store one end card template per client. When you run the workflow again for the same client with different characters, reattach their existing end card node. When you onboard a new client, produce one end card for them and store it. After that, every future run for that client requires zero end card work.

Export settings

The Save Outputs node on the far right of the graph writes the final files. Configure it to output all three model variants so you have A/B test material ready for the ad network without any additional work.

File naming: use a consistent convention such as clientname_model_YYYYMMDD.mp4
Export all three variants: Kling, Veo, Seedance
Video specs: H.264 encoding, AAC audio if music is included, maximum file size per network (AppLovin: 50MB, Meta: 4GB, but aim for under 30MB for fast loading)
Resolution: 1080x1920 for 9:16, 1080x1080 for 1:1, 1920x1080 for 16:9
Frame rate: match whatever was set in the animation stage (24fps or 30fps)

Once exported, upload all three variants to the ad network as separate creatives and tag them clearly (Kling variant, Veo variant, Seedance variant). Let them run for at least three to five days before drawing conclusions on which model produces the best performance for this client and genre. Over time, this data will tell you which model to default to and which to disable on future runs.

Reusing the workflow for a new game

The entire value of this pipeline is its reusability. Once the workflow is configured and producing good output for one client, adapting it for a new client should take less than ten minutes of setup time.

Here is the complete list of what changes and what stays the same:

If a new game has a very different game art style - for example, a dark fantasy RPG versus a pastel puzzle game - you may also want to update the environment prompt and the animal reference images to match the new tone. But for most casual mobile game players, the default cute-animal style will work across the board.

What 100 variants actually look like and what it costs

Let’s do the math, because this is where the workflow stops feeling like a cool experiment and becomes a competitive advantage.

A single run of this workflow produces the following outputs: three character transformations, each animated by three video models (Kling, Veo, Seedance), in up to three aspect ratios (9:16, 1:1, 16:9), with an end card appended. That is already 27 distinct video files from one workflow run. Add in prompt variations for the environment - kitchen, outdoor, candy world, and you are at 80 to 100 unique creatives before you have touched the character inputs a second time.

Now the cost. Video generation with Kling, Veo, and Seedance runs at roughly $0.05 to $0.15 per clip, depending on the model and clip length. With nine clips per full run (three characters × three models), that is approximately $0.45 to $1.35 in generation cost per run. Upscaling adds cents. The image transformation stage adds a few more. A fully loaded cost per workflow run - all nine clips, all three stitched variants, exported and ready to upload - sits somewhere between $2 and $5.

At $5 per run and 20 runs to dial in the workflow for a new client, you have spent $100 to produce a fully iterated, production-ready creative suite. A single revision round with a freelance motion designer typically costs more than that.

The production readiness point is worth stating explicitly. These are not rough concept videos. The output of this workflow is upload-ready: correct resolution, correct frame rate, end card composited, and file size within network limits. There is no post-processing step, no handoff to an editor, no round of feedback. You run the workflow, you review the outputs, and you upload the ones that pass QA. The whole cycle from brief to live creative can happen in an afternoon.

This changes the economics of creative testing in mobile UA. Instead of producing three to five creatives per month and hoping one lands, you can produce 50 to 100 variants, let the algorithm find the winners, and kill the losers after 48 hours. The creative feedback loop that used to take weeks now takes days. That is not a workflow improvement. That is a different way of doing UA.

EXAMPLE VIDEOS FROM THE WORKFLOW

What are you waiting for?

Softlaunch your game PROPERLY

🧪 Last Asylum: Plague global launch case study

Matej Lancaric — Tue, 07 Apr 2026 15:56:07 GMT

When I started receiving ten new creatives per day on Facebook after a game launched cold into the global market in early 2026, I knew something worth talking about was happening. That game was Last Asylum: Plague, published by 37GAMES GLOBAL, the overseas arm of Chinese entertainment giant 37 Interactive Entertainment. The game didn’t soft-launch.

It didn’t test quietly in one territory (or a few). It simply arrived and arrived at scale.

The episode of Two and a Half Gamers we recorded with Jakub Remiar and Felix Braberg to dissect Last Asylum: Plague is one of the most information-dense we’ve produced, covering the game’s surprisingly deep onboarding loop, its eclectically sourced 4X mechanics, the company’s publicly disclosed financials, and most strikingly - the revelation that 37 Interactive Entertainment reported using AI in the production of 70% of its video ad creatives and 80% of its 2D art assets. My mind was pretty much blown. Let’s unpack all of it.

How to use AI in creatives

Who Is 37 Interactive Entertainment?

To understand Last Asylum: Plague, you first have to understand the company behind it, because this game did not emerge from a startup operating on instinct. It came from a publicly traded, A-share-listed conglomerate with nearly fifteen years of institutional knowledge in mobile game publishing.

37 Interactive Entertainment, officially Sanqi Interactive Entertainment (三七互娱), was founded in September 2011 by Li Yi Fei and Zeng Kaitian. It is headquartered in Guangzhou and listed on the Shenzhen Stock Exchange under the ticker 002555. The company has grown from a domestic browser game publisher into what it describes as “a global top 20 listed game company and a key cultural company in China,” with inclusion in the CSI 300 index, the MSCI Index, and Goldman Sachs’ New China Nifty 50.

