CPI, CPA, ROAS — every mobile marketing team tracks at least one of these, and most track all three. But the metric you optimize toward determines the quality of users you acquire, the campaigns you scale, and ultimately whether your UA budget produces real business outcomes or just impressive-looking dashboards.
The wrong metric doesn't just waste money — it actively misleads you. A campaign optimized for CPI will find the cheapest installs, which are rarely the most valuable users. A campaign optimized for ROAS without enough data will stall during the learning phase. Choosing the right performance marketing KPI for your specific situation is one of the highest-leverage decisions in mobile user acquisition.
This guide breaks down what CPI, CPA, and ROAS actually measure, when each one is the right optimization target, and which mobile marketing metrics matter most at each stage of campaign maturity.
CPI is the cost of acquiring a single app install. It's calculated as:
CPI = Total Ad Spend ÷ Total Installs
CPI is the most widely reported metric in mobile UA because it's simple, universally available, and allows direct comparison between networks and campaigns. Every ad network reports it; every media plan includes a CPI target.
CPI tells you how efficiently your campaigns are generating installs. A falling CPI means you're getting more installs for the same spend. A rising CPI means you're reaching audiences that are harder or more expensive to acquire.
CPI tells you nothing about the quality of the users being acquired. A campaign generating installs at $2 CPI from users who never open the app a second time is more expensive — in business terms — than a campaign generating installs at $8 CPI from users with 40% D7 retention and strong monetization rates.
CPI optimization is appropriate in two scenarios: early campaign testing, when you need install volume to feed algorithmic learning phases, and broad awareness campaigns where downstream conversion data isn't yet available. Once you have 50+ post-install conversion events, move to a deeper optimization target.
CPA is the cost of acquiring a specific user action beyond the install. The "action" is defined by the advertiser and should correspond to a meaningful business outcome. Common CPA events include:
- First registration or sign-up
- First purchase or subscription start
- First deposit (iGaming, fintech)
- Tutorial completion with first session engagement
- First meaningful in-app action (first bet, first trade, first booking)
CPA = Total Ad Spend ÷ Total Qualifying Actions
CPA tells you how efficiently your campaigns are acquiring users who take the action that matters to your business. It connects UA spending directly to conversion outcomes, not just install volume.
CPA doesn't tell you how much those converted users are worth over time. A campaign with low CPA for first registrations might still have poor LTV if the registered users don't monetize, return, or upgrade. CPA is a leading indicator of quality, but it's not a complete picture.
CPA optimization is the right default for most performance campaigns once sufficient install data is available (typically 50+ events in the conversion window). It aligns campaign optimization directly with business conversion goals and produces measurably better downstream quality than CPI optimization in most verticals.
ROAS measures the revenue generated for every dollar spent on advertising. It's calculated as:
ROAS = Total Revenue Attributed to Ads ÷ Total Ad Spend
A ROAS of 3.0 means every $1 of ad spend generated $3 in revenue. A ROAS below 1.0 means the campaign is losing money on a direct revenue basis.
ROAS directly connects advertising investment to revenue outcomes. It's the most business-aligned metric in mobile UA and the one that most accurately reflects whether a campaign is profitable.
ROAS has significant measurement challenges in mobile UA:
- Time lag: Most apps don't recover their user acquisition cost within the attribution window. D7 ROAS is rarely positive; D30 ROAS might be; D90 ROAS is where most apps reach payback.
- Attribution gaps: iOS privacy restrictions and MMP attribution windows mean some revenue is inevitably unmeasured.
- Category differences: Apps with subscription models, high-ticket purchases, or delayed monetization need longer measurement windows than apps with immediate in-app purchases.
ROAS optimization works best on mature campaigns with sufficient revenue event data and predictable LTV curves. It's the right target for campaigns that have progressed through CPI and CPA optimization stages and have enough data to train algorithmic models on revenue signals.
CPI, CPA, and ROAS are all trailing metrics — they tell you what happened. LTV (Lifetime Value) is the metric that tells you what a user is worth, which determines how much you can afford to spend to acquire them.
Maximum Sustainable CPI = LTV × Target Payback Margin
If your average user LTV at 90 days is $15 and your target is to recover UA cost within 90 days, you can afford to spend up to $15 per install — regardless of what any other network or competitor is spending. This LTV-driven approach to CPI targeting prevents two common mistakes: overpaying for users who don't monetize, and under-spending on channels that consistently deliver high-LTV users.
Optimizing for CPI throughout the campaign lifecycle. CPI is a starting point, not a destination. Campaigns that never progress to CPA optimization consistently underperform on downstream metrics.
Measuring ROAS too early. Evaluating ROAS in the first 7 days of a campaign produces misleading results in most app categories. Most revenue events occur weeks or months after install.
Using blended metrics instead of source-level data. A blended CPA of $8 across a campaign can hide sources delivering $4 CPA from high-LTV users alongside sources delivering $8 CPA from users who never monetize. Source-level metric analysis is essential.
Ignoring retention as a leading indicator. D1 and D7 retention rates are the fastest available proxies for user quality and LTV. They're available within days of install and predict downstream monetization behavior reliably enough to use as early optimization signals.
Setting CPA targets without LTV data. A CPA target set arbitrarily — "let's try $5 per registration" — may be profitable or catastrophically unprofitable depending on what those registered users are worth. CPA targets should be derived from LTV analysis, not guessed.
- CPI: Volume and reach efficiency. Use for testing and learning phases.
- CPA: Conversion quality. Use as the primary optimization target for most campaigns.
- ROAS: Revenue efficiency. Use for mature campaigns with sufficient revenue data.
- D1/D7 Retention: Leading quality indicator. Monitor continuously at the sub-source level.
- LTV: The foundation for setting all other targets. Measure at D30 and D90 minimum.
What is the difference between CPI, CPA, and ROAS? CPI measures the cost per install, CPA measures the cost per meaningful user action (registration, purchase, deposit), and ROAS measures revenue generated per dollar of ad spend. Each metric is appropriate at a different stage of campaign maturity.
Which mobile marketing metric should I optimize for? For most campaigns, CPA is the right optimization target once sufficient conversion data is available. CPI is appropriate for early testing; ROAS is appropriate for mature campaigns with predictable LTV.
What is a good ROAS for mobile apps? Target ROAS depends heavily on your app category, revenue model, and payback period. Most subscription and high-ticket apps target D90 ROAS of 150-200%. Direct-purchase apps may target D30 ROAS of 100%+.
How do I calculate LTV for mobile app users? LTV is typically calculated as average revenue per user over a defined time window (D30, D90, D365). For subscription apps, LTV = average subscription value × retention rate. For transactional apps, LTV = average transaction value × average number of transactions over the measurement period.
What is the difference between CPA and CAC? CPA (cost per action) refers to the cost of a specific in-app event. CAC (customer acquisition cost) is the total cost of acquiring a paying customer, which may include multiple CPA events across a longer funnel. In mobile UA, CPA is the more commonly used metric at the campaign level.
CPI, CPA, and ROAS each measure a different layer of campaign performance. The teams that scale effectively are the ones who understand which metric to trust at each stage — and who never let a single metric tell the whole story.
At Traffy, we set up full-funnel attribution from install through LTV for every campaign we run, ensuring that optimization decisions are always connected to real business outcomes. If you want help building a metric framework for your mobile UA, talk to our team.