Understand CPI, CPA, ROAS and ROI with a clearly labeled example. Learn why cheap installs may not deliver valuable users and how to compare payback windows.
Is install cost, retention or return the main issue? Share metric definitions and the observation window through Telegram to discuss diagnosis.
CPI measures advertising cost per install; it does not show whether a user stays or pays. Assess return alongside ROAS, retention and qualified conversions at the same cohort maturity. Low CPI with weak retention calls for a review of creative expectations and product experience. Sudden revenue drops also require checks for tracking and attribution changes.
CPI is ad spend divided by installs. CPA is spend divided by a specified qualified event. ROAS is attributed revenue within a defined window divided by ad spend. ROI also involves profit and the wider cost base. Define D7 retention precisely: returning on day seven and returning at any point within seven days are different measures.
Suppose cohort A spends $1,000 for 1,000 installs and $300 in attributed D7 revenue: CPI is $1 and D7 ROAS is 30%. Cohort B spends $1,000 for 500 installs and $600 in D7 revenue: CPI is $2 and D7 ROAS is 60%. B has costlier installs but more attributed revenue within that window. These are teaching assumptions, not LLF client results, and they do not establish profitability.
Users acquired today do not have complete D7 outcomes. Compare cohorts using the same windows, currencies and attribution logic. Review sample size, the effect of a few unusually large spenders, refunds and delayed reporting. Where conditions are incomplete, retain uncertainty instead of declaring a winner.
High clicks and weak conversions call for checks of message consistency, loading and events. Low cost with poor retention calls for an examination of audience expectations and early experience. Stable retention with changing revenue requires payment-path and event checks. Once evidence supports an action, record the creative or budget change for review.
It depends on net revenue, margins, the payback window, platform shares and other costs. There is no universal threshold for every game.
Consider user quality, revenue maturity and sample size. Higher CPI may acquire more valuable users or simply be inefficient; comparable measurement is needed.
Is install cost, retention or return the main issue? Share metric definitions and the observation window through Telegram to discuss diagnosis.