In mobile games, most monetization decisions happen earlier than anyone would like. The data is thin, the stakes are high, and waiting for perfect answers is rarely an option.

Euan Cowen, Director of Growth at Super Banana, and ex-Triple Dot Studios, spends his time making those calls. In this interview, he breaks down how hybrid monetisation in puzzle games is assessed early, scaled responsibly, and adjusted as real behaviour starts to show.

 

In puzzle games with hybrid monetization, what early-funnel signals do you rely on to forecast long-term ad-mon revenue before D7?

In puzzle, the goal’s often to form a reliable view of your cohorts as early as possible. There’s an inherent trade-off between speed and accuracy, but with a strong framework, you can make confident decisions around long-term monetisation expectations. 

The signals to prioritise early are simple: revenue and revenue growth, with the earliest absolute values often mattering less than the trajectory. A D0 or D1 ARPU might look soft, but if the growth curve is strong, engagement is high, and ad impressions and CPMs are healthy, then long-term monetisation can far outperform initial expectations. It’s the shape of that curve, not the starting point, that ultimately drives lifetime value.

When forecasting, be smart with your data granularity. App-level data can be misleading unless you understand your UA traffic mix; channel- or campaign-level data is more actionable but comes with smaller sample sizes. I suggest building a low-level model that rolls up. This captures nuance while creating a more malleable and adaptive framework. How “low-level” your model can and should be is informed by the data. Each team must decide for itself.

As for when a cohort is “mature” enough to forecast from, that hinges entirely on the reliability of your historical data. In some cases, strong D0→D1 revenue growth might give you enough conviction to act; more often, you’ll want to look at D1→D3 and D1→D7, while continuing to monitor everything beyond. What really matters is understanding your own variance patterns, the behaviour of your UA channels, and the revenue multiples you typically see as cohorts develop.

Hybrid monetisation adds a second layer of complexity. You’re no longer evaluating a single revenue behaviour, but the interaction between IAA and IAP dynamics. These patterns often vary significantly by platform, source, geo, optimisation type, and campaign objective. Your aggregate split might look like a clean 50:50, but that can hide vastly different compositions across traffic types.

Within your modelling, it’s advisable to start by looking at these components separately, using the principles mentioned earlier. Once that’s done, you can then start layering in the hybrid context:

A) Track the monetisation split early and understand how it evolves. Are IAA and IAP growing in parallel, diverging, or converging? 

B) Understand how your product is segmenting users, whether by design or incidentally. Difficulty profiles, bespoke IAP offers, IAA exposure, level-flow pacing (and more) all play a role.

Ultimately, the tighter the alignment between ad-mon, UA, and product, the earlier and more accurately you can forecast long-term revenue. Early signals matter, but only when interpreted within the full context of how your game attracts, segments, and monetises its players.

How do you balance ad-density with level-flow pacing to avoid suppressing retention in casual puzzle titles?

Before optimising this trade-off, the first question every publisher should ask is: what are we trying to achieve? The puzzle space is vast, and both retention curves and viable ad-density strategies vary dramatically across subgenres. Are you optimising for long-term LTV or faster payback? What kind of user experience are you creating? Which UA channels are, or are expected to, drive your traffic? Getting clear on these fundamentals should come before rushing into ad setup.

It’s fair to say that increasing ad frequency can hurt retention, but that statement lacks refinement. If you have banners from D0, interstitials at every turn, and rewarded placements saturating the UI, players will churn quickly. But no ads at all means leaving revenue on the table. The challenge is finding the sweet spot. If you’re just starting out, looking to the market for inspiration is a safe bet.

A smart ad setup considers timing, placement, invasiveness, and cadence. User tolerance isn’t one-size-fits-all, so even in a simple configuration, decisions around ad-type introduction, cooldowns, geo segmentation, and rewarded value exchange should be intentional. Get these foundations right and you can monetise in a way that aligns with user tolerance, expectations, and intent.

To understand how much you can bend without breaking, A/B testing is essential. A solid experimentation framework (a topic in its own right) allows you to quantify exactly what your audience will bear before it becomes too much. Tolerance varies significantly across platforms, countries, acquisition channels, and cohort compositions. What works for one segment may be unacceptable for another.

