Vytis Bareika on Scaling Apps: ROAS, Retention, and the Real Cost of Growth

Anastasiia Malyhina

Spending $1 million a day on advertising sounds like a significant achievement. But what if the biggest mistake at that scale is not spending enough?

That might seem counterintuitive when most marketing teams are trying to reduce acquisition costs, improve ROAS, and make every dollar work harder. Yet once a product has proven its economics, limiting spend can mean leaving substantial growth opportunities on the table.

In the latest episode of the Retention Podcast, Oleg Lesov talks with Vytis Bareika, Founder & CEO of Defined Chase, a performance marketing agency, and former Head of Digital Marketing at Nordcurrent. Having worked with global brands, mobile game publishers, and companies managing substantial advertising budgets, Vytis has seen how growth strategies change as businesses move from early testing to large-scale acquisition.

The conversation covers predictive LTV, cohort-based ROAS, Web2App funnels, geographic expansion, attribution, creative production, and AI. But much of the discussion comes back to the same challenge: how do you know when your business is ready to spend more, and what should you fix when growth starts slowing down?

Why Cheap Installs Can Become an Expensive Mistake

Performance marketing for mobile apps requires a different mindset from traditional digital advertising.

In many traditional businesses, marketers focus on relatively immediate results: cost per acquisition, return on ad spend, and the revenue generated from a particular campaign. Mobile apps introduce a longer timeline. Users may install today, subscribe next week, make additional purchases months later, or continue generating revenue for years.

That changes how acquisition performance should be evaluated.

Vytis explains that when he moved from working with traditional global brands to mobile gaming, the biggest difference was the importance of cohort analysis and lifetime value. Instead of looking only at the cost of bringing someone into the product, marketers need to understand what happens after acquisition and how much revenue those users eventually generate.

This is particularly important for games with in-app purchases, where a relatively small group of highly engaged customers can account for a significant share of revenue.

Vytis has worked with a game that accumulated more than 500 million downloads. At that scale, historical data makes it possible to connect early performance with longer-term outcomes. A team might know, for example, that a cohort generating 20% ROAS by day seven has historically reached profitability after several months.

That relationship allows marketers to make acquisition decisions before the full revenue picture is available.

The difficulty is that not every business has years of historical data. New apps may only have a few weeks of results, making long-term predictions much less reliable.

For those companies, Vytis recommends starting with a proper mobile measurement setup. Tools such as AppsFlyer, Adjust, Singular, Kochava, or Branch help teams connect acquisition channels with downstream user behavior and revenue.

From there, the focus should move beyond basic acquisition metrics.

CPM, CPC, and CPI are useful for understanding campaign mechanics, but they don't tell the complete story. A campaign generating installs at $2 may perform worse financially than one generating installs at $10 if the second campaign attracts customers who spend significantly more.

As Vytis puts it, you get what you ask for. Optimize for cheap installs, and the platforms will find cheap installs. Optimize for revenue, and the acquisition strategy starts moving toward customers who are more valuable to the business.

Why Mobile App Marketing Is a Different Game
Why Mobile App Marketing Is a Different Game

The Most Expensive Mistake Might Be Not Spending Enough

One of the more surprising moments in the conversation comes when Oleg asks Vytis about his most expensive performance marketing mistake.

His answer isn't a failed campaign or an expensive targeting decision.

It's the opportunity he missed by not scaling faster.

Vytis recalls managing advertising spend of approximately $1 million per day while his team continued encouraging him to increase the budget. Looking back, he believes the company could have captured more growth by spending more aggressively.

Of course, this doesn't mean every business should immediately double its advertising budget.

The key is understanding the relationship between acquisition costs, payback periods, and available cash flow.

Different companies can tolerate very different payback windows. Some large gaming publishers are comfortable waiting one or even two years to recover acquisition costs. Other app businesses expect to recover their investment within 30 days or, at most, a few months.

Those differences directly affect how aggressively a company can compete for users.

A business that can wait two years for profitability may be willing to pay much more for the same customer than a startup that needs its acquisition investment back within a month.