Its publishing infrastructure is substantial. Under the 37 Interactive umbrella sit multiple distinct platforms: 37Games (China domestic), 37 Online Games, 37 Mobile Games, and 37GAMES (overseas publishing). For Last Asylum: Plague, the relevant entity is 37GAMES GLOBAL, with the game developed by Cloudwalker Ltd. The company has offices in Beijing, Shanghai, Anhui, Hubei, Hong Kong, and overseas hubs across Europe, America, Japan, South Korea, and Southeast Asia. As of December 2024, the workforce stood at 3,249 employees, down 5.74% year-over-year — a number that becomes interesting in the context of AI adoption I’ll discuss later.

I pulled the financials directly from the company’s public financial statements during the podcast, and they were genuinely striking. For the full-year 2024, 37 Interactive reported revenues of approximately 17.44 billion yuan (roughly $2.4 billion USD) with a net profit of approximately 2.67 billion yuan ($371 million USD). In the first half of 2025 alone, revenues reached around 1.2 billion USD, with net profit of $200 million. Their overseas revenue for 2024 was reported at $800 million — but as Jakub pointed out, Sensor Tower attributed only around $474 million to observable Western store revenues.

The gap reveals something we kept coming back to on the show: they generate 4 to 5 times more revenue than we see.

Western app store data is, for Chinese publishers, the tip of the iceberg.

How much money they really make?

The short answer: Sensor Tower’s $4B is roughly 22% of their real lifetime revenue. The financial reports tell a very different story.

From the annual reports, 37 Interactive’s revenue history goes from $47M in 2010 all the way to $2.42B in 2024 (CompaniesMarketCap), with the biggest jump coming in 2015 (up 662%) when they went public and supercharged mobile expansion. Add it all up year by year, and you get approximately $18.4B in total reported revenue since founding.

The Sensor Tower $4B figure is only the Western/overseas app store-visible slice. The 2024 split makes it concrete: overseas income had triple-digit growth for two consecutive years around 2020–2021 (EqualOcean), but even at peak overseas scale, Western stores represent a minority. On the podcast, we confirmed: $800M overseas revenue in 2024 per their financial report, $474M visible on Sensor Tower - meaning even within “overseas,” Sensor Tower misses nearly half. And that’s before accounting for the $1.6B+ that comes purely from China.

A few things worth noting in the chart:

The 2020–2021 ramp is when Puzzles & Survival went global and overseas revenue hit ¥969M in H1 2020 alone, up 95% YoY (PR Newswire). That’s when Sensor Tower’s slice started becoming meaningful. Before 2019, 37 Interactive's revenue was essentially China-only and completely invisible to any Western tracking tool.

The plateau since 2021 (~$2.3–2.5B/year) shows the company has been stable rather than explosive in recent years globally, which makes Last Asylum: Plague an important new bet to reignite top-line growth in overseas markets specifically. Let’s finally talk about the game…

Last Asylum: Plague - The Game Itself

Last Asylum: Plague launched globally on February 26, 2026, notably with zero soft launch. The game is a 4X strategy title developed by Cloudwalker Ltd. and published by 37GAMES GLOBAL.

Within five weeks of launch, it had accumulated 5.7 million downloads on Android alone, with 3.1 million downloads in a single month, ranking #1 in the strategy category.

The thematic premise is a medieval plague setting: players assume the role of a plague doctor tasked with managing a makeshift hospital, healing the sick, defending against rat-borne infection vectors, and ultimately building a medieval 4X city-state.

The aesthetic borrows heavily from the atmosphere of Dark War Survival - gloomy, hand-painted, plague-dark, but the mechanical DNA is sourced from multiple directions simultaneously.

This is what we identified on the show as Last Asylum’s central design innovation: it is not merely a reskin, but a synthesis of the best-performing mechanics from every major 4X game in the current market, assembled into a single coherent progression system.

The Onboarding (My Perfect Hotel): The game’s first hours are essentially a reskin of My Perfect Hotel. Players manage wards, assign nurses, restock pharmacies, and expand rooms in a familiar idle-management loop. This is not superficial - it is the core engagement mechanic for the early game and the primary source of “Stars,” the currency that gates progression into the 4X layer. As Jakub noted, “Stars come only from this mechanic first. Later, you can get Stars elsewhere. But first, you really need to play this. It’s an actual game.” The My Perfect Hotel onboarding isn’t a bait-and-switch - it’s structurally integrated as the progression gating system into the headquarters building.

The Road (River Game): Once the player exits the hospital onboarding, they encounter a road system lifted directly from River Game. The road unlocks segments of territory that expand the buildable 4X base footprint. This is the map-expansion mechanic that gives Last Asylum its spatial depth.

The City Layout (King Shot): The spatial arrangement of the city - headquarters at top, with the road stretching down and outward beneath it - is architecturally King Shot. We described it on the show as “one to one King Shot.”

The Invasion Timer (King Shot/River Game hybrid): Red invasion events with countdown timers create urgency and map pressure. This is King Shot-style tension design, but modified with River Game’s territorial unlock framework.

The Gacha Characters (Frozen City): The hero roster and gacha system for character acquisition follow the Frozen City standard, which has become broadly adopted in 4X games featuring auto-battle mechanics.

The practical outcome of this synthesis is a game that veteran 4X players cannot meaningfully accuse of being “fake.” There is no misleading advertising in the traditional sense, because the My Perfect Hotel mechanic genuinely exists in the game, is genuinely required for progression, and is genuinely well-designed. Felix put it simply: “It’s not fake anymore.” The transition from casual management game to 4X city-builder is, as we all agreed on the show, “extremely smooth.” Players who don’t know what they’re getting into may not even realize they’ve crossed the bridge.