When scaling 9-figure UA budgets, which monetization-linked cohort behaviours most reliably predict when a channel can be profitably scaled?

Referring back to Q1, the starting point is still revenue, revenue growth, and how that curve develops across cohort days. Those signals sit at the core of any prediction. Alongside that, understanding user-base composition is critical. Modelling from a granular level up, rather than relying on blended averages, helps clarify how your IAA versus IAP mix influences cohort behaviour.

But the UA perspective isn’t just about maximising LTV. ROAS rules, therefore the relationship between LTV and CPI becomes critical. But, the simplest indicator of profitable scale potential is existing profitability. If cohorts are meeting or exceeding your target ROI (LTV/CPI) within your breakeven window, you usually have room to scale. Usually matters here. Some channels run into inventory limits or inelastic supply quickly and require caution. Others improve with scale, meaning you may sometimes take a bet on scaling even when CPI initially exceeds LTV. Knowing the channel, forecasting LTV, and using both together should guide decision-making. 

When assessing a channel, the first step is understanding the traffic and campaign types. In the best-case scenario, a mature and stable channel, CPI, ARPU, revenue growth, RPM, RPM decay, ad frequency, payer rate, retention, and playtime are all relatively consistent. When that holds true, LTV forecasting becomes straightforward.

Fundamentally, most monetisation-linked cohort behaviours are already reflected in the ARPU and revenue-growth signals used for forecasting. However, there are cases where it pays to dig deeper. ARPU is cumulative, while revenue growth is directional, and both require some level of maturity to be reliable. Overlaying supporting metrics, such as RPMs, can provide additional context. For example, a step-change in RPMs may suggest stronger long-term returns than initially forecasted for existing cohorts. This is especially relevant for the seasonal period we’re currently in the middle of.

Without going metric by metric, it’s important to consider what changes in these signals imply for profitability and scaling decisions. Understanding the relationships between UA and monetisation KPIs is key. Shifts in monetisation mix, such as changes in payer rate or rewarded, interstitial, and banner balance, can produce meaningfully different LTV curves. 

If you’re testing a new channel, you won’t have perfect data early on. In those cases, smart extrapolation matters. Consider what kind of traffic the channel is likely to bring, and which existing sources behave most similarly. Using proxy data from comparable channels helps fill early gaps and avoids overreacting to immature cohorts.

Ultimately, profitable scaling comes from combining monetisation behaviour, cost efficiency, and channel-specific dynamics. When those line up, scaling becomes a measured decision rather than a gamble.

From a startup perspective, what’s the core playbook small studios should follow to design & launch an effective ad monetization setup?

As with earlier questions, the starting point is strategy. Before touching tooling or placements, small studios need to be clear on what they’re optimising for. Are you trying to break even quickly, maximise long-term LTV, or support a hybrid UA strategy? Those decisions should inform everything that follows.

Once that’s defined, the playbook is about making a few high-leverage decisions early, without over-engineering.

The first is mediation selection. This is one of the most important choices you’ll make, and one that’s costly to reverse. In the casual puzzle space, most studios will naturally gravitate toward AppLovin MAX, which means leaning primarily on header bidding rather than traditional waterfalls. That choice has implications across both monetisation and UA, including access to AppLovin’s ad-ROAS (and hybrid) optimised campaigns. Whatever you choose, it’s important to understand the trade-offs, talk to peers, and make an informed decision rather than defaulting blindly.

Next comes ad format selection and initial configuration. Start simple. Focus on a small number of well-placed formats, typically rewarded video first, then interstitials, then banners and others, and be deliberate about when and where they appear. Early parameters like ad introduction timing, cooldowns, refresh rates (banners) and skip behaviour don’t need to be perfect, but they should be intentional and aligned with the experience you want to create. Mediation support for cross ad-type cooldowns is often limited, so this is an area where developer involvement and lightweight custom tooling can make a real difference.

From there, partner selection and auction setup matter more than many teams realise. Even early on, ensuring a competitive auction and filtering out low-quality demand can materially impact both revenue and user experience. Demand partners have different strengths across platforms and ad formats, so broad onboarding helps mitigate blind spots. This also includes getting the basics right, such as app-ads.txt and consent flows, so demand can compete fairly and compliantly from day one. Publications like the Appsflyer Performance Index are a good proxy for the demand volumes of partners.

