Core Premise
"UA Accretive" ≠ Company Profitable
UA's Job
- + Generate positive contribution margin on spend
- + Find and scale profitable channels
- + Optimize creative performance
- + Report data accurately and quickly
NOT UA's Job
- - Cover fixed costs
- - Extend LTV curve
- - Fix product problems
- - Make the company profitable
The Decision Tree
LTV shape is the input you can't control. Everything else flows from it.
Input
Product LTV Curve
→
Gates
Channel Viability
→
Tactics
Channel Execution
→
Fuel
Creative Strategy
Focus on Open Doors
Your LTV shape determines which channels are viable. Don't waste budget "training" channels that can't work with your curve. Focus all energy on maximizing the doors that are open.
1
Know Your LTV Shape
Your LTV curve determines which channels can work
Click a state to see channel recommendations for that LTV shape:
Current State
Front-Loaded
70% of LTV by D7. Curve flattens after D30. Need fast feedback loops.
Viable: AppLovin, Kashkick, Unity (test)
Target State
Mid-Weighted
Breakeven by D30. Growth continues to D60. Opens more doors.
Unlocks: + Unity scale, Meta test
Goal State
Extended
D60 > D30 × 1.2. Sustainable growth to D90+. All channels viable.
Unlocks: + Google, Meta scale, iOS
Matching Feedback Loops to LTV Curves
| Channel | Optimization Window | Best For | Avoid For |
|---|---|---|---|
| AppLovin | D1-D7 signal | Ad-monetized, casual games | — |
| Unity Ads | D1-D7 signal | Similar DNA to AppLovin | — |
| Meta/Facebook | D1-D7 signal | Broad audiences, creative testing | — |
| 7-42 days | IAP-heavy, extended LTV curves | Front-loaded ad-monetized games | |
| Incentivized | Immediate signal | Casual/hypercasual, volume at low CPI | Retention-dependent games |
Attribution Windows Reference
Industry-standard attribution and learning periods by network (as of Dec 2025). See ATTRIBUTION_WINDOWS_REFERENCE.md for sources.
| Network | Click Attribution | View Attribution | Learning Period |
|---|---|---|---|
| AppLovin | 7 days | 1 day | Real-time (AXON) |
| Unity Ads | 7 days | 1 day | ~7 days |
| Meta/Facebook | 7 days (max) | 1 day (max) | ~7 days |
| 30 days (default) | Configurable | 7 days - 6 weeks | |
| Incentivized | Immediate | N/A | Immediate |
Google: Tested and Not Viable for Current LTV Shape
We tested Google App Campaigns and they did not work for our front-loaded LTV curve. Google's learning period (7 days to 6 weeks depending on platform) and optimization for longer payback windows make it unsuitable for ad-monetized games where 70%+ of LTV occurs by D7. Revisit only after product delivers LTV extension (D60 > D30 × 1.2).
Phase Transition Criteria
| Transition | Criteria | Owner |
|---|---|---|
| Front-Loaded → Mid-Weighted | D60 ROAS > D30 ROAS × 1.1 for 3 consecutive cohorts | Product (elder game features) |
| Mid-Weighted → Extended | D60 ROAS > D30 ROAS × 1.2 for 3 consecutive cohorts | Product (monetization depth) |
Phase Transitions Are Product Metrics
UA cannot fix the LTV curve. Phase transitions require product changes (elder game content, monetization features, retention mechanics). UA's job is to maximize the current phase, not to force a transition.
2
Channel Unlock Matrix
LTV shape determines which doors are open
| LTV Shape | AppLovin | Kashkick | Unity | Meta | |
|---|---|---|---|---|---|
| Front-Loaded (Selected) | PRIMARY | PRIMARY | TEST | LOCKED | LOCKED |
| Mid-Weighted | SCALE | MAINTAIN | SCALE | TEST | TEST |
| Extended | MAINTAIN | REDUCE | SCALE | SCALE | SCALE |
Viewing: Front-Loaded
AppLovin
PRIMARY
AXON thrives on D0-D7 signals and ad ROAS
Min spend: $10-15K/week for 10+ payers/day
Optimization: Total ROAS (Ad + IAP)
Kashkick
PRIMARY
Immediate feedback loop, incentivized users
Guardrail: D1 retention >15% or pause
Watch: Time-to-Complete for fraud
Unity Ads
TEST
Similar DNA to AppLovin, different audience
Entry: AppLovin stable at 50%+ D7
Structure: ROAS D7 campaign
Google App Campaigns
LOCKED
Tested - did not work for front-loaded LTV
Unlock: D60 > D30 × 1.2
Warning: Do not spend to "train"
Meta/Facebook
LOCKED
Requires creative velocity we can't support yet
Unlock: 4-5 creatives/week capacity
Prerequisite: Mid-weighted LTV
Don't Waste Budget on Locked Channels
Spending $40K to "train" Google when your LTV shape can't support it is burning money. Focus all energy on maximizing PRIMARY channels until product delivers an LTV extension.
