Scaling App User Acquisition: How to Grow Ad Spend Without Breaking ROAS

Scaling app user acquisition is the process of increasing paid media spend on an app while keeping cost per install, cost per action and return on ad spend inside the targets that make the growth profitable. Most apps can buy more installs tomorrow by raising a budget. Very few can buy more profitable users at the same rate, because every extra euro pushes the bidding algorithm into a less efficient part of the auction. This guide explains how Admiral Media approaches that problem across Google App Campaigns, Meta, TikTok and Apple Ads, with the step sizes, measurement windows and case study results behind each recommendation.

Two definitions matter. Vertical scaling is raising the budget of a campaign that already works. Horizontal scaling is adding new campaigns, countries, channels, creatives or optimization events to reach users the existing setup cannot. Healthy growth uses both, in a specific order. Based on Admiral Media’s work managing more than €500M in mobile ad spend for over 150 mobile brands, the order matters more than the size of any single budget increase.

Why Scaling App User Acquisition Breaks Efficiency

Scaling breaks efficiency because ad platforms buy the cheapest qualifying users first, so each additional unit of budget reaches users who are harder to convert. That is not a sign the campaign is broken. It is the normal shape of an auction, and the job of a scaling plan is to move that curve outward faster than spend climbs along it.

Three mechanisms explain most of the efficiency loss the Admiral Media team sees when a client raises budgets aggressively.

1. Diminishing marginal returns in the auction

An App campaign with a target CPA or target ROAS ranks every available impression by predicted value. At a low budget it only needs to win the top slice of that ranking. When the budget doubles, it must also win impressions further down, where predicted conversion rates are lower and bids have to be higher to compete. Average CPI rises because the marginal CPI, the cost of the last install bought, rises much faster than the average.

2. Learning resets

Every major platform treats a large budget or target change as new information and re-enters a learning or assessment period. Google’s own App campaign for installs checklist states plainly: “Do not change your budget by more than 20% up or down,” and it describes learning and assessment phases lasting 7 to 14 days. TikTok’s guidance for value-based optimization asks advertisers not to change budget or target ROAS during a learning phase of at least seven days. Meta’s delivery system also re-enters learning after significant edits to an ad set, and budget changes can count as significant edits. A budget jump that forces a reset throws away the signal the algorithm spent weeks collecting.

3. Creative and audience saturation

More budget means more impressions against the same creatives and the same pockets of the audience. Frequency rises, click-through rate decays, and the cost of the next conversion climbs even if the auction itself has not changed. Admiral Media’s research on the creative fatigue curve covers this decay in detail. The practical point for scaling is simple: budget and creative volume have to grow together. TikTok’s value-based optimization tips say the same thing in platform language, recommending that advertisers upload 2 to 5 more videos when they increase the ad budget.

None of these mechanisms can be removed, only managed: size steps to avoid resets, feed creative before saturation, and add horizontal capacity before vertical scaling hits the steep part of the curve.

Signals That an App Is Ready to Scale

An app is ready to scale when its unit economics are proven at the current budget, its measurement is trustworthy, and its campaigns have enough conversion volume for the algorithm to learn from more spend. Scaling before those three conditions hold usually multiplies a problem instead of a result.

In Admiral Media’s onboarding audits, the readiness check covers five questions:

1. Is payback acceptable at today’s spend? If a cohort acquired this month does not pay back inside the window the business can finance, more of that cohort only deepens the hole. The CAC payback period guide explains how to set that window.
2. Does tracking capture the events that predict revenue? If the optimization event is an install while revenue comes from trials or purchases, the algorithm is scaling toward the wrong user.
3. Is conversion volume high enough? Google’s checklist recommends that the in-app action used for optimization occurs at least 30 times per day, preferably more. TikTok names 50 conversions as the most significant indicator of passing the learning phase.
4. Is there a creative pipeline? A campaign running on two or three winning ads has a short runway once spend rises.
5. Is the product ready for the volume? Onboarding, paywall and retention problems are amplified at scale. Admiral Media’s work with Inshallah showed this directly: cheap Android installs looked attractive, but analysis showed high churn by Day 5 from those users, with revenue below targets.