Finally, while startups don’t need heavy infrastructure upfront, they should plan for iteration. A basic testing mindset, some lightweight controls, and a rough roadmap for how monetisation will evolve as data matures can help avoid painful rework later. You don’t need everything on day one, but you do need to know where you’re heading.

In short, an effective startup ad-monetisation setup isn’t about complexity. It’s about making a small number of good decisions early, staying aligned with product and UA goals, and leaving yourself room to learn and adapt as real data comes in.

What monetization levers do you consider most controllable post-launch for lifting LTV without materially changing core gameplay?

If you’re set up with real-time bidding, like on Applovin MAX, partner selection and auction health become your primary non-intrusive levers. From a user’s perspective, it makes little difference which network serves the ad, but for a publisher, ensuring strong competition in the auction can materially lift CPMs. Regularly reviewing partners, onboarding high-demand buyers, and keeping bidding pressure high is one of the lowest-risk ways to improve ad LTV.

Within bidding setups, bid floor management is especially powerful. Setting sensible CPM floors ensures low-value impressions simply don’t show, helping filter out weaker demand and often improving both monetisation efficiency and user experience, without material changes to core gameplay. Most platforms allow geo-tier-level floors, and with more backend support you can introduce more dynamic approaches.

If you’re running a waterfall-based setup, control shifts toward waterfall structure and ordering. Basic levers here include optimising waterfall depth, segmenting waterfalls by geo or platform, and ensuring your highest-value demand gets first look at impressions. Even small tweaks to ordering or segmentation can generate meaningful LTV gains over time.

Across both setups, rewarded video placements remain one of the safest levers to pull. Unlike interstitials or banners, rewarded ads can be deeply integrated into gameplay through revives, boosters, retries, and multipliers without disrupting flow. The key is that the value exchange feels natural. When rewarded ads align with player intent, you can often increase engagement and revenue without materially impacting retention.

My final point is a slight cheat answer, but it’s UA traffic quality. While this doesn’t improve LTV at a user level, acquiring higher-LTV users through premium UA channels will lift blended LTV. This typically comes at a CPI cost and may not always be the optimal UA strategy, but it’s worth calling out for teams optimising toward headline LTV metrics for fundraising, M&A, or external benchmarking. Value- or ROAS-optimised campaigns in high-ARPU markets can shift that number quickly.

How do you evaluate whether rewarded video placements are enhancing, or cannibalising, your in-app purchase revenue in puzzle games?

At its core, like so many in mobile gaming, this question is about the counterfactual. You’re trying to understand what would have happened to in-app purchase behaviour if rewarded video had not been present. In practice, the most reliable way to answer that is through controlled experimentation.

Where possible, A/B testing is the best approach. By isolating rewarded placements for a test cohort and comparing key metrics such as IAP conversion, ARPPU, total revenue, retention, and engagement against a control group, you can determine whether rewarded video is genuinely incremental or whether it’s shifting value away from purchases. It’s important to evaluate total LTV rather than individual revenue streams in isolation, as some degree of IAP suppression may still be acceptable if overall value increases.

If you don’t yet have the framework or scale to run clean A/B tests, a secondary approach is segmented rollout. Introducing rewarded placements to a specific user segment, while keeping traffic sources and cohort composition as consistent as possible, allows you to compare performance against a historical baseline. This method is less precise, but when interpreted carefully it can still provide useful directional insight.

If neither of those options is feasible, then judgement and product instinct become more important. In those cases, it helps to critically assess the value exchange you’re offering. Does the rewarded outcome meaningfully replace a purchase, or is it more likely to engage users who were never going to convert anyway? From a player’s perspective, the exchange should feel reasonable. Unless you’re deliberately applying price anchoring tactics, disproportionately generous rewards are more likely to cannibalise IAP.

Benchmarking can also help anchor decisions. Understanding typical purchase rates, rewarded video engagement, and IAA versus IAP revenue splits within your genre and target markets provides useful context. If rewarded engagement is unusually high while payer conversion declines relative to comparable titles, that’s often a signal worth investigating.

