WEPR INSIGHTS

App Growth: Should You Optimise for CPI or Cost per Activated User?

CPI is a useful buying-efficiency starting point, not a business outcome. Learn how to define activation, build a cost funnel and choose the right optimisation target by product stage.

Low CPI does not automatically mean app growth is working. If people download but do not register, complete a meaningful action or retain, low CPI may simply buy low-value installs. A higher CPI may produce better users. The useful question is whether the cost metric fits product stage and business model.

Define activation first

Activation should represent early product value, not just registration: completing setup, creating a first project, placing a first order, using the product on three days, reaching meaningful consumption or inviting a collaborator. Product, growth and sales should agree on the definition and keep it stable long enough for comparison.

Build a cost funnel

Use spend → click → install → first open → registration → activation → retention → payment or qualified lead. Track count, conversion rate and cost at every layer. CPI is spend divided by installs; cost per activated user is spend divided by users meeting the activation definition. Do not compare different attribution windows, markets or event definitions as if they were identical.

When CPI helps and when activation should lead

During early creative, channel and install-flow tests, CPI can reveal expensive clicks, mismatched store pages or installation failure. Once install volume is stable, optimising only to CPI teaches a platform to find the easiest download rather than the person likely to retain. Shift gradually to activation, trial, first order or qualified lead when the key event is defined, the MMP can return it and the sample is large enough. Use an observation window appropriate to the channel and sales cycle.

Include product experience

Low activation is not always a media problem. Check whether ad promise, store screenshots, first open, registration, permissions and onboarding agree. A five-minute promise followed by a long form distorts quality assessment.

Weekly review

By channel, country, device and creative, report spend, clicks, installs, registrations, activations, day-seven retention and cost at each stage. Label attribution window and data freshness, and sample-check MMP, ad-platform and product records. Treat high-CPI/low-activation, low-CPI/low-retention and high-CPI/high-value outcomes differently. WEPR agrees activation definition and attribution path before discussing budget expansion.