If any answer is “no”, fix that bottleneck first.

The Admiral Media Scaling Ladder Framework

The Admiral Media Scaling Ladder Framework is a seven-step sequence for growing app ad spend in controlled increments, where each step must hold its efficiency targets before the next one starts. It combines vertical budget steps with planned horizontal expansion so that spend never outruns the signal, creative or measurement that supports it.

The Admiral Media Scaling Ladder Framework

1. Lock the efficiency guardrail. Define the CPA, ROAS or payback threshold that a scaled campaign must hold, and the measurement window it is judged on. Without a written guardrail, every review becomes an argument about whether a worse number is “acceptable for scale”.
2. Prove the base. Run the campaign at a stable budget until it exits learning and delivers the guardrail for a full evaluation window. Google suggests giving a campaign about 10 days to settle before judging it.
3. Climb in small vertical steps. Raise budgets in increments of roughly 20%, in line with Google’s guidance, and wait for each step to stabilize before the next. Small steps keep the campaign out of a full learning reset.
4. Feed creative ahead of spend. Add new concepts and variants of winners before frequency climbs. A useful habit is to schedule creative launches on the same day as budget steps, so the algorithm has new inventory when it needs to buy deeper into the audience.
5. Move the optimization event down the funnel. Once a campaign has enough volume on a shallow event such as installs or sign-ups, shift optimization toward trials, first purchases or value. Deeper events let the algorithm separate valuable users from cheap ones as spend grows.
6. Expand horizontally when vertical returns flatten. When the marginal CPA of a budget step breaches the guardrail, stop climbing and add capacity: a new channel, country, language, campaign type or audience structure.
7. Validate incrementality and reallocate. At scale, platform-reported results overlap. Test which spend is actually incremental and move budget toward the channels, operating systems and geos that produce the most profitable users.

The leverage comes from discipline: one change at a time, enough time between changes, and a clear rule for when to stop climbing and start expanding.

How Fast Can You Increase an App Ad Budget?

On Google App Campaigns, the platform’s own guidance is to avoid changing budgets or targets by more than 20% at a time, and to let a campaign settle for 7 to 14 days around learning and assessment. Applied consistently, those small steps still compound into large growth over a quarter.

Google’s tips for maximizing your App campaign repeat the caution against “changing the budget by >20%, or changing your CPI by >20%.” Many marketers read this as a speed limit that makes scaling slow. The arithmetic says otherwise.

Five consecutive 20% steps take a budget to about 2.49x its starting level. Nine steps take it to about 5.16x. If each step is held for about a week, nine steps fit inside roughly nine weeks. In other words, a team that never breaks the 20% rule can still grow a single campaign’s budget fivefold in a quarter, provided each step holds the guardrail.

That pace is consistent with what Admiral Media has documented in real accounts. Admiral Media scaled TIER’s user acquisition budget 5x in less than 3 months while expanding into new channels, as described in the TIER case study. The lesson is not that a specific step size produced that result. It is that fivefold growth inside a quarter is achievable without reckless jumps, because disciplined steps compound.

Two practical notes:

• Steps compound on the current budget. A 20% step on a doubled budget is a much larger absolute move, so watch marginal CPA, not the percentage.
• Steps can go down. If a step breaches the guardrail for a full evaluation window, step back rather than waiting for the campaign to “grow into” the budget.

Platform Guardrails for Scaling Budgets

Each ad platform publishes its own thresholds for budget size, learning duration and conversion volume, and a scaling plan should respect the strictest one that applies. The table below summarizes the published guidance the Admiral Media team uses as a starting point.