Vytis mentions that, in his experience, day-seven ROAS below approximately 20% often makes him cautious about scaling. However, he emphasizes that this is not a universal threshold. The acceptable result depends on the business model, retention curve, historical data, and cash-flow requirements.

The real decision is whether the expected return justifies the capital being committed.

This becomes especially relevant for subscription apps. Unlike games where a high-spending customer might make purchases worth thousands of dollars, subscriptions often have a more predictable revenue ceiling.

That makes improving LTV particularly important.

Vytis points to upsells and additional offers as opportunities, especially immediately after an initial purchase. His observation is that customers can be particularly receptive to related purchases during the first 24 hours after buying something.

For subscription businesses, this creates an opportunity to increase the value of an acquired customer without relying entirely on future renewals.

But the offer still needs to make sense for the customer. The goal is to identify additional value that fits the original purchase, rather than simply adding more payment steps.

Web2App Funnels: When More Friction Can Actually Help

Web2App funnels have become an increasingly common acquisition strategy, extending beyond health and fitness into utilities, tracking products, and other app categories.

Vytis sees two major advantages.

First, web-based payments can change the economics of acquisition by offering alternatives to in-app purchasing. Second, quiz funnels give companies more room to explain the customer's problem and position the product as a solution before asking for payment.

However, he doesn't believe that longer funnels are automatically better.

In fact, Vytis generally prefers simpler user journeys. Every additional step creates an opportunity for someone to leave.

The exception is what he calls good friction.

For some businesses, attracting as many leads as possible isn't the objective. A large number of unqualified prospects can waste sales resources and send poor-quality optimization signals back to advertising platforms.

In those cases, qualification questions can improve the acquisition process.

Vytis describes a setup where users answer questions before progressing through the funnel. Qualified users continue, while others are directed to a different endpoint. The advertising conversion event is triggered only for the qualified group.

The idea is to teach the advertising platform what a valuable prospect looks like rather than rewarding it for generating as many low-quality leads as possible.

This approach requires care, particularly around accurate event definitions, consent, and platform requirements. But the underlying principle is useful: conversion volume isn't always the best objective when conversions have very different business value.

The conversation also explores how tracking becomes more complicated when the customer journey moves between a website and an app.

Web pixels, server-side events, mobile attribution, and consent settings all need to work together. For purchase-driven products, the most important events might include checkout, purchase, and subscription activation.

For businesses where revenue happens much later, marketers may need to optimize for an earlier event that correlates strongly with future purchases.

A free-trial start, for example, can become a useful acquisition signal if the company understands how frequently trial users convert into paying customers and how much those customers are worth.

The important part is connecting the event being optimized to actual business outcomes.

Good Friction: How to Qualify Better Users
Good Friction: How to Qualify Better Users

When Growth Stalls, Look Beyond the Advertising Account

Increasing acquisition spend is only one way to grow. Companies can also expand into new countries, reach new audiences, improve retention, or change their monetization strategy.

Vytis has experience running campaigns globally, including markets that many app companies overlook while concentrating on the United States and other high-income countries.

His approach is to group markets not only by geography but also by their economic characteristics.

Countries with similar LTV can sometimes be managed together even if they are geographically distant. This makes it easier to optimize campaigns around comparable acquisition costs and expected returns.

Different markets also serve different purposes.

India and Indonesia, for example, can be useful for technical launches because relatively inexpensive installs allow teams to test infrastructure, identify errors, and observe early user behavior at scale.

Japan and South Korea can be attractive markets for gaming products, although their customer expectations and creative requirements may differ substantially from Western markets.

Latin America offers lower acquisition costs in many categories, but lower purchasing power means companies still need to evaluate the economics of each product carefully.

Vytis also notes that AI has made localization significantly easier, although translating a creative isn't always enough. Cultural references, local celebrations, and customer expectations still matter.

Yet geographic expansion won't solve every growth problem.

