❌Why most Soft Launches FAIL & how to avoid the trap

The 3-Stage Bluepring for soft Launch success in 2025

If you think a quick one-month soft launch is enough, think again. Most mobile games that rush or skip a proper soft launch end up face-planting at global release. In my own blunt words: “If you have a one-month soft launch, you can't really test shit. [You] can't really assess anything, and you will get… really partial data.”

A soft launch needs 3–6 months of hard data gathering to validate your game’s product-market fit, retention, monetization, and UA strategies. Consider it your last chance to minimize risk – or else your global launch might just be a global flop.

In this BS guide (inspired by my Soft Launch Bible), we’ll break down the 3-stage soft launch framework – Tech Test, Retention, Monetization – along with the geo rollout, KPIs, creative testing cadence, and ultimate kill-or-scale decisions at each step. Strap in, because we’re not sugarcoating anything.

By the end, you’ll know exactly how to soft launch your game in 2025 like your studio’s survival depends on it (because honestly, it does).

Why a Soft Launch must last 3–6 Months (Not 1)

Before diving into stages, let’s get one thing straight: a meaningful soft launch takes time. You need enough runway to observe retention curves and LTV (lifetime value) over weeks and months, not just a few days. I insist on a soft launch duration of at least 3 months, ideally up to 6, to capture metrics beyond Day 7: “I really want to know how the retention curve looks… not only day one, day three or day seven, but also day 14, day 30… ideally day 60, 90”. Early retention can fool you; a game that looks decent at D7 might fall off a cliff by D30. Without a longer test, you’ll never see it coming.

Equally important is mapping out the LTV curve and ROAS evolution. You might hit a positive Day 7 ROAS, but what about Day 30 or 60? If you call it early, you won’t know if players monetize long-term or just spike early. A soft launch gives your team time to implement improvements, run live ops events, and then observe how those moves affect LTV and retention over multiple cycles. It’s an iterative process, not a one-and-done sprint.

Ideal benchmark KPIs for a globally launchable game are famously “40/20/10” – roughly 40% Day-1, 20% Day-7, 10% Day-30 retention. In other words, a healthy game retains about half its players a week out, and half of those again by a month. Many top-grossing titles follow this pattern. In fact, I often shoots for 50% D1 if he can. He also emphasizes that by the end of soft launch your LTV should exceed your CPI – basic economics: if LTV isn’t higher than cost per install, you’re not ready to scale.

That said, “40/20/10” is a guideline, not gospel. Different genres have different retention profiles. I point out that puzzle RPG hits like Empires & Puzzles soft-launched around ~40% D1 and ~19% D7, whereas Puzzle & Survival saw ~31% D1 and ~12% D7 – yet both games make millions. The key is not to slavishly chase a magic number but to ensure your retention drop-off is proportionate and the curve “looks healthy.” If you see D1 50% but D7 3%, you’ve got a serious problem in your mid-term retention funnel.

In my words: “If you have day one 50%… but day seven is 3%, that sucks… there’s all these problems in the onboarding, in the funnel… You need to fix that.” A good soft launch will surface these issues early, giving you a chance to fix UX, tutorial, difficulty spikes, etc., before you pour money into global UA.

The Three Stages of Soft Launch: Tech, Retention, Monetization

Soft launch isn’t a monolithic phase – it’s broken into three distinct stages, each with its own purpose, metrics, and “gates” you must clear to proceed.

Stage 1 is the Tech Test, focusing on technical stability and analytics setup.

Stage 2 is the Retention test, honing the core gameplay and early retention loop.

Stage 3 is Monetization, where you validate that the economics make sense with real spend. Let’s unpack each stage and what you should accomplish before moving forward.

Stage 1: Tech Test – “Do our pipes leak?”

The first stage is all about tech stability and data integrity. Before worrying about whether players like your game, you must ensure the damn thing works. That means no crash-fest, no broken analytics, and no server meltdown. Pick a cheap, low-stakes market to test in – think “Tier 4” countries like the Philippines, Mexico, Brazil, Indonesia, or India. These regions offer inexpensive installs at low volume, so you can get a few hundred or a thousand players without burning cash. Why cheap geos? Because Stage 1 is not about performance metrics; it’s a dry run for your tech. As I put it, “tech test is only and only for testing the tracking and the technical stability of the game. Nothing else.”

In this phase, instrumentation is king. Make sure your tracking SDKs, attribution (MMP) integration, backend events, and analytics pipelines are firing on all cylinders. A missing event or a data mismatch between Firebase and your dashboard can completely derail your ability to measure success. “All of these small technical mistakes can destroy basically the whole soft launch,” I warn. Double-check that crash reporting is in place and monitor your crash rate and ANR (app not responding) rate closely. Google Play will actually punish your app’s visibility if it detects high crash/ANR rates, so stability isn’t just a player experience issue; it affects discoverability, too.