Platform	Minimum budget guidance	Change size guidance	Learning and evaluation	Volume signal
Google App Campaigns (installs)	Daily budget at least 50x target CPI	No more than 20% up or down	Learning and assessment 7 to 14 days; give the campaign about 10 days to settle	Conversion lag up to 5 days, especially on iOS
Google App Campaigns (in-app actions)	Daily budget at least 10x target CPA	No more than 20% for budget or target	Same settling period	Optimization action at least 30 times per day
TikTok value-based optimization	Daily budget 30x CPA, or 3x a manual campaign budget	No budget or target ROAS changes during learning	At least 7 days of learning	50 conversions as the key learning-phase signal
Meta app campaigns	Enough budget to generate consistent optimization events	Large budget edits can count as significant edits	Learning re-enters after significant edits	Event volume determines learning exit
iOS measurement (SKAdNetwork 4)	Not a budget rule	Not a change rule	Postbacks arrive 24 to 144 hours after windows close	Larger crowds unlock richer postback data tiers

Sources: Google’s App campaign checklist and App campaign tips (linked above), TikTok’s tips for value-based optimization for app, and Apple’s SKAdNetwork postback documentation (linked below). Meta guidance is stated qualitatively.

The budget multiples matter in a less obvious way: a campaign funded below 10x its target CPA struggles to generate enough daily conversions to learn. In Admiral Media’s audits, many “this campaign does not scale” complaints turn out to be campaigns that were never funded enough to finish learning in the first place.

Vertical vs Horizontal Scaling: Choosing the Right Lever

Vertical scaling is the fastest lever while a campaign’s marginal CPA is still inside the guardrail. Horizontal scaling becomes the better lever once it is not. Most stalled accounts the Admiral Media team takes over have pushed vertical scaling past that point and are paying auction premiums for users a new campaign or channel could reach more cheaply.

Scaling lever	What it changes	Best used when	Main risk	Metric to watch
Vertical budget steps	Spend on a proven campaign	Marginal CPA or ROAS still inside guardrail	Learning resets, rising marginal cost	Marginal CPA per step
Deeper optimization event	What the algorithm buys	Shallow event has enough daily volume	Volume drops below learning thresholds	Daily count of the new event
Creative volume	Inventory the algorithm can match to users	Frequency rising, CTR decaying	Testing spend on weak concepts	New concept win rate, CTR trend
New channel	Audience pool and auction	Core channel’s marginal returns flatten	Measurement overlap, setup cost	Incremental CPA vs core channel
New country or language	Market size	Product and payments work locally	Localization quality, lower LTV	Country-level payback
Bidding strategy change	How bids are set	Revenue data is reliable	Spend drops while targets recalibrate	ROAS at stable spend

Admiral Media’s work with TIER shows the combination in practice. The Admiral Media team audited TIER’s existing Facebook campaigns, mapped which channels fit the target personas, prepared creative briefs and an ad copy database in 8 languages, and then added Google and Snapchat to the stack. The result was +297% new customers, 2 new channels beyond Facebook and a 5x budget in under three months. The operating rule described in the case study is worth copying: scale the best-performing creatives and build variations, pause non-performers, look for new audiences and channels, then “measure profitability, KPIs, double in size and repeat.”

Horizontal expansion into new markets deserves its own plan. The app market expansion guide covers how Admiral Media selects and enters new countries without diluting payback.

Move the Optimization Event Down the Funnel as You Scale

The single most effective scaling move in Admiral Media’s case studies is shifting optimization from a shallow event to a deeper one once volume allows. Deeper events give the algorithm a sharper definition of a valuable user, so additional spend buys better users instead of just more users.

The pattern appears again and again:

• TIER: campaigns started with install optimization and moved down the funnel to sign-up optimization once enough learnings and data had accumulated in the account.
• Himalaya: Admiral Media started with sign-up optimization, moved to trial optimization, and used multiple levels of proxy event campaigns to understand drop-off and build a minimum daily event count for each campaign.
• Inshallah: after months of community building in the US, the ad networks had enough data to find likely purchasers. In September 2023 Admiral Media shifted US campaigns to purchase optimization, which produced a significant revenue spike.
• Fast commerce: for a leading UK grocery delivery app, Admiral Media moved from install optimization to sign-up optimization and later to first-order optimization.

The fast commerce account is the clearest illustration of why this matters for scaling. Admiral Media took that client from zero to seven figures in monthly user acquisition spend. Installs grew, but the events closer to revenue grew far faster.

Paid installs rose 64%, while paid purchases rose 1,210%, paid conversions 2,178% and paid users 5,152%, with paid CAC down 91%. If the campaigns had stayed on install optimization, the extra budget would most likely have bought more installs from users who never ordered. Moving the event down the funnel is what turned spend growth into revenue growth.