At the time we recorded the episode, Last Asylum: Plague was generating approximately $300,000 per day in revenue in its first month, tracking at a comparable early-game pace to King Shot, which generated approximately $4 million in its first month. If Last Asylum maintains that pace, it has a credible path toward becoming one of the biggest 4X launches of 2026. Look at this now!

$8 mil vs $4 mil. This is going to be an interesting year!

37 Entertainment’s AI Disclosure: The numbers that change everything

The most extraordinary part of researching this episode was reading 37 Interactive Entertainment’s publicly disclosed financial and operational report live on air. The AI adoption statistics within that report are, to put it plainly, a signal of where the entire mobile gaming industry is heading.

The key figures disclosed by the company for its overseas publishing operations:

80% of 2D art assets are AI-assisted
70% of all video creatives involve AI generation
AI localization covers 85% of overseas titles
AI development tools are improving R&D efficiency by 30%

For context: most Western game studios hedge with language like “we’re exploring AI tools in certain pipelines.” 37 Interactive is disclosing, in a public financial report, that AI-generated or AI-assisted content accounts for the overwhelming majority of its creative production. That is not a pilot program. That is operational infrastructure.

The creative volume produced for Last Asylum: Plague confirms these numbers in practice. In the first 30 days following launch, the game ran 3,333 unique creatives on Facebook alone, and that figure covered only one network. With 2,500 videos across platforms and 22 playable ads, the output volume is simply not achievable at this cost structure without AI-generated production pipelines. Also, 22 playables? Come on!

The creatives I went through on the show covered an astonishing range of concepts: horror hooks, township-style mechanics, township building simulations, King Shot–style pressure ads, Witcher-referenced fantasy imagery, White-Out Survival visual callbacks, garlic-versus-plague medieval absurdism, and “Get Out”-style jump-scare hooks. All within a single game’s creative library. All within a single month of launch.

The quality is notably high. The AI-generated video content we reviewed was not cheap or inconsistent - Jakub called it “very, very good quality AI.” Static image creatives drawn in the same pipeline were similarly polished.

The plague doctor character design - visually distinctive, immediately recognizable across creatives - is what Felix identified as a character that “sticks out so well in creatives. You recognize it.” The AI toolchain is not undermining brand coherence; it is scaling it.

On the text side, some of the English-language copy in the creatives was imperfect. But honestly - who cares? It doesn’t really make any difference. There is the headline, and the hook works before the text is read. The AI-generated imagery does the UA work.

The competitive implications for the industry

On the show, Jakub, Felix, and I spent considerable time on the competitive consequences of what 37 Interactive has built, and the conclusion was uncomfortable.

The core concern is efficiency amplification. Chinese mobile game studios are already operating with larger, faster, and more systematized creative teams than most Western publishers. 37 Interactive generated $2.4 billion in revenue in 2024 with 3,249 employees - a revenue-per-head ratio that most Western studios cannot approach. Now, those same teams are producing creative volume that would require ten times the headcount in a non-AI workflow.

The headcount reduction of 5.74% year-over-year at 37 Interactive, in the same period they dramatically scaled AI adoption, is not a coincidence.

As Jakub put it: “If Chinese studios are able to squeeze their giant creative teams with AI, then it’s kind of a game over. What can you do then? Because they’re efficient to begin with and they’re gonna be even more efficient now.”

My own pointed observation during the episode was about playable ads specifically - Last Asylum launched with only 22 playables against 2,500 videos. This represents the one genuine gap in their creative strategy.

If 37 applies the same AI-assisted production scaling to playables that it has applied to video, the competitive advantage widens further.

Build Playables with AI

My advice to anyone competing in this space: convert some fraction of your video production capacity into playable ad production. That is currently the underinvested format in Last Asylum’s library, and playables remain the highest-engagement creative format across most ad networks. Even converting 20% of their video output into playables would be a significant competitive escalation.

What makes the Chinese publisher case distinct is not just the adoption rate but the willingness to disclose and, implicitly, to compete on AI adoption as an operational advantage. Where Western studios treat AI tooling as a sensitive topic - amid ongoing debates about job displacement and ethical concerns, 30% of Western developers surveyed in the 2025 GDC State of the Industry Report expressed negative views of AI’s impact, a 12-point increase from the prior year.

37 Interactive publishes its AI adoption metrics in its financial reports as a competitive differentiator. They brag about it. In China, it’s efficient.

37 Entertainment’s Game Design Philosophy

One pattern runs through 37 Interactive’s history that Last Asylum: Plague extends rather than departs from: the deliberate identification of what works in the market, followed by synthesis and refinement rather than pure originality.

Global Launch your game like a BOSS

Puzzles & Survival didn’t invent the hybrid 4X/Match-3 genre, but it executed it better than its predecessors and then embedded the match-3 mechanic so deeply into the UA creative strategy that it became inseparable from the game’s market identity. As documented in the Naavik deep-dive on Puzzles & Survival, Sally Peng, Vice President of 37GAMES, described the philosophy directly: “We always try to use actual gameplay in our game as the creative focus for our advertising. We also give feedback and communicate with our R&D team on what the creative team is doing, and they integrate everything back into the game.”

Last Asylum: Plague extends this philosophy with a broader synthesis of surface area. The game doesn’t just take one mechanic from one competitor - it takes the best-performing mechanic from every relevant competitor and integrates them into a coherent progression system where each borrowed mechanic has a genuine function. The My Perfect Hotel loop gates 4X progression. The King Shot invasion timer creates urgency. The River Game road system provides spatial expansion. The Frozen City gacha provides character depth. None of these is a decorative feature. They are load-bearing pillars of a design that passes what might be called the honest onboarding test: the game plays exactly as its creatives suggest.

