How to Lower CPI for a Mobile Game (Without Buying Worse Users)

Cost per install is the number every mobile game marketer is asked about first, and the one most often improved in ways that cost money later. Lowering CPI is worth doing — every saving lets the same budget reach more players — but only if the players stay the same. This guide goes through the levers that genuinely reduce cost per install, in roughly the order they pay off, and the checks that stop a cheaper install from becoming a worse one.

What CPI is made of

On impression-priced inventory, CPI is the price of reaching people divided by how many of them install. Written out:

Even when you bid on a cost-per-install basis, the same arithmetic applies behind the scenes: the network estimates how many installs your ad will produce from an impression and bids accordingly. A creative with a higher IPM lets the network win impressions at your target price more often, which is why the creative is usually the biggest lever of all. IPM and CPI compared across formats goes deeper on how to read the two together.

A worked illustration

The numbers here are invented to show the mechanics, not to suggest what you should expect. Say a campaign buys impressions at $10 per thousand and its creative produces 5 installs per thousand impressions: CPI is $2. A new creative that produces 8 installs per thousand on the same inventory brings CPI down to $1.25 without any change in price. Cutting the price of impressions by a fifth instead, with the old creative, only reaches $1.60 — and usually means accepting worse inventory to get it. That is why most teams look at the install rate before they look at the bid.

Lever 1: the creative

Most of the variance in CPI between campaigns comes from the ads. A better creative raises IPM, and a higher IPM lowers CPI on the same inventory.

• Hook in the first seconds. Viewers decide almost immediately; show the most compelling moment of the game straight away. See the first three seconds.
• Show the real game. Footage or interaction from the actual product converts more honestly than a staged concept.
• Use more than one format. Video, playable and interactive end card each reach viewers differently and qualify for different placements.
• Make the call to action obvious and reachable. A convinced viewer who cannot find the button is a lost install.
• Refresh before fatigue. Every creative's performance decays as an audience sees it repeatedly; creative fatigue covers how to spot and plan for it.

Lever 2: store page conversion

Every install passes through a store page. If the page converts poorly, you pay for clicks that do not become installs, and some networks also learn that your traffic converts badly.

• Match the page to the ad. A viewer who tapped on a puzzle mechanic should land on screenshots of that mechanic.
• Use custom product pages on iOS and custom store listings on Google Play to give each creative theme its own page. Custom product pages with playable ads shows how.
• Test the icon, the first screenshots and the preview video; most visitors never scroll.
• Keep ratings healthy and reply to reviews; they are visible at the moment of decision.

Lever 3: bidding and optimisation goal

The campaign settings decide what the network is trying to buy. Optimising for installs gets the most installs for the money; optimising for an in-app event or for value gets fewer, more expensive installs that are worth more. A lower CPI on an install-optimised campaign is not automatically better than a higher one on a value-optimised campaign.

• Let new campaigns finish learning before judging them; early CPI is noisy.
• Avoid frequent bid changes, which restart the learning and raise cost.
• Raise bids where the cohorts are worth it, rather than cutting bids everywhere.

Lever 4: geos and audiences

Installs cost very different amounts in different countries, and so do the players. Lower-cost markets can deliver volume cheaply, and they suit ad-monetised games or soft launches; higher-cost markets often bring higher value per player. Lowering blended CPI by moving spend to cheaper countries is only a gain if those countries' players still pay back for your monetization model.

Localising the creative is a cheaper and less risky lever than changing countries: an ad in the viewer's language tends to convert better in the same market. Localising playable ads covers what to translate.

Lever 5: placements and networks

Inventory differs by network and by placement. Rewarded placements, interstitials, social feeds and search all deliver different users at different prices, and each network sees a different slice of the market.

1. Check where installs actually come from inside each network's reporting, by placement and app where available.
2. Make sure your creatives qualify for the placements you want — many in-app placements carry playables, which a video-only campaign cannot enter.
3. Expand to networks where your game's audience is under-served rather than doubling spend where you already saturate.
4. Exclude inventory that produces installs with no engagement at all.

Lever 6: a testing system

Lower CPI comes from a steady flow of new creatives tested against the current best, not from a single good ad. Keep a control, change one thing at a time, give each variant enough spend to reach a conclusion, and move on. The creative A/B testing guide covers sample size and what to test first.

The trap: cheap installs that are bad users

The fastest way to lower CPI is to show people something more exciting than your game: a mechanic it does not have, a fail state that is not in it, a reward that never comes. It works on the first metric and fails on every one after it. Players who were promised a different game leave quickly, ROAS drops, and networks increasingly treat misleading creative as a policy violation. Misleading ad policy explains where the lines are.

So pair every CPI improvement with a quality check:

• Day-one and day-seven retention by creative, not just by campaign.
• Early in-app events — tutorial complete, first level cleared.
• Cohort ROAS at a fixed day; see ROAS and LTV for mobile games.
• Cost per retained user, which combines both: spend divided by players still active at day seven.

If CPI fell and cost per retained user rose, the campaign got worse.

Where playable creative fits

Playables sit at the intersection of the two goals. They tend to raise install intent among people who enjoy the loop and filter out people who do not, because the viewer has already played before deciding. That can mean a lower cost per retained user even in cases where a video wins on raw CPI — which is why they should be judged on the deeper numbers.

The obstacle has usually been production cost per variant. Playable Ads Maker removes most of it: build in the browser from a game template or from scratch, export a ready-to-upload file for every major network, and produce enough variants to keep a real testing cadence. The template library is the quickest place to start.