The trade-off is volume: a deeper event happens less often, and below the platform’s threshold the campaign stalls in learning. That is why the Scaling Ladder places this move after vertical scaling has built volume. Admiral Media’s guide to app event optimization explains how to choose the event and when to make the switch.

Bidding Strategy at Scale: From Cost Targets to Value

As spend grows, cost-based bidding (target CPI or target CPA) tends to buy the cheapest users who complete an action, while value-based bidding (target ROAS) buys users weighted by predicted revenue. For apps with reliable revenue data, moving toward value bidding is often what allows ROAS to rise while spend scales.

ChatPDF is the clearest example in Admiral Media’s portfolio. ChatPDF asked Admiral Media to scale paid acquisition during its low season while improving ROAS and CAC efficiency. The Admiral Media team consolidated overlapping ad sets, tested target CPA against target ROAS, added value rules and an LTV signal, and ran weekly creative concept tests with three variants per winner. The case study’s key learning states it directly: value bidding plus audience analytics increased ROAS while spend scaled.

Admiral Media managed ChatPDF’s Google and Meta campaigns with value-based bidding and a cadenced creative engine, achieving a 320% increase in ROAS, 142% more subscriptions and a 38% lower CAC on Google, and a 280% increase in ROAS, 171% more subscriptions and a 45% lower CAC on Meta.

Value bidding is not a universal answer. It needs enough purchase or revenue events to model value, and switching to it usually causes a temporary drop in delivery while targets recalibrate. Admiral Media typically introduces it as a parallel test against the existing cost-based campaign rather than a hard switch, and sets the initial ROAS target close to what the account already achieves so delivery does not collapse. For subscription apps with long payback, the predictive LTV bidding approach extends this further by feeding early signals that predict lifetime value.

Creative Supply Is the Hidden Scaling Constraint

Creative volume caps how far a campaign can scale, because the algorithm can only reach new users efficiently if it has ads that resonate with them. When spend rises faster than creative supply, frequency climbs and performance decays even if bids and budgets are set perfectly.

Admiral Media builds creative supply as a system rather than a sequence of one-off ads. In the fast commerce account, the Admiral Media team brainstormed concepts and analyzed market creative twice a week, produced tens of storyboards, and launched tests with an accelerated testing strategy designed to reach statistical significance within one or two days. Winners were scaled, varied and fed back into the process.

Admiral Media’s work with NeuroNation followed the same test-and-learn logic: the team categorized communication ideas, tested them against target audiences in every market, and introduced the pRank methodology to identify likely winners and losers faster. Admiral Media managed NeuroNation’s campaigns with this structured testing approach, achieving a 117% increase in ROAS, 66% more installs, 42% higher net cohort revenue and a 39% lower CPI (January to August 2019). NeuroNation’s co-founder summarized the scaling outcome on the NeuroNation case study page: the team was able to scale paid social spend by more than 200% while achieving its target ROAS.

Three creative rules Admiral Media applies when scaling:

• Ship creative with every budget step. New ads give the algorithm fresh inventory to match against the less responsive users that higher budgets reach.
• Iterate on winners before inventing new concepts. Variations of a proven concept have a higher hit rate and keep the learning the campaign already has.
• Test concepts, not only formats. Inshallah’s growth came from finding the “Halal Love” concept, which outperformed family, privacy and security angles, and from leaning into user-generated content once it beat static images and non-UGC video.

Scaling a New Channel From Zero

Scaling a new channel from zero follows a different path from scaling an existing campaign: the first job is building a clean, approved account and enough conversion history, and only then does budget growth begin. Rushing spend into an unproven account typically produces rejections, unstable delivery and misleading early results.

Admiral Media’s work with the audio streaming app femtasy on TikTok shows the sequence. femtasy had faced recurring ad approval problems that put its TikTok campaigns at risk. Admiral Media set up a new ad account for German user acquisition, introduced a rigorous pre-approval process for landing pages and creatives in coordination with TikTok and the client, and started with a deliberately conservative content approach before moving to more dynamic angles and testing formats such as Spark Ads, carousels and Smart Creative Ads. Within five months of building the new account, Admiral Media scaled femtasy’s TikTok spend from €0 to more than €30,000 per month while maintaining a positive ROAS, with the best creatives reaching a 6% click-through rate, according to the femtasy case study.