This is meaningful in an era where regulators and platforms are increasingly scrutinizing “fake ad” practices. 37 Interactive isn’t simply building bait-and-switch games - they are building hybrid games where the bait is the game.

Conclusion: The 4X Playbook

I said it at the opening of the episode: this is almost like a how-to-launch-a-4X-game playbook. Seriously, the new one.

Softlaunch your game PROPERLY

By the end of recording, the evidence had accumulated: zero soft launch, 3.3 million downloads in a month, $300K daily revenue tracking toward a Kingshot-comparable first month, 3,333 Facebook creatives in thirty days, 70% AI video production, a structurally honest onboarding loop built from four different games, and a publicly listed parent company generating $2.4 billion in annual revenues with AI tools reducing headcount requirements while increasing creative output. Now the Last Asylum is trending event better then Kingshot launch

We ended the episode with a question: Is this the biggest game of 2026? Jakub’s answer: “Maybe we’re looking at the biggest game of the year. Never know with these things.” Mine: It’s gonna go great. They have everything in place. In terms of UA diversification, creatives, and game design. Please, don’t f*ck it up!

What 37 Interactive has built with Last Asylum: Plague is not just a game launch. It is a demonstration of what mobile game publishing looks like when you combine institutional 4X design knowledge, full-funnel creative process, AI-scaled production pipelines, and zero risk aversion at launch. In the current competitive environment, that combination is a very different kind of plague (pun intended XOXOXOXO): fast-spreading, hard to contain, and very good at finding the gaps in your defenses.

Improve your UA skills NOW

Brutally Honest #207 by Matej Lancaric

Matej Lancaric — Mon, 06 Apr 2026 20:10:07 GMT

Hey, FAM! What’s up?

🥃 Drink talk

Recently created a bundle with all the information you need to excel in the UA world.

🦩Tip of the week

I recently started running new Google campaigns for one game. Its a hybrid casual game (by monetization standards - AKA 50 vs 50 split Ads/IAPs).

all best practices are pointing into 7 day window for Purchase event and 1day window for Ad impression event. Well, I also shared this recently:

The above table is a bit controversial given the fact Google recommends usually 7-14 days window. But by DEFAULT, they turn on 90 days.

The Trade-off:

Shorter Window = Faster feedback loop, faster learning, but lower LTV cap.
Longer Window = Slower feedback loop, but captures true Whale LTV.

But then I started thinking about Applovin campaigns and how both IAP, Ads oh well even blended ROAS campaigns are optimizing for 28 day window.

I set my Purchase for 28 days and Ad impression for 7 days. So far, the results are amazing. Next step would be creating new ad account to run different attribution windows against each other. You should try that too!

This newsletter is brought to you by FastSpring:

Handling direct-to-consumer (D2C) payments and webstores for over 20 years.
Launch a D2C store in record time.
Includes Steer Safe™ for seamless and secure in-game steering to D2C checkout.
Powering top mobile publishers around the world, like Rovio, TechTree Games, + a LOT more
Fast-growing, profitable, and financially secure.
Headquartered in the US with offices from Amsterdam to Singapore to deliver local service on a global scale.
A no-drama D2C partner with real-world experience at scale, FastSpring is a partner you can trust with your players.

Stop managing payments, start scaling your game. Get a personalized D2C strategy session at fastspring.gg

🐊 Cornelius: The Suspicious Investi-gator

True detective-like exploration of creative trends and historical influences. Delving deep into the world of creatives to uncover hidden patterns and inspirations:

𝗧𝗵𝗲 𝗿𝗶𝘀𝗲 𝗼𝗳 𝘀𝗼𝗽𝗵𝗶𝘀𝘁𝗶𝗰𝗮𝘁𝗲𝗱 𝗼𝗻𝗯𝗼𝗮𝗿𝗱𝗶𝗻𝗴 𝗶𝘀 𝗻𝗼𝘁 𝗮 𝗷𝗼𝗸𝗲. The latest 4X Last Asylum launch is living proof. 𝗧𝗵𝗲 𝗠𝘆 𝗣𝗲𝗿𝗳𝗲𝗰𝘁 𝗛𝗼𝘁𝗲𝗹 onboarding isn’t a bait-and-switch - it’s structurally integrated as the progression 𝗴𝗮𝘁𝗶𝗻𝗴 𝘀𝘆𝘀𝘁𝗲𝗺.

𝗧𝗵𝗲 𝗢𝗻𝗯𝗼𝗮𝗿𝗱𝗶𝗻𝗴 = 𝗠𝘆 𝗣𝗲𝗿𝗳𝗲𝗰𝘁 𝗛𝗼𝘁𝗲𝗹

The game’s first hours are essentially a reskin of My Perfect Hotel.

Players manage wards, assign nurses, restock pharmacies, and expand rooms in a familiar idle-management loop.

This is not superficial - it is the core engagement mechanic for the early game and the primary source of “Stars,” the currency that gates progression into the 4X layer.

As Jakub noted in our podcast, “Stars come only from this mechanic first. Later, you can get Stars elsewhere. But first, you really need to play this. It’s an actual game.”

