A cheap CPI does not mean a profitable campaign. What decides whether a mobile game or app campaign makes money — and how far you can scale it — is LTV (Lifetime Value).
In this guide you will learn how to calculate LTV, how to track it with analytics, how to predict it early, and how to use it to scale user acquisition (UA) campaigns. Every step includes a worked example.
What is LTV and why does it matter?
LTV is the total revenue one installed user generates over their lifetime in your app or game — from ads, in-app purchases (IAP) and subscriptions.
Profitable UA ⇔ LTV > CPI
What a user pays you must be more than what you pay to acquire them.
- LTV : CPI below 1 — you lose money on every install, and scaling makes it worse.
- LTV : CPI above 1.2 — you have margin, and budget becomes your growth lever.
Method 1: Retention × ARPDAU (the quick formula)
LTV = ARPDAU × Lifetime Days
Lifetime days = 1 (install day) + D1 + D2 + … + Dn retention
Example — a hybrid-casual puzzle game:
| Day | D1 | D3 | D7 | D14 | D30 |
|---|---|---|---|---|---|
| Retention | 40% | 23% | 15% | 11% | 7% |
- Lifetime over 30 days (sum of the full daily retention curve): 4.8 active days per user
- ARPDAU (ads + IAP ÷ daily active users): $0.12
- 4.8 × $0.12 = $0.58 LTV at D30
Method 2: Cohort revenue (what UA teams actually use)
Cumulative ARPU (Dn) = Cohort revenue by Dn ÷ Installs
Example: a fully matured March cohort of 10,000 installs from the same game.
| Day | Cohort revenue | ARPU per install |
|---|---|---|
| D0 | $1,200 | $0.12 |
| D3 | $2,300 | $0.23 |
| D7 | $3,100 | $0.31 |
| D14 | $4,100 | $0.41 |
| D30 | $5,800 | $0.58 |
| D90 | $9,500 | $0.95 |
At D30 this gives the same answer as Method 1 — but it comes straight from your data, so you can split it by campaign, country and platform.
Predicting LTV early with multipliers
You can not wait 90 days to decide on a campaign. Instead, use a multiplier from your matured cohorts.
Predicted D90 LTV = D7 ARPU × Multiplier
Multiplier = D90 ARPU ÷ D7 ARPU, taken from mature past cohorts
- From the March cohort: $0.95 (D90) ÷ $0.31 (D7) = 3.0× multiplier
- New campaign on day 7: D7 ARPU $0.28 × 3.0 = $0.84 predicted D90 LTV
Important: build a separate multiplier per platform, geo and ad network. An iOS Tier-1 user and an Android Tier-3 user decay very differently.
Hybrid monetization: split LTV into Ads + IAP
| Revenue source | How it is calculated | D90 LTV |
|---|---|---|
| Ad LTV | Impressions per user × eCPM ÷ 1000 (tracked via ILRD) | $0.57 |
| IAP LTV | Payer conversion × net ARPPU (after store fee): 2.5% × $15.20 | $0.38 |
| Total | $0.95 |
Why split it?
- Ad revenue shows up fast (D0–D7), while IAP revenue has a long tail.
- An IAP-heavy network can look weak at D7 and strong at D60.
- It tells you the right bidding goal: ad-ROAS or IAP-ROAS.
How to track LTV with analytics
- MMP (attribution): AppsFlyer, Adjust or Singular show which network, campaign and creative brought each user. Use SKAN / AdAttributionKit for iOS.
- Ad revenue SDK (ILRD): AppLovin MAX, LevelPlay or AdMob send impression-level ad revenue to your MMP and Firebase.
- Product analytics: Firebase / GA4 events such as
tutorial_complete,level_complete,ad_impressionandpurchase(with value and currency). - Warehouse and dashboard: BigQuery → Looker Studio or Tableau for cohort ARPU, retention and ROAS by campaign × geo × day.
From LTV to ROAS: the daily metric
D7 ROAS = D7 ARPU ÷ CPI
Break-even D7 ROAS = 1 ÷ Multiplier (for example, 1 ÷ 3.0 = 33%)
| Campaign | CPI | D7 ROAS | Multiplier | Break-even | Pred. D90 ROAS | Decision |
|---|---|---|---|---|---|---|
| A · US Android | $0.80 | 39% | 3.0× | 33% | 117% | ✅ Scale |
| B · BR Android | $0.25 | 36% | 2.6× | 38% | 94% | ⚠️ Optimize |
| C · UK iOS | $1.20 | 26% | 3.4× | 29% | 88% | ❌ Cut |
Campaign B has the cheapest CPI and still misses break-even. Campaign C has the highest-value users and still loses money. The only real win is CPI below LTV.
When and how to scale a campaign
- Prove it first: early ROAS (D0/D3) tracking above target for 3–5 cohorts, with enough volume (around 100+ installs a day).
- Scale in steps: raise budget 20–30% every 2–3 days. Big jumps reset the algorithm’s learning.
- Upgrade the bid goal: CPI → AEO (event) → tROAS / value bidding once you have around 50+ conversion events per week per campaign.
- Expand horizontally: new geos, networks and lookalikes — each with its own ROAS target.
- Feed creatives: refresh every 1–2 weeks. Creative fatigue is the most common reason ROAS drops while scaling.
Worked example: scaling Campaign A (+25% every 3 days)
| Step | Daily budget | CPI | Pred. D7 ROAS* | Action |
|---|---|---|---|---|
| Day 1 | $500 | $0.80 | 39% | Scale |
| Day 4 | $625 | $0.82 | 38% | Scale |
| Day 7 | $780 | $0.85 | 36% | Scale |
| Day 10 | $975 | $0.89 | 35% | Hold |
| Day 13 | $1,220 | $0.97 | 32% | Step back |
CPI rises as you scale into less-engaged audiences. At $1,220 a day the campaign drops below its 33% break-even, so go back to $975, add fresh creatives, and push again. The result: about 2× the budget, still profitable.
5 common LTV mistakes
- Using only blended LTV. Averaging all geos hides the campaigns that burn money.
- Forgetting ad revenue. Without ILRD, hybrid-casual LTV is heavily understated.
- Mixing gross and net. Remove store fees (15–30%) and VAT from IAP before comparing to CPI.
- Using old multipliers. Refresh them with every newly matured cohort — updates change the curve.
- Judging too early. Do not kill an IAP-heavy campaign on D1 ROAS alone.
Quick recap
- Calculate: ARPDAU × lifetime days, or cohort ARPU
- Track: MMP + ILRD + Firebase events + BigQuery
- Predict: D7 ARPU × multiplier = D90 LTV
- Scale: +20–30% steps while ROAS stays above break-even
All numbers in this post are illustrative examples for learning purposes.
Over to you 👇
💬 What D7 ROAS target do you use for your games? Share it in the comments below.
👥 Know someone who works in UA or growth? Send this post to them.
♻️ Found it useful? Share it with your team on LinkedIn, X or WhatsApp.