During the tech test, keep the user acquisition very limited – just enough installs to spot issues. I might run a small Facebook or Google campaign to gather a few hundred players (even a couple hundred can surface major crashes or tutorial blockers). You’re not optimizing for KPIs here; you’re verifying that when you do hit the gas, the wheels won’t fall off. If something critical breaks (e.g. players stuck on a loading screen, purchase events not logging), fix it fast and restart the tech test until your core instrumentation is bulletproof.

One note of caution: when testing in places like the Philippines, don’t panic about low retention numbers – they will be artificially low due to device and network quality. I have seen many teams freak out because their Day-1 retention was 20% in the Philippines instead of 35%+ like they’d expect in a Western country.

Relax – that’s normal for a low-cost region. In fact, it can be so extreme that I sometimes digs deeper into device breakdowns. If you find that many players on ancient devices struggle to play your game, consider excluding low-end devices from your UA targeting.

For one project, I did exactly this – cutting out users with less than 2–3 GB RAM or old OS versions – and saw a huge jump in apparent retention without changing the game at all: “Instead of seeing 20–25% day one retention in Philippines, I was able to see 40% right away and I didn't change anything in the game itself.” The lesson: optimize your test audience to match the technical needs of your app. After a week or two in Stage 1, you should have confidence that your game is stable, events are tracking, and you can trust the data. Now the real fun begins.

Stage 2: Retention – “Do players love it enough to stick around?”

Stage 2 shifts the focus to product/market fit and retention. This is where you answer the million-dollar question: do players actually enjoy your core loop and come back the next day? Now that your tech is stable, you move to higher-quality markets to get a read on genuine retention metrics. I recommend going for Tier 2 countries here – not the absolute cheapest, but still cost-efficient and representative of your target audience. Good examples include Poland, Brazil, or the Netherlands (for a mid-core game targeting a male audience, I often use those). These regions give a decent mix of engaged users without the ultra-high CPIs of Tier 1 markets. As he notes, Poland is one of his favorite soft launch countries – an “emerging EU market” with a high gamer population and good monetization potential.

At this stage, you’re aiming to hit the retention KPI targets that you believe are necessary for global launch. For example, maybe you set a goal of 40% D1 and 15% D7 for your genre. Now you iterate on the game to get there. Measure everything: tutorial completion, level drop-off, session lengths, and user feedback. If D1 is low, dig into where players churn in the first session (maybe a confusing tutorial or difficulty spike). If D7 is the issue, perhaps mid-game content is lacking. The team should continuously push updates and improve the first-time user experience (FTUE) during this stage. You might go through multiple build iterations – each time, gather another cohort of users and see if metrics improve.

Keep in mind that statistical significance matters. Don’t celebrate or despair over a cohort of 30 players. Aim for at least a few hundred players per build before judging a retention metric. I suggest something on the order of 200+ new users per day to get stable Day-1 retention figures, and more for Day-7+. Yes, this means spending more on UA, but it’s better than making decisions on noise. By the end of Stage 2, you ideally want to see retention trending near your targets (or confidently improvable to targets). If after several iterations you’re still way off – e.g., D1 plateaued at 20% with no upward movement – that’s a major red flag. We’ll talk kill decisions soon, but suffice to say, Stage 2 is where many games die (and rightly so) if players just don’t stick.

Geo sequencing tip: You can sequence markets within Stage 2 as well. Often teams start in one country (say Poland), then expand to a couple more (e.g., add Brazil, Denmark or the Netherlands) to see how retention holds across geos. Also, at some point in Stage 2, you’ll likely enable iOS (if you intend to launch on iOS). Usually, you start on Android (easier to iterate and no review process), but once the Android retention looks good, bring in iOS in one of your test countries and make sure the metrics line up (and sort out the whole SKAN conversion value schema on iOS).

UA Channel rollout: Early in Stage 2, stick to one or two reliable channels (I like to start with Facebook, since it’s broad and efficient). Once you have a stable build or two, add Google UAC to reach a different mix of players. By the second or third iteration, you might throw in a wildcard channel like TikTok or even Unity Ads to diversify traffic. There’s a balance to strike: you want enough channels to get volume and not overfarm one source, but not so many that you can’t pinpoint which changes in metrics are due to game improvements versus UA changes.

“I usually start with Facebook, add Google, and as a wild card, TikTok [in retention stage].”

Perhaps more importantly, optimize your campaign objectives as the test progresses. In the first weeks, you’ll run basic Mobile App Install campaigns (optimized for installs) to get users in the door. But these will bring a lot of low-quality players. After a couple of iterations, I switched to App Event Optimized (AEO) campaigns, optimizing for deeper funnel events (e.g. reach level 5 or complete tutorial). This naturally finds players who are more engaged, boosting your retention numbers.