The My Perfect Hotel onboarding isn’t a bait-and-switch - it’s structurally integrated as the progression gating system into the 𝗵𝗲𝗮𝗱𝗾𝘂𝗮𝗿𝘁𝗲𝗿𝘀 𝗯𝘂𝗶𝗹𝗱𝗶𝗻𝗴.

With this onboarding, players never guess they are actually playing a 4X game.

Yours truly, Cornelius.

Increase your creative win rate: https://payhip.com/b/tu4nk

💀 Dr. Doom is not amused

One of my industry friends amused me with his direct and honest TL;DRs, so we agreed to share it here, too! My friend remains anonymous as Dr.Doom!

“This shit never disappoints. Nintendo is again trying to patent the act of throwing a “sub characters with balls” in order to fight Palworld and then Palword changes the way how they summon the characters 😃 (which they already did, duh!) This time Nintendo itself get rejected by US court. This is not Tokyo District court, welcome to the USA 🤣” Sincerely, Dr.Doom (https://www.gamesindustry.biz/us-patent-office-rejects-nintendos-summon-subcharacter-and-let-it-fight-patent)

🎉 “Ha-ha” moment

Because life is too short to be serious.

Thank you for reading Brutally Honest by Matej Lancaric. This post is public so feel free to share it.

🎮 two & a half gamers new episode

This might be one of the 𝗹𝗲𝗮𝘀𝘁 𝗳𝘂𝗻 games making money right now. Still, made $𝟮𝗺𝗶𝗹+ 𝘀𝗶𝗻𝗰𝗲 𝗹𝗮𝘂𝗻𝗰𝗵 and growing

Brilliant Sort is a perfect example of how mobile games are evolving beyond traditional “fun”. The gameplay is repetitive, slow, and often frustrating. 𝗕𝘂𝘁 𝘁𝗵𝗮𝘁’𝘀 𝗲𝘅𝗮𝗰𝘁𝗹𝘆 𝘄𝗵𝘆 𝗶𝘁 𝘄𝗼𝗿𝗸𝘀. It taps into a psychological loop of organizing, cleaning, and completing structured tasks - something players naturally find 𝘀𝗮𝘁𝗶𝘀𝗳𝘆𝗶𝗻𝗴. One could say 𝗔𝗦𝗠𝗥! Wink wink...

What really stands out is the monetization design. The game intentionally creates friction, forcing players to watch ads or spend currency just to make basic progress. Combined with UA using highly polished (and often fake) creatives, it creates a strong acquisition-to-monetization pipeline.

𝗝𝘂𝘀𝘁, 𝗳𝗼𝗿 𝘁𝗵𝗲 𝗹𝗼𝘃𝗲 𝗼𝗳 𝗴𝗼𝗱, 𝗱𝗼𝗻'𝘁 𝗽𝗹𝗮𝘆 𝘁𝗵𝗶𝘀 𝗴𝗮𝗺𝗲! Its super annoying..

📺 Watch on YouTube

👕 Get our Merch: 25gamers.com/shop

🤑 Paid subs only

Sneak peek at what you get for your support:

🦄 Diary of a Dancing Unicorn

Today was a good day.

Hey hey, here is the playable bible.. I love this guy, we worked together in the past. Insights from this bible could help the team..

Oooh, I love you too.. We meet again! ❤️❤️❤️

OMG, you are here in the slack, no idea.. lets chat.

Oh yeah, lets do that!..

Today was a good day.

Share Brutally Honest by Matej Lancaric

🤺 Quote of the week

“We love details which male audience may find boring or friction.” - Asli, linkedin comment

Brilliant Sort Review

🧞 This Week @ Playablemaker

“I have been doing this research for the last 2 weeks, and your tool looks by FAR the best on the market.” Potential Client, Demo call

Get a 20% discount for 3 months with code: DANCINGUNICORN

Try now!

👾 THE AI (R)EVOLUTION

The AI revolution is here, and it’s transforming the way we approach mobile games marketing.

From creating scroll-stopping ad creatives to optimizing user acquisition strategies, AI is the game-changer we can’t ignore

Top AI image generation models right now:

1. Google Nano Banana 2: A fast image model built for high-quality outputs, strong text rendering, and solid character consistency without slowing the workflow down.

🔗 https://gemini.google.com

2. GPT-Image 1.5: The model behind ChatGPT Images, with some of the best instruction following and editing performance right now. Especially strong at keeping lighting, composition, and facial details intact.

🔗 https://chatgpt.com/images

3. Google Nano Banana Pro: A more advanced version aimed at higher-end creative work, with better reasoning, multilingual text rendering, and high-resolution output.

🔗 https://gemini.google.com

4. Wan-2.7-Image: Alibaba’s latest image model focused on detailed control, portrait customization, and very strong multilingual text rendering.

🔗 https://wan.video

5. Reve v1.5: A model built for sharp, high-resolution images with strong detail retention and polished-looking outputs.

🔗 https://app.reve.com

6. Grok Imagine: xAI’s image model built for fast, high-resolution generation at scale, with a clear focus on speed and throughput.

🔗 https://grok.com/imagine

7. Flux 2 Max: Black Forest Labs’ top FLUX model, built for photorealism, strong prompt adherence, and reliable consistency across generations and edits.

🔗 https://bfl.ai/models/flux-2-max

8. Hunyuan Image 3.0: Tencent’s large-scale image model, positioned as a serious high-end contender for photorealistic and prompt-accurate generation.

🔗 https://hunyuan.tencent.com/image/zh

9. Seedream 4.5: ByteDance’s image model focused on cinematic visuals, stronger instruction following, better text rendering, and high-resolution generation and editing.

