How Cost Per Action Works
Cost Per Action is a performance-based pricing model where advertisers pay only when a user completes a specific, predefined action inside the app. Unlike impression-based or click-based models, CPA shifts the financial risk from the advertiser to the publisher or ad network. You are not paying for eyeballs or taps, you are paying for outcomes.
The "action" in CPA is flexible and defined by the advertiser. For an e-commerce app, it might be a first purchase. For a fintech app, it could be completing KYC verification. For a subscription app, it might be starting a free trial. The specificity of the action is what makes CPA powerful, it aligns ad spend directly with business outcomes rather than vanity metrics.
CPA operates downstream from CPI in the conversion funnel. A user first sees an ad (impression), clicks it, installs the app, and then completes the target action. Each step introduces friction and drop-off, which is why CPA is always higher than CPI for the same campaign. The gap between your CPI and CPA reveals how effectively your app converts new installs into valuable users.
Calculating and Benchmarking CPA
The CPA formula is straightforward: divide your total campaign spend by the number of completed target actions. If a campaign costs $10,000 and drives 500 registrations, your CPA is $20 per registration. But the simplicity of the formula masks the complexity of getting the inputs right.
Accurate CPA calculation depends entirely on reliable attribution. You need to know which actions were driven by which campaigns, and that requires proper post-install event tracking with correct attribution windows. If your attribution setup miscounts conversions, either over-attributing organic actions to paid campaigns or missing legitimate conversions, your CPA numbers will be misleading and your optimization decisions will be wrong.
Benchmarking CPA requires context. A $50 CPA is excellent if your average customer generates $500 in lifetime value, but it is unsustainable if your LTV is $60. Industry benchmarks provide a starting reference, gaming apps typically see CPAs of $5–$15 for in-app purchases, while fintech apps might see $50–$200 for account funding events. But your own LTV-to-CPA ratio is the metric that actually matters. A healthy ratio is typically 3:1 or higher, meaning your LTV should be at least three times your CPA.
CPA vs Other Pricing Models
Understanding where CPA fits among other pricing models helps you choose the right approach for each campaign objective. CPM (Cost Per Mille) charges per thousand impressions and works best for brand awareness campaigns where reach matters more than conversions. CPC (Cost Per Click) charges per ad click and suits campaigns focused on driving traffic. CPI (Cost Per Install) charges per app install and is the most common model for user acquisition.
CPA sits at the bottom of this funnel and carries the highest per-unit cost but the lowest risk. When you pay on a CPA basis, you are guaranteed that every dollar spent resulted in a meaningful user action. This makes CPA ideal for performance-focused campaigns where you have clear revenue targets and need predictable unit economics.
The trade-off is volume. Ad networks and publishers prefer models that pay them earlier in the funnel because conversion rates at each stage reduce their earning potential. A network might deliver 100,000 impressions that generate 1,000 clicks, 200 installs, and 20 purchases. On a CPM basis they earn on all 100,000 impressions. On a CPA basis they earn only on the 20 purchases. This means CPA campaigns often receive lower priority in ad auctions and may struggle to scale compared to CPI or CPC campaigns.
Optimizing CPA for Mobile Growth
Reducing CPA is not just about negotiating better rates with ad networks, it requires optimizing the entire funnel from impression to conversion. Start with targeting. The most impactful CPA reduction comes from reaching users who are more likely to convert. Use lookalike audiences based on your highest-value existing users, and leverage post-install event data to train ad network algorithms on what a valuable user looks like.
Creative optimization is the second lever. Ad creatives that accurately represent your app's value proposition attract users with genuine intent, which improves post-install conversion rates and lowers CPA. Misleading creatives might lower your CPI but will inflate your CPA as unqualified users churn before completing the target action.
Linkrunner helps growth teams connect the dots between ad spend and post-install actions with precise attribution across every campaign and channel. By tracking the full journey from click through install to conversion event, you can identify which campaigns, creatives, and audiences deliver the lowest CPA, and reallocate budget accordingly. This closed-loop measurement is essential for CPA optimization because without it, you are optimizing blind.
CPA in a Privacy-First Landscape
The shift toward privacy-first measurement frameworks like SKAdNetwork and Privacy Sandbox has complicated CPA tracking significantly. These frameworks limit user-level data, making it harder to attribute specific post-install actions to specific ad clicks. Growth teams need to adapt their CPA measurement strategies accordingly.
On iOS, SKAN's conversion value schema allows you to encode post-install events, but the limited bit space means you cannot track every possible action with full granularity. Prioritize encoding your primary CPA event, the action that most directly correlates with revenue, and use coarse conversion values for secondary events. Timer extensions in SKAN 4.0 and later provide additional windows to capture actions that happen days after install.
Probabilistic modeling becomes more important when deterministic attribution is restricted. By analyzing aggregate patterns, install timing, campaign-level conversion rates, and cohort behavior, you can estimate CPA at the campaign level even without user-level attribution. Combine this with media mix modeling to validate your CPA estimates against actual revenue outcomes. The teams that build robust measurement stacks combining deterministic signals, probabilistic models, and incrementality testing will maintain accurate CPA visibility regardless of platform-level privacy changes.