Channel diversification can also change the economics outright. For the dating app PURE, Admiral Media tested a programmatic DSP against an established self-attributing network with separate budgets and platform-specific ads. The DSP delivered a CPI of $2.44, four times lower than the network’s $9.43, while exceeding D7 ROAS goals, which led to launches in new markets.

Measurement at Scale: Reading Results You Can Trust

Measurement becomes harder as spend grows, because results from multiple channels overlap and iOS data arrives late and aggregated. A scaling decision is only as good as the data behind it, so the evaluation window must match the measurement delay of each platform.

On iOS, SKAdNetwork 4 splits attribution into three conversion windows. Apple’s documentation states that the first window spans days 0 to 2, the second days 3 to 7, and the third days 8 to 35, and that postbacks arrive after a random delay of 24 to 48 hours for the first postback and 24 to 144 hours for the second and third.

The implication for scaling is concrete. A budget step taken on Monday cannot be judged on iOS revenue data by Wednesday, because the second postback, which carries day 3 to 7 activity, may not arrive until the following week. Google’s checklist makes the same point for App campaigns, noting that conversion lag may take up to 5 days, especially on iOS.

Apple’s documentation adds a second, less discussed effect. SKAdNetwork assigns each install a postback data tier based on crowd size, and only the higher tiers return fine-grained conversion values, longer source identifiers and, at Tier 3, a country code. Larger campaigns tend to generate larger crowds, so scale can improve the granularity of iOS data. That is one reason fragmenting a scaling budget across many tiny campaigns on iOS often makes measurement worse rather than more precise.

Inshallah shows why this matters in budget allocation. Admiral Media set an initial conversion value schema and refined it from SKAN revenue data. That analysis showed TikTok’s CPI volatility directly affected earnings, and when revenue was compared with Facebook, Facebook proved more profitable, so budget was reallocated with confidence. Combined with the finding that iOS users delivered significantly higher revenue and nearly twice the retention of Android users, the reallocation helped Admiral Media grow Inshallah’s US iOS subscriptions by 824% and US iOS revenue by 1,253%, as shown in the Inshallah case study.

At larger budgets, the last measurement question is incrementality: how much of the reported result would have happened anyway. Platform dashboards each claim credit for overlapping users, and the overlap grows with spend. Admiral Media recommends incrementality testing and a blended efficiency view such as blended ROAS and MER before committing the next large tranche of budget.

Diagnosing a Scaling Stall

When efficiency drops after a budget increase, the cause is usually one of a small number of identifiable problems, and each has a different fix. Raising or cutting budget blindly treats all of them the same way.

Symptom after a budget step	Likely cause	What Admiral Media checks	Typical action
CPA spikes for a few days, then recovers	Normal learning or assessment after the change	Size of the change, days since edit	Wait out the evaluation window; keep future steps near 20%
CPA rises steadily, CTR falls, frequency rises	Creative saturation	Age and share of spend of top ads	Launch new concepts and variants of winners
Installs grow, purchases flat	Optimization event too shallow	Daily volume of deeper events	Move optimization down the funnel when volume allows
Spend will not rise despite budget increase	Target too tight for the larger audience	Bid or ROAS target vs recent actuals	Loosen target in small steps, or widen inventory
Platform ROAS rises, business revenue flat	Attribution overlap, low incrementality	Blended revenue vs total spend	Run an incrementality test, rebalance channels
Strong installs on one OS or geo, weak revenue	Low-value segment absorbing budget	Retention and revenue by OS and country	Reallocate toward the most profitable segments

The Inshallah account illustrates the last row: cheap Android users churned heavily by Day 5, and shifting the main KPI from CPI to ROAS while reallocating toward iOS users on Facebook and TikTok made the account scale profitably.

