CPI vs CPV: Cost Per Install vs Cost Per View
In short: Both are early-funnel cost metrics pricing cheap, low-commitment actions - CPI prices an app install, CPV prices a video watched past a threshold - and neither correlates well with what happens afterward. CPI-optimized installs frequently never reopen the app again, while CPV-optimized views say nothing about comprehension or purchase intent. They also come from entirely different campaign types: app-install campaigns for CPI, video ad formats for CPV, so they rarely appear side by side in the same report. Use CPI only for app-install campaigns and pair it immediately with a retention metric; use CPV only for video creative and pair it with completion rate.
By the AdFlint research team · Fact-checked against current Google and Meta platform behavior · Last reviewed July 2026
CPI
Ad spend divided by app installs, the average cost of one install attributed to your advertising by the store or measurement partner.
It prices the download and nothing after it. Installs are the cheapest event in a mobile funnel, so campaigns optimized toward this metric reliably find people who install once and never open the app again. Attribution is also contested between networks and SDKs. The common misreading is judging campaigns on it when retention and post-install revenue say the opposite.
Full definitionCPV
Ad spend divided by video views, where the platform's definition of a countable view, a duration or interaction threshold, determines the denominator.
Its entire meaning sits in that threshold. Google counts a view on skippable in-stream at thirty seconds, or the full ad if shorter, or an interaction; feed placements elsewhere count far briefer watches. The same creative therefore reports very different numbers by surface. The frequent misreading is comparing across platforms or formats without checking what each one calls a view.
Full definitionSide by side.
The differences that actually change what happens in your account.
| CPI | CPV | |
|---|---|---|
| What counts as the event | An app install attributed to your ad by the app store or a mobile measurement partner. | A video watched past the platform's countable-view threshold. |
| Applies to which campaign types | App install or app promotion campaigns specifically. | Video ad formats, in-stream or in-feed, across most platforms. |
| Attribution mechanism | Store-level or SDK-based attribution (mobile measurement partners), often contested between networks. | Platform-native view tracking within the ad delivery system itself. |
| What happens after the counted event | Nothing guaranteed - many installs never reopen the app. | Nothing guaranteed - a counted view doesn't confirm sound-on, comprehension, or recall. |
| Cheapest point in its funnel | Yes - installs are typically the cheapest, least committed event in a mobile app funnel. | Relatively cheap, but views still sit above raw impressions in commitment. |
| What it's good for measuring | Whether app-install campaigns are efficiently driving downloads. | Whether video creative is holding attention long enough to deliver its message. |
| Failure mode | Judging install campaigns as successful on CPI alone while retention and post-install revenue tell the opposite story. | Comparing CPV across platforms or formats without checking what each counts as a view. |
What actually separates them.
CPI only exists for app-install campaigns and depends on store or SDK attribution, while CPV exists for any video ad format and depends on the platform's own view-tracking definition.
CPI measures the cheapest, least committed event in a mobile funnel - a download - while CPV measures a somewhat more engaged action, a sustained watch past a set threshold.
Attribution for CPI is frequently contested between ad networks and measurement partners over which one gets install credit; CPV attribution is native to the platform serving the ad and isn't split between competing attributors the same way.
Optimizing toward CPI can actively select for low-intent users who install once and churn, since the algorithm is only rewarded for the download; optimizing toward CPV selects for people who watch longer, which is a marginally stronger, though still weak, intent signal.
CPI has no format restriction - it applies to any ad type driving app installs, image, video, or playable - while CPV is inherently tied to video content and doesn't apply to static formats at all.
Which one should you use?
Use CPI when
- You are running an app-install or app-promotion campaign and need a top-of-funnel efficiency check.
- You are comparing install cost across ad networks or campaigns before layering in retention data.
- You are early in a mobile launch and install volume itself is the immediate goal, with post-install metrics tracked separately.
- You are auditing whether attribution between your ad network and mobile measurement partner is roughly consistent.
Use CPV when
- You are running video creative and want to know if it holds attention past a meaningful watch threshold.
- You are comparing video variants within the same platform and placement where the view threshold is constant.
- You are building awareness with video and using watch-through as a proxy for message delivery, not conversions.
- You want an early signal on whether a video is too long or front-loaded before committing more budget.
Common questions.
Is a low CPI always a good sign for an app campaign?
Not by itself. A low CPI can mean the campaign is efficiently finding installs from an audience that opens the app once and never returns, since the metric only prices the download, not what happens after. Pair it with a retention or post-install event before judging the campaign.
Can CPI and CPV both apply to the same campaign?
Rarely in a meaningful way, since CPI belongs to app-install campaign objectives and CPV belongs to video ad formats; a video ad promoting an app install could technically generate both a view and an install, but most platforms report toward the campaign's primary objective rather than both metrics equally.
Why do my CPI numbers not match between my ad network and the app store?
Install attribution is handled by a mobile measurement partner or SDK, and different networks can claim credit for the same install under different attribution windows and models, especially with last-click or multi-touch rules. Reconcile against one source of truth, usually your measurement partner, rather than trusting a single network's self-reported number.
Why does my CPV vary so much between platforms for the same video?
Each platform defines a countable view differently - a full or extended watch on some skippable in-stream formats versus a much shorter exposure threshold in feed placements elsewhere. The same creative will report very different CPV depending on where it ran, so don't compare the raw numbers across surfaces without checking the definition.
Or stop choosing between them.
AdFlint picks the setting, writes the ads, and keeps optimizing inside the Google and Meta accounts you already own.