Expect a bump in D1/D7 once you move to AEO targeting quality users. Finally, by the end of Stage 2, you might even test purchase optimization on a small scale (if your game has IAP) – though big spend on purchase campaigns usually comes in Stage 3. The point is to simulate how your UA will behave at scale: “as soon as we move towards app event optimized campaigns, we will see way better retention… it’s a better targeting option”.

Speaking of creatives, this is where you also establish your creative testing cadence. Early soft launch is not an excuse to run one old video ad for months and call it a day. Even in a handful of test geos, creative fatigue is real. You’re hitting the same audience over and over in, say, the Philippines or Poland – they will get banner blindness unless you refresh

.

I recommend aiming for roughly 10–15 new video ads per month during soft launch, plus a few playable ads once you have confidence in the gameplay. Keep the concepts gameplay-oriented – don’t run misleading “fake ads” just to juice click-throughs.

That will only bring players who churn immediately because the game isn’t what they expected. As I warned, “if you use fake ads, it’s going to tank your retention data… it’s misleading”. Instead, test a variety of gameplay highlights, short video cuts, maybe some 3D character clips – different hooks that still reflect the actual game. By doing this, you’ll not only avoid poisoning your metrics, but you’ll also learn what themes resonate with your real audience. The goal is to exit Stage 2 with a few proven ads and a clear idea of your CPI in a mid-tier market at target retention. This will set the stage for scaling up in the next phase.

Stage 3: Monetization – “Will this game actually make money?”

If you’ve made it to Stage 3, give your team a small pat on the back. It means your retention KPIs are on track and you haven’t killed the project yet. But now comes the moment of truth: can this thing monetize effectively and hit ROI? Stage 3 focuses on revenue, LTV, and proving the business case for a global launch. The focus shifts to testing your monetization model (IAP, ads, or hybrid) in conditions that closely mimic a global launch.

First, you’ll open up in a couple of Tier-1 markets – typically English-speaking, high-GDP countries where players spend well. Many studios have historically used Canada or Australia, but I actually prefer the UK and Germany for this stage. Why? Canada and Australia are fine, but they’re relatively small – it’s easy to saturate them and drive up CPI quickly, plus their revenue profile might not scale linearly.

The UK and Germany, on the other hand, offer larger populations and strong purchasing power, and they can serve as excellent proxies for US/EU performance. (You can always include Canada/Australia too for extra data points, of course – the more Tier-1 coverage, the better your model generalizes.) By the end, your Stage 3 geo mix might be something like UK, Germany, Canada, Australia, and maybe a Nordic or two.

Now, crank up the UA spend to a volume that approaches a launch scenario. You should also unlock all major ad networks now – Facebook, Google, TikTok should already be in play from Stage 2, but now add the heavyweights of performance marketing: AppLovin, IronSource (LevelPlay), Unity Ads, DSPs like Moloco or Mintegral, etc.

This gives you a broad spread to see how ROAS and LTV hold up across sources. Expect CPIs to rise in Tier-1 (maybe 2- 3x what you paid in Stage 2’s geos) – that’s normal. Higher CPI comes with typically higher-quality users. In fact, you’ll likely switch your campaigns fully to purchase optimization or even value-optimization at this point. No more proxy events – you want to acquire payers. Yes, those campaigns cost more, but Stage 3 is about measuring real revenue potential.

“You go for purchase [campaigns] – higher quality, even higher CPI… value optimized campaigns [can drive great performance but with super high CPI as well],” I note.

The mix might look like: run purchase-optimized ads on Facebook/Google, maybe value-opt on one channel to see LTV ceiling, continue some AEO here and there for comparison.

Your key metrics in Stage 3 are D1/D7 ARPU, conversion rate, and long-term ROAS/LTV projections. Essentially, can you get to LTV > CPI with a reasonable payback period?

I often use Day 30 or Day 60 ROAS as a benchmark for soft launch success – for example, if you can hit 20-30% D30 ROAS in soft launch and see solid retention, you might extrapolate to 80-100% by D180 (depending on the genre).

Every game’s targets will differ, but now is when you find out if your monetization design (pricing, economy, ad placements, etc.) holds water. If not, better to find out now than after burning marketing budget globally.

ne more thing to layer in here is Live Ops and monetization features. By Stage 3, you should be testing things like events, sales, bonus offers – the stuff that boosts revenue. I actually recommend trying to have some live ops “mini-events” and special offers running by the end of soft launch. For instance, schedule a weekend event or introduce a starter pack sale. See how it moves the needle on revenue and engagement. This helps simulate a post-launch content cadence and can goose your LTV to where it needs to be. It’s also a great way to discover what events might drive spikes or where your economy could break under pressure.