A 90-Day Scaling Plan

A practical scaling plan for most apps runs in three phases over roughly a quarter: stabilize, climb and expand. The phases are a planning structure, not a promise of outcomes, and each phase gate depends on the guardrail holding.

Days 1 to 30: Stabilize the base

Audit tracking and events, confirm the guardrail and its evaluation window, consolidate overlapping campaigns, fund each core campaign to at least the platform’s minimum budget multiple, and build a creative backlog. Do not scale yet.

Days 31 to 60: Climb

Raise budgets in steps of about 20%, holding each step for its evaluation window (longer on iOS because of postback delay). Launch new creative with each step, track marginal CPA per step, and test a deeper optimization event in parallel once volume allows.

Days 61 to 90: Expand and validate

When budget steps stop holding the guardrail, shift to horizontal growth: a new channel, country or value-bidding test. Run an incrementality check on the largest channel before the next tranche and reallocate toward the best-payback segments.

Admiral Media’s scaling work with Himalaya followed this broad shape: analyze the tech stack and define proxy events, find winning creatives and channels through rapid testing, find the best audiences per channel, country and language, then scale while maintaining results. The outcome was +875% installs, +676% users acquired and a 79% lower CAC, according to the Himalaya case study.

Frequently Asked Questions

How much should I increase my app ad budget at a time?

Google recommends not changing an App campaign’s budget by more than 20% up or down at a time, and not changing the target CPI by more than 20% either. Small steps avoid a full learning reset and let you see the marginal cost of each increase. Consecutive 20% steps compound quickly: nine of them take a budget to about 5x its starting level. Hold each step long enough to evaluate it, which is longer on iOS because SKAdNetwork postbacks arrive with a delay.

Why does my CPI go up when I scale my app campaigns?

CPI usually rises with spend because ad platforms buy the most likely converters first, and additional budget has to reach users who are less likely to install. The marginal cost of the last install rises faster than the average. Learning resets after large budget changes and creative fatigue from higher frequency add to the increase. Smaller budget steps, fresh creative and horizontal expansion into new channels or markets are the main ways to slow it.

What is the difference between vertical and horizontal scaling in user acquisition?

Vertical scaling means increasing the budget of a campaign that already performs. Horizontal scaling means adding new campaigns, channels, countries, languages, audiences or creative concepts to reach users the current setup cannot. Vertical scaling is faster while marginal CPA stays inside your target, but it flattens as the campaign saturates its audience. Horizontal scaling takes more setup but opens new auctions and audience pools, so most mature accounts combine both.

How many conversions does a campaign need before I scale it?

Platforms publish different thresholds. Google recommends that the in-app action used for optimization occurs at least 30 times per day and that the daily budget is at least 10x the target CPA. TikTok’s value-based optimization guidance names 50 conversions as the most significant indicator of passing the learning phase, with at least seven days of learning. A campaign below these levels is usually still learning and is not a stable base to scale from.

When should I switch from target CPA to target ROAS bidding?

Switch to target ROAS when your app reliably reports purchase or revenue events in enough volume for the platform to model value. Value bidding tends to help most when spend is growing, because it weights users by predicted revenue instead of treating every conversion as equal. Test it in parallel with the existing cost-based campaign and set the starting ROAS target close to current performance to avoid a sharp drop in delivery. Admiral Media used this approach for ChatPDF, where ROAS rose 320% on Google and 280% on Meta.

How long should I wait before judging a budget increase on iOS?

On iOS, wait at least a week before judging a budget step on revenue, and longer if your key events happen after day 7. Under SKAdNetwork 4, the first postback covers days 0 to 2 and arrives 24 to 48 hours after that window, while the second covers days 3 to 7 and can arrive up to 144 hours later. Early reads mostly reflect first-window activity. Google also notes that App campaign conversion lag can take up to 5 days, especially on iOS.

Can an app scale ad spend 5x without losing efficiency?

Yes, when spend growth is paced and supported by creative, deeper optimization events and new channels. Admiral Media scaled TIER’s user acquisition budget 5x in less than three months while adding two new channels and growing new customers by 297%. The results depend on the app’s unit economics and product quality, so the safe approach is to set an efficiency guardrail and only take the next budget step when the current one holds it.