🔗 https://seed.bytedance.com/en/seedream4_5

10. Qwen-Image-2512: A newer Qwen image model with better text rendering, more natural textures, and more realistic human generation than previous versions.

🔗 https://chat.qwen.ai

🥦 Before you go

Ask Matej AI a question. It’s actually quite good!

We produced one more episode at 2.5 Gamers. Check it out now:

We started a new segment. Give it a go:

🍿 Fun starts in two & a half gamers Slack! Join now

Feel free to join here!

🧙 Thanks for reading this week’s newsletter!

Please do me a favor and share this with someone in your network who you think would benefit from the insights and brutal honesty. It would mean a lot to me!

Take care, everyone! See you next Monday!

👇 Reach out for feedback, questions, or say hi!

Give me some love (or hate), & Subscribe everywhere! ♥️

Playable Ads Trends in Mobile Games & Apps - February 2026

Matej Lancaric — Fri, 03 Apr 2026 08:04:14 GMT

For those who read only summaries, here are the playable trends February in a nutshell:

February - a month of the big guns of playable ads
- So, you want to know what the big guns do? You want to be like them, and last month’s Flambé wasn’t enough? Well, guess what - in February, one of the industry big shots said: “I am going ALL IN”. Need any hints? Kingdom, Royal, Match3, Kings, Kingdom of match3... I guess you already know:)

February – when Royal Kingdom flipped the switch

So, you want to know what the big guns do? You want to be like them and last month’s Flambé breakdown wasn’t enough? In February, one of the industry’s loudest royalty IPs basically said: “Screw it, I’m going ALL IN on playables.” Need hints? Kingdom, Royal, Match3, Kings, Kingdom of match3… yeah, that one.

Brutally Honest #206 by Matej Lancaric

Matej Lancaric — Thu, 02 Apr 2026 14:24:11 GMT

Hey, FAM! What’s up?

🥃 Drink talk

Yes, I was traveling for the last 2 weeks, but I am back in civilization. Lets goo!

As you know, I hate reporting. Yes, I hate making reports. Not the insights - I love the insights. I hate the mechanical act. If you run UA at any kind of scale, you probably know exactly what I mean. And you know that this is a giant WASTE of time. Thats why I did this:

🦩Tip of the week

Keep getting questions about the Applovin campaigns setup. So here you go:

Do not create 20 campaigns. Fragmentation is the enemy. AXON needs consolidated data to learn.

Improve your UA

Android & iOS

I usually start with the WW campaign anyway. Then you end up with this campaign structure:

iOS WW D7 optimization (AdROAS/BlendedROAS/IAP ROAS)

iOS WW D28 optimization (AdROAS/BlendedROAS/IAP ROAS)

GP WW D7 optimization (AdROAS/BlendedROAS/IAP ROAS)

GP WW D28 optimization (AdROAS/BlendedROAS/IAP ROAS)

Just don’t forget to add country-level TARGET ROAS so you control the traffic. (E.g your target ROAS on D28 is 120%, put 120% across all geos, but 100% target where you want to get more traffic - US, KR, DE, etc)

WITH CPM Billing

When you set your Bidding Strategy to “Spend budget evenly throughout the day”, you are on CPM billing.

How it works: Spend is paced, and the budget is capped. Based on what is written in the AXON tooltip, AXON will disregard strict ROAS targets and simply achieve the best ROAS possible for the set budget. But that is not usually true.

Give it some time to get proper learnings.. Usually it takes 2 weeks until you can see whats up …

This newsletter is brought to you by FastSpring:

Handling direct-to-consumer (D2C) payments and webstores for over 20 years.
Launch a D2C store in record time.
Includes Steer Safe™ for seamless and secure in-game steering to D2C checkout.
Powering top mobile publishers around the world, like Rovio, TechTree Games, + a LOT more
Fast-growing, profitable, and financially secure.
Headquartered in the US with offices from Amsterdam to Singapore to deliver local service on a global scale.
A no-drama D2C partner with real-world experience at scale, FastSpring is a partner you can trust with your players.

Stop managing payments, start scaling your game. Get a personalized D2C strategy session at fastspring.gg

🐊 Cornelius: The Suspicious Investi-gator

True detective-like exploration of creative trends and historical influences. Delving deep into the world of creatives to uncover hidden patterns and inspirations:

🚨𝗜 𝗪𝗔𝗦 𝗪𝗥𝗢𝗡𝗚!🚨 Top companies 𝗱𝗼𝗻'𝘁 produce 60 playables a month. They produce way more!👇

Royal Kingdom, Match villains, and Pixel Flow (yes, bringing this game again, I know, I know) are doing hundreds of playables per month.

𝗖𝗿𝗲𝗮𝘁𝗶𝘃𝗲 𝘃𝗼𝗹𝘂𝗺𝗲 𝗯𝗲𝗮𝘁𝘀 𝗲𝘃𝗲𝗿𝘆𝘁𝗵𝗶𝗻𝗴. Top studios are winning because they ship 10+ playables a week!

𝗛𝗼𝘄?

𝗦𝘁𝗮𝗿𝘁 𝘄𝗶𝘁𝗵 “𝗽𝘂𝗿𝗲 𝗴𝗮𝗺𝗲𝗽𝗹𝗮𝘆 𝗼𝗻𝗹𝘆” 𝗽𝗹𝗮𝘆𝗮𝗯𝗹𝗲𝘀
This is how to do it right: mirror the real experience as closely as possible. Use the actual camera angle, UI style, pacing, and difficulty curve from early levels, and highlight what truly differentiates your game – meta depth, build choices, boss patterns, or team synergies.