If all goes well, by the end of Stage 3 you’ll have a dashboard full of green lights: retention cohorts looking healthy, LTV projected above CPI, crashes near zero, and UA strategies proven out across channels. In short, you have a potential hit on your hands. But if things don’t go well… it might be time to have the tough conversation.

Know When to Pull the Plug (Kill Triggers)

Not every game is destined to make it to global launch – and that’s okay. The whole point of a rigorous soft launch is to fail fast on games that aren’t meeting their targets, before you waste millions on a worldwide release. The worst mistake a studio can make is falling prey to the sunk-cost fallacy and pushing out a game that showed clear warning signs in soft launch.

MY advice? Be brutally honest with the data. If the game is **“not hitting the goals that you set, and you’re not moving the needle anywhere, kill it – kill it with fire”*.

It may sound harsh, but as he adds, “I know it’s hard, but you should definitely do it, because if you kill the game, you can actually still save the company and then pivot to something else. Most people don’t do it, and then the company’s gone.”

So what are the specific kill triggers to watch? It will vary by studio, but a few likely red lines:

• Retention below viable thresholds – If, despite all Stage 2 efforts, your D1 is languishing in the 20s or D7 in single digits, it’s unlikely to magically fix itself. Also, look at the retention ratio (D7 vs D1). I suggest that if your D7 is much below ~30–40% of D1, the curve probably isn’t healthy enough to scale. For example, D1 30% and D7 10% (33% ratio) might be borderline; D1 30% and D7 5% (17% ratio) is almost certainly a kill signal. There are exceptions by genre, but broadly, a steep early drop-off = no go.

• LTV < CPI – By the end of Stage 3, you should have a solid estimate of LTV. If you can’t find a scenario where LTV will realistically exceed CPI (even with optimizations, live ops, better creatives, etc.), then the game’s economics don’t work. Maybe you have great retention but very low monetization (common in purely ad-monetized games or ultra-casual). Or maybe some players pay, but overall conversion is too low. If the numbers don’t add up, it's better to cut losses. You’re not running a charity; the game has to make money.

• No improving trend – Soft launch is about iteration. If you’ve done multiple updates and your key metrics flatlined or only inched up, ask yourself if you’ve hit a local maximum. Sometimes teams keep hoping “one more feature” will suddenly boost retention 10 points – a rare miracle. Set a limit for the number of iterations or the time you’ll spend. If you’ve thrown your best efforts at the game for 6 months and it’s still off-target, that’s a sign. (Conversely, if metrics are steadily improving and just haven’t hit the goal yet, you might extend the soft launch a bit. Use judgment, but don’t let hopeless projects drag on due to optimism alone.)

• Comparative benchmarks – Look at similar games’ soft launch trajectories. If by month 3 you have half the retention or revenue that comparable titles had at the same point, you’re likely too far behind. I often research top games in the genre during soft launch to know what “great” looks like. If you’re nowhere close, you either pivot the game’s design or kill it.

It’s painful to kill a game into which you’ve poured months of work. But that’s a sunk cost you cannot recover. What you can save is the remaining runway and resources to start a new project that might fare better. The mobile graveyard is littered with studios that bet the farm on a game with clear early signs of failure. Don’t join them. As I bluntly put it, the choice at the end of a soft launch is usually go global or go home – “most importantly, you go live or you kill the game. Pretty much that’s as easy as that.”

Soft Launch like you mean it

You’ve made it through this gauntlet of a guide – congratulations! By now, it should be clear that soft launching a game in 2025 is not for the timid or the lazy. It takes careful planning, relentless analysis, and the guts to make hard decisions. The upside? It dramatically increases your chances of global success and could literally save your studio from a costly failure.

So if you’re a growth lead or product manager gearing up for a soft launch, here’s my challenge to you: embrace the process fully. Set up those stage gates and stick to them. Hold your team (and yourself) accountable to the data. Schedule your 3-6 month roadmap with deliberate tech tests, retention tuning, and monetization experiments. Invest in creative testing early so you’re not scrambling with unproven ads on launch day. And steel yourself for the possibility that the best outcome might be to walk away and try something else – failing fast is better than failing big.

In 2025’s hyper-competitive market, you don’t get a second chance to make a first impression on players. A well-executed soft launch ensures that when you do hit that “Go Global” button, you know your game is ready to roll. As my experience shows, a soft launch isn’t just a checkpoint – it’s a make-or-break process that separates the winners from the wannabes. So launch smart, iterate hard, and don’t be afraid to kill your darlings. Your future self (and your finance team) will thank you.

Now go forth and soft launch like you mean it. Your game’s success might depend on it. Happy soft launching! 🚀