Ultimately, rewarded video isn’t inherently positive or negative for IAP revenue. Its impact depends on placement design, reward value, and audience composition. The goal is to ensure rewarded ads expand the monetisation funnel, rather than simply shifting value from one revenue stream to another.

In markets with lower ARPU, which monetization experiments typically deliver the highest lift without harming UA efficiency?

In lower-ARPU markets, it’s worth looking beyond user behaviour and headline ARPUs to consider differences in platform and device composition compared to major markets. These regions often skew more heavily toward Android and lower-end or older devices, which can influence both performance and monetisation behaviour. As a result, the most effective monetisation experiments tend to focus on efficiency and quality rather than simply pushing more volume. Small improvements compound quickly in these regions, but aggressive changes can just as easily damage retention or UA performance.

One of the most impactful levers is bid floor optimisation. In low-CPM environments, a long tail of low-value, often low-quality ads can start to dominate delivery. These ads rarely contribute meaningfully to revenue and can actively harm user experience. Introducing sensible bid floors helps filter out that weakest demand, often improving engagement and retention while having a neutral or even positive impact on overall ad LTV.

Closely related to this is partner specialisation. Global demand partners do not always perform equally well across all geographies. In lower-ARPU markets, testing specialised networks with strong local or regional inventory can deliver meaningful CPM uplift. Even small improvements here can materially improve returns without changing ad frequency or placement strategy.

On the IAP side, localised and affordable pricing can be a powerful experiment. Adapting price points to local purchasing power, rather than relying on tier-one defaults, can encourage more users to convert through smaller, more frequent purchases. In many lower-ARPU markets, microtransactions at accessible price points outperform fewer, higher-priced offers, improving payer conversion without materially impacting UA efficiency.

User segmentation is another area where experimentation tends to pay off. Lower-ARPU markets often have very different IAP and IAA dynamics compared to “tier-one” regions. Segmenting users based on early payer signals, ad engagement, or progression allows you to tune experiences more precisely. This might mean adjusting rewarded offers, experimenting with bundles, or shifting the balance between ad exposure and purchase opportunities for specific cohorts.

The direction of these experiments should be informed by what’s actually limiting value. If payer conversion is extremely low but ad engagement is healthy, leaning slightly more into ad-driven configurations can make sense. If CPMs are weak but payer intent exists, experimenting with pricing, offers, or bundles may deliver better results. When both payer conversion and CPMs are low, that’s often a signal to reassess UA efficiency or long-term viability in that market rather than force monetisation changes.

Overall, the highest lifts in lower-ARPU markets usually come from targeted optimisation rather than sweeping changes. Improving demand quality, matching partners to geographies, and aligning monetisation strategy to local user behaviour allows teams to extract incremental value without undermining retention or UA efficiency.

How do you integrate ASO learnings into monetization strategy, specifically when store-page experiments reveal shifts in user intent or payer propensity?

The first step is being clear on strategy and optimisation goals. ASO is not just a UA lever; it’s an early part of the user experience and one of the first signals of user intent. When store-page experiments shift who you attract, monetisation and engagement can easily change in tandem.

I tend to treat ASO learnings in the same way as a product change. When an ASO variant wins, it’s important to look beyond install rate and assess the impact on post-install metrics. How are ARPU, payer conversion, retention, and churn affected? A store page that attracts more users but materially lowers payer propensity or early engagement may still be valid, but only if the downstream economics support it.

Consistency is another important consideration. Do the ASO learnings hold across platforms, countries, and acquisition sources, or are they highly localised? In many cases, an ASO variant that performs well in one market or store can behave very differently elsewhere. Understanding where results generalise and where they don’t helps avoid overfitting monetisation decisions to a narrow slice of traffic. Custom product pages make it increasingly feasible to test store fronts more precisely.

At scale, highly segmented ASO strategies can become resource intensive. Running different store pages by geo, platform, or traffic type requires dedicated ownership and tooling. That said, with the right staffing and clear cost-benefit thinking, this approach can become a meaningful advantage.

Ultimately, ASO should not be viewed in isolation. When store-page experiments shift who you acquire, monetisation strategy needs to respond. Teams that get the most value from ASO treat it as an upstream monetisation lever, not just a conversion optimisation exercise.