𝗧𝗲𝗺𝗽𝗹𝗮𝘁𝗲𝘀
The key to starting with templates is picking the right one. DUH! Jokes aside, you should pick a universal one, such as a 3x3 puzzle, character selection, video carousel, image carousel, spot-the-difference, or make your choice. Well, make your choice wisely! Badumtss!

Mini games
As mentioned previously in my playable trends, Flambé is a great example of how to start. 𝗦𝗽𝗼𝗶𝗹𝗲𝗿 𝗮𝗹𝗲𝗿𝘁: 𝟭𝟬 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁 𝗰𝗼𝗻𝗰𝗲𝗽𝘁𝘀:

Cutting

Customer Serving

Sorting

Threading

Cleaning up

Freezing families

Full screen board merge 2

Tables serving

Panic in the kitchen

Order taking

Get the playable bible now!

Yours truly, Cornelius.

Increase your creative win rate: https://payhip.com/b/tu4nk

💀 Dr. Doom is not amused

One of my industry friends amused me with his direct and honest TL;DRs, so we agreed to share it here, too! My friend remains anonymous as Dr.Doom!

“Layoffs continue and ppl are already proclaiming that its now worse than the big game industry crash, powered by the notorious E.T. video game flop in the 80s. Just wait when they will learn that games are made 5x faster and 10x cheaper in China this time. Say goodbye to the US video game industry 🍿😎” Sincerely, Dr.Doom (https://www.pcgamer.com/gaming-industry/the-industrys-in-a-really-horrible-place-brenda-romero-says-we-were-there-in-the-80s-for-the-crash-and-this-is-definitely-crashier/)

🎉 “Ha-ha” moment

Because life is too short to be serious.

Thank you for reading Brutally Honest by Matej Lancaric. This post is public so feel free to share it.

🎮 two & a half gamers new episode

🧶 𝗬𝗮𝗿𝗻 Loop: Knit Puzzle by 𝗦𝗽𝘆𝗸𝗲 𝗚𝗮𝗺𝗲𝘀: Huge hit making $300k/day nobody is talking about!👇

Yarn Flow was launched in December 2025. Already renamed to Yarn Loop: Kniz puzzle (I wonder why 😉). Already surpassing $150K/day in IAP revenue by late March, and that doesn’t even include ad revenue. 🤑

That’s serious velocity for a 4-month-old game.

This game is published under Combo Games 🤯. Most people stuill don’t realize this is a Spyke Games title. It’s published under Combo Games, which appears to serve as an experimental arm for testing new creative directions.

The largest Pixel Flow game 🦄 (recently acquired by Scopely) validated the ceiling. The second-largest? 𝗔𝗹𝘀𝗼 𝗧𝘂𝗿𝗸𝗶𝘀𝗵. That’s ecosystem depth 🇹🇷.

The next question is obvious: can Turkish studios continue to dominate the Flow genre?

Watch the full episode:

📺 Watch on YouTube

👕 Get our Merch: 25gamers.com/shop

🤑 Paid subs only

Sneak peek at what you get for your support:

🦄 Diary of a Dancing Unicorn

Today was a good day.

Hey, so this tool for no-code playables is super expensive. 600 EUR/month.

Wait, what?

Yeah, we can’t commit to something like this for a year. Its 5000 EUR/year

Do you realize that we live in a world where you pay 5,000 EUR for 1 playable?

Wait what?

Yeah?!

Oh, I didn’t know.

I am a bit surprised now.

Anyway, are there any other tools that we can explore?

Sure, here: X, Y, Z, but they are like 5x more expensive…

Omg, now the 600 EUR/month doesn’t sound bad ….

Today was a good day.

Share Brutally Honest by Matej Lancaric

🤺 Quote of the week

“Yeah, I’ve been monitoring it like a hawk 😂” - John, WhatsApp

Yarn Loop Review

🧞 This Week @ Playablemaker

“Sorry it took us a while, but after trying different platforms, talking with our finance department, and so on, we've decided to start building with you!”

Get a 20% discount for 3 months with code: DANCINGUNICORN

Try now!

👾 THE AI (R)EVOLUTION

The AI revolution is here, and it’s transforming the way we approach mobile games marketing.

From creating scroll-stopping ad creatives to optimizing user acquisition strategies, AI is the game-changer we can’t ignore

🥦 Before you go

Ask Matej AI a question. It’s actually quite good!

We produced one more episode at 2.5 Gamers. Check it out now:

We started a new segment. Give it a go:

🍿 Fun starts in two & a half gamers Slack! Join now

Feel free to join here!

🧙 Thanks for reading this week’s newsletter!

Please do me a favor and share this with someone in your network who you think would benefit from the insights and brutal honesty. It would mean a lot to me!

Take care, everyone! See you next Monday!

👇 Reach out for feedback, questions, or say hi!

Give me some love (or hate), & Subscribe everywhere! ♥️

I Automated my weekly UA reporting using AppsFlyer MCP: Here’s how!

Matej Lancaric — Wed, 25 Mar 2026 14:01:04 GMT

Every week, I do the same thing. I open AppsFlyer. Then I open it again for a different client. Then again. I’m currently running UA for around 10 companies, which means 10 separate accounts, 10 dashboards, and 10 sets of numbers to pull before I can say anything useful to anyone.