When a product stops growing, Vytis starts by examining retention. Are users leaving after the first day, the first week, or several months? Are they running out of content? Are they receiving enough value to renew their subscriptions?

For gaming products, content exhaustion can be a major reason for declining engagement. For other apps, the issue might be that users no longer see a reason to return.

He also looks at whether acquisition teams have exhausted their existing audience.

Many campaigns repeatedly target people who already understand the category and are close to making a purchase. That can work for a while, but eventually the available audience becomes saturated.

Reaching new customers requires different creative angles and a broader explanation of the problem.

Instead of only targeting people who are actively searching for a parking app, for example, marketers might explore different situations where parking creates frustration and introduce the product to people who haven't considered using an app to solve it.

That means creative strategy needs to evolve alongside the business.

New audiences often require new messages, and new messages may require different onboarding or lifecycle journeys. A campaign that attracts customers through one particular problem should ideally lead into an experience that continues addressing that problem.

Otherwise, acquisition creates expectations that the product fails to carry forward.

Attribution Isn't the Truth. It's a Way of Looking at the Business.

As acquisition becomes more sophisticated, measuring performance becomes harder.

A customer might discover a product through a video, search for the brand several days later, click a paid advertisement, receive an email, and eventually purchase.

Which channel deserves the revenue?

The answer depends on the attribution model.

Vytis is skeptical of relying exclusively on first-click or last-click attribution because different channels play different roles in the customer journey.

He gives an example of a health-related product where a channel appeared highly successful under middle-touch attribution but performed poorly under last-click attribution.

Neither result necessarily tells the complete story.

Oleg raises a similar issue from lifecycle marketing. A customer may click an email, later interact with a paid advertisement, and then make a purchase. Depending on the attribution window, the email platform could receive credit for revenue that was also influenced by paid acquisition.

This can make channel-level performance appear stronger than its actual incremental contribution.

For Vytis, attribution is ultimately a business decision involving trade-offs. There is no single model that perfectly captures every influence on a purchase.

That's why he values incrementality testing.

Instead of relying only on attributed conversions, companies can temporarily stop advertising in a particular market, region, or channel and measure what happens to overall results.

He shares an example of a client that decided to stop bidding on branded search terms. The assumption was that customers would simply click organic results instead.

The experiment was supposed to last eight weeks, but the client asked to restore the campaigns after approximately three or four weeks.

Direct and organic traffic declined, and the business lost revenue. The paid brand campaigns had been contributing more value than expected.

The lesson is that attribution reports and actual business impact aren't always the same thing.

For companies operating across many channels, Vytis also sees value in blended ROAS, blended CPA, and media mix modeling. But he cautions against making measurement unnecessarily complicated.

A model that is imperfect but understandable and verifiable can sometimes be more useful than a sophisticated system nobody can explain.

This matters particularly when connecting acquisition with lifecycle marketing. Teams need a shared understanding of where customers come from, how they behave, and what drives additional revenue.

Without that foundation, marketing departments can end up competing over attribution credit instead of improving the customer journey.

AI Is Accelerating Creative Production, but Ideas Remain the Bottleneck

AI is changing performance marketing in ways that go far beyond generating ad copy.

Vytis sees companies using AI to build landing pages, produce creative variations, develop internal tools, and automate operational tasks that previously required significant resources.

Video generation is becoming particularly interesting.

He has experimented with running local AI models on his own hardware, including generating hundreds of video variations overnight. Although production costs and technical requirements still vary, these experiments suggest that large-scale creative generation could become much more accessible.

For companies managing substantial advertising budgets, this changes the economics of experimentation.

The traditional challenge was often how many creatives a team could afford to produce. AI reduces some of those production constraints.

But it introduces another problem: deciding what is worth producing.

Vytis argues that companies still struggle to develop genuinely new hypotheses, identify promising creative directions, and move beyond the audiences and messages that already work.

Generating another hundred variations of the same concept doesn't necessarily help a business reach a new market.

The creative advantage increasingly comes from understanding customers, finding new angles, and knowing which experiments are worth running.