I hate reporting. Yes, I hate making reports. Not the insights - I love th…

Brutally Honest #205 by Matej Lancaric

Matej Lancaric — Wed, 18 Mar 2026 21:26:31 GMT

Hey, FAM! What’s up?

🥃 Drink talk

Launching an Applovin campaign? Here you go:

🦩Tip of the week

Lol, Andromeda here, Andromeda there.. Still.. Well, just iterate your creatives and put them live. And test!

Yes. Even a simple headline on top of your best-performing creative can work. But test those as well. This one is performing better then the other 4.

👀 Never Ending Creative trends #3

Matej Lancaric — Tue, 17 Mar 2026 23:18:32 GMT

Welcome to Never Ending Creatives - a fast, structured breakdown of what’s performing across today’s ads and why it matters. We have strategy games, idle clickers, hybrid casual, puzzle, and so much more!

If you are new here, we recommend starting with our methodology section to get the most value from these insights. Brought to you by PvX Partners & brutally honest myself!

UA Creative Trends

February delivered a clear mandate for idle-clicker UA: stop summarizing gameplay, start telling a story. This month’s data unequivocally points to a powerful shift towards emotionally charged, highly engaging narratives that put players directly into the action. Forget generic highlights; we’re seeing huge wins from content that challenges, frustrates, and ultimately rewards.

Bottom Line
What’s Next? Interactive Storytelling Your creatives should be felt, not just seen. Aggressively invest in looping arcs, lean into raw emotions, and explore interactive elements like choice branches and puzzle challenges to drive deeper connection and installs.

Increase creative win rate by 50%

This past month underscored the importance of connecting directly with the viewer. Creatives increasingly used techniques like the ad speaking directly to you or showing parallel views simultaneously to build immersion. The focus also intensified on competitive progression, highlighting the drive to build or farm faster than rivals and showcasing scenarios where players either win big or lose everything. Emotionally, ads leveraged everything from player failures and frustrations to inspiring victories, moments of envy, and even clever humor to keep audiences hooked.

Read the monthly report now

Extra example only for you:

🧠 MVP UA Template: I want to launch my first Applovin campaign. Now what?

Matej Lancaric — Sun, 15 Mar 2026 19:04:37 GMT

Hey there, UA manager, company founder, or indie developer! I know the feeling. Today, I am going to talk about the best growth engine out there. Axon by Applovin! Buckle up..

Let’s be honest. If Meta is a chaotic playground and Google Ads is a black box, Applovin (ehm ehm, AXON) feels like a fully autonomous predator drone. Check out also the other highly popular MVP UA guides:

You don’t steer it. You don’t aim it. You just give it a target, feed it creatives, and get out of the way. With the AXON 2.0 algorithm, manual optimization is basically dead. If you try to oversegment or micromanage it, you will get crushed by competitors who let the machine do its job.

The truth is, Applovin is an algorithm-first, creative-second network. But that does not mean creative is not crucial. It matters a lot, and playable ads are extremely important. Don’t even start Applovin campaigns before reading the Playable bible:

AXON is more than an ad engine. Operating at a massive scale, over $11B in annual media spend, it functions as a closed-loop system. Its deep integration with MAX mediation provides exclusive access to high-intent mobile users (especially casual gamers) through rewarded video and full-screen interstitials. Interstitials are quite important!

Get Mobile UA 101: Scaling guide

The model doesn’t just optimize for installs, it forecasts entire user value curves: engagement, conversions, purchase behavior, and LTV. Its accuracy enables real-time capital allocation with surgical precision.

Here is your Minimum Viable Campaign Setup (MVCS) for Applovin in 2026. Simple, structured, and actually designed for how AXON works today:

🦄 "I AM MATEJ NOW" E-books Bundle release

Matej Lancaric — Wed, 11 Mar 2026 18:25:47 GMT

Everything I know from 12 years of experience about User Acquisition and Creatives is here in this bundle. You will immediately improve your UA operations after reading all my e-books. And, I will pretty much lose my job afterwards. Still fucking worth it..

Buy NOW!

Well, yes. Go for it and improve your UA & creative skills!

Become Matej Now

Brutally Honest #204 by Matej Lancaric

Matej Lancaric — Mon, 09 Mar 2026 18:46:32 GMT

Hey, FAM! What’s up?

🥃 Drink talk

Here is the thing, I have been pushing content to more digestable way: E-Books! New one is out!

Oh yeah, coming back to writing global launch case studies! Check out the latest:

🦩Tip of the week

Creative testing checklist

Define the test parameters in writing:

Hypothesis stated (what are you testing and why?)
Creative briefs …

Brutally Honest #203 by Matej Lancaric

Matej Lancaric — Wed, 04 Mar 2026 18:26:56 GMT

Hey, FAM! What’s up?

🥃 Drink talk

Yes, I just released a Playable Bible! Get it while it’s hot.

🦩Tip of the week

Emotions .

🧶 𝗬𝗮𝗿𝗻 𝗙𝗹𝗼𝘄 by 𝗦𝗽𝘆𝗸𝗲 𝗚𝗮𝗺𝗲𝘀: Huge $4mil hit nobody is talking about!

Matej Lancaric — Tue, 03 Mar 2026 18:35:35 GMT

It’s been a while since I published a global launch case study. So you can bet this is the real deal! As usual, expect UA & creative deep dive! Yarn Flow was launched in December 2025. Already surpassing $100K/day in IAP revenue by late February, and that doesn’t even include ad revenue. 🤑

I calculated a serious $4 mil in revenue since December.