That also changes how teams should think about AI adoption.

Vytis is enthusiastic about AI agents and uses them for different functions, including development, finance, sales, and administrative work. Yet he also recognizes a problem familiar to anyone who has started using coding agents extensively: the temptation to do everything at once.

When AI makes it possible to launch several projects simultaneously, it becomes easy to confuse activity with progress.

Oleg describes a similar experience, sometimes running seven or eight agents on unrelated tasks and finding that constant context switching reduces the quality of his decisions.

Vytis has responded by concentrating more deliberately on the company's main constraint.

Rather than starting ten initiatives, he prefers identifying the biggest bottleneck, solving it properly, and then moving to the next one.

It's a useful distinction for marketing teams. AI can increase execution capacity, but that capacity only creates value when it is directed toward the right problem.

How is AI changing creative production?
How is AI changing creative production?

Will AI Reduce the Need for Performance Marketers?

Despite the growing capabilities of AI, Vytis doesn't expect demand for experienced performance marketers to disappear in the near future.

His conversations with large advertisers suggest almost the opposite.

Companies managing six-, seven-, and eight-figure monthly budgets are still looking for more talent. They have more opportunities to test, more channels to manage, and more decisions to make than their existing teams can comfortably handle.

AI can help people execute faster, but it doesn't remove the need for someone to decide which opportunities deserve investment.

Vytis believes particularly strong professionals will become even more valuable because AI allows them to expand their impact. Someone who understands acquisition economics, creative strategy, attribution, and the product can use automation to accomplish significantly more.

At the same time, less experienced marketers will need to develop those fundamentals rather than relying entirely on tools.

AI may help them learn, analyze, and execute, but they still need to understand why a campaign is working, whether a result is meaningful, and what should happen next.

The conversation ends with an interesting piece of career advice.

If Vytis could change something about his earlier career, he would start building a business sooner and invest in organic content much earlier.

He compares professional visibility to a compounding investment. Building an audience through consistent knowledge sharing can eventually generate speaking opportunities, partnerships, and customer relationships that would otherwise be difficult to create.

For a performance marketer who has managed enormous paid advertising budgets, it's a notable reminder that not every valuable growth channel needs to be purchased.

In a Nutshell

Vytis Bareika's experience shows why scaling mobile apps requires much more than improving campaign-level metrics.

A low CPI means little if the users don't generate revenue. A strong ROAS figure can be misleading when attribution ignores other channels. And an advertising budget that looks enormous might still be too conservative if the product has proven its economics and the company can afford the payback period.

The discussion also highlights how closely acquisition and retention are connected. Understanding customer value requires cohort analysis, reliable measurement, and a clear picture of what happens after installation. When growth slows, the answer may lie in retention, product content, new audiences, or monetization rather than another advertising optimization.

Web2App funnels and AI add new opportunities, but neither eliminates the need for good judgment. Funnels work best when their steps help identify valuable customers, while AI is most useful when it accelerates meaningful experiments rather than simply increasing output.

Perhaps the strongest takeaway is that performance marketing is becoming both more technical and more strategic.

Companies have more data, more automation, and more ways to reach customers than ever before. But the ability to interpret that information, understand the business economics, and decide where to invest remains the difference between buying traffic and building sustainable growth.

As Vytis's experience suggests, sometimes the right decision is to spend less, fix the product, or change the audience. And sometimes the biggest opportunity is having the confidence to spend more.

Natalya Ustymenko

|

February 16, 2023

George Johnson

|

November 22, 2025

Ready to Gain Real
Competitive Advantage?

See Your Messages Come to Life!

Get a firsthand look at how your messages will appear on user devices with our FREE simulator.

Preview Now

Evaluate and Elevate Your Push Notification Engagement

Are your pushes as compelling as they could be? Find out with our free preview tool. Instantly see how your notifications will display on iOS and Android, in both expanded and collapsed views.

This is your chance to fine-tune your messages to ensure they're not just seen but felt. Make every character and every push work harder for your engagement goals

Start the Test

Request a demo

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.