3
Channel Execution
How to maximize each open channel
AppLovin Playbook
Campaign Goal: Total ROAS (Ad Revenue + IAP)
Do NOT split into separate Ad ROAS and IAP ROAS campaigns at this budget level. One unified campaign. Let AXON optimize for total value.
Kashkick Playbook
Incentivized Traffic Quality Warning
Incentivized users are task-oriented. They often ignore ads or churn immediately after reward. Watch retention closely.
Unity Playbook (When Ready)
Core KPIs & Thresholds
| Metric | Target | Floor (Pause If Below) | Notes |
|---|---|---|---|
| D7 ROAS | >50% | <40% | Primary health metric |
| D30 ROAS | >100% | <80% | 143% needed for post-fee breakeven |
| IPM | >15 | <8 | Creative north star |
| CPI | <$3.00 | >$5.00 | By geo (T1 vs T2) |
| D1 Retention (Kashkick) | >20% | <15% | Fraud/quality signal |
4
Creative Engine
Strategy → Formats → Production → Testing
4.1 Strategy & KPIs
North Star: IPM (Installs Per Mille)
IPM directly determines your effective CPI. High IPM = lower bid required = better ROAS. On AppLovin/Unity, IPM is the primary signal AXON uses to optimize delivery.
Competitor Intelligence (Weekly)
Use ad intelligence tools (SensorTower, AppMagic, Facebook Ad Library) to identify top-performing creatives in the casual genre on AppLovin/Unity. Clone and adapt winning concepts rather than inventing from scratch.
4.2 Format Strategy (AppLovin/Unity)
For SDK networks (AppLovin, Unity), prioritize interactive formats. Static rarely scales for casual games.
Video + IEC
50%
15-25s gameplay + end card
Playables
45%
HTML5 mini-games
Static
5%
Native banners only
| Format | When to Use | Technical Specs |
|---|---|---|
| Video + IEC | Primary format. Video qualifies user, IEC converts. | 15-25s video, <10MB total |
| Full Playable | High-engagement inventory. Best IPM drivers. | <5MB for fast load (critical for IPM) |
| IEC (Interactive End Card) | Attached to every video. Simple 1-click interaction. | <2MB, loads after video |
| Static | Low-cost testing, native banner inventory only | Standard IAB sizes |
Never Run Video Without an IEC on AppLovin
A video without an Interactive End Card is wasted budget. The IEC is where conversion happens. Video alone has 50-70% lower conversion rates.
4.3 Messaging Strategy
For SDK networks, gameplay-focused messaging outperforms conceptual hooks. Prioritize these patterns:
Gameplay Fail
"Harder than it looks" moments
Triggers "I can do better" instinct
Progression Fantasy
High-level skins, power, rewards
Shows aspirational end state
Comparative Difficulty
"Noob vs Pro" gameplay
Social proof + challenge
Simplified Mechanics
Hypercasual presentation of core loop
Broadest audience appeal
Playable/IEC Optimization
80% of playable installs come from the end card, not gameplay. Optimize the fail state + CTA. The playable should frustrate users into clicking, not satisfy them. High-contrast CTA button is critical.
4.4 Production Engine
Weekly Velocity
3-4
new concepts
Test Budget
20%
80% on winners
Kill Threshold
<70%
of benchmark IPM @ 2k impr
Cost Efficiency (Survival Phase)
At $10-15K/week, agency pricing is not viable. Use these alternatives:
| Method | Best For | Cost Advantage |
|---|---|---|
| Creative Resurrection | Re-skin winners (new first 3s, colors, CTA) | Uses proven mechanics, lowest risk |
| Competitor Cloning | Adapt top performers from ad intelligence | Pre-validated concepts |
| UGC Platforms (Billo, etc.) | Video creatives with human element | 10x velocity at 1/10th agency cost |
| AI Tools (Midjourney, Runway) | Motion backgrounds, variants, voiceovers | Near-zero marginal cost |
4.5 Testing Protocol
4.6 Kashkick Creative (Incentivized)
Different Creative Strategy for Incentivized Traffic
Kashkick users are paid to play. Your creative must show how to progress quickly to the milestone, not just "fun" gameplay. Explicitly show the path to reward completion.
5
Reporting & Cadence
What to track, when to report, who owns what
Weekly Reporting Cadence
| Report | Frequency | Owner | Audience |
|---|---|---|---|
| UA Performance Dashboard | Weekly (Monday) | UA Manager | CEO, Transcend |
| Creative Performance | Weekly (Monday) | UA Manager | Creative Team |
| Cohort ROAS Update | Bi-weekly | UA Manager | CEO, Transcend |
| Channel Deep Dive | Monthly | UA Manager | CEO, Transcend |
Required Dashboard Metrics
Daily
Spend, Installs, CPI
By channel
Weekly
D7 ROAS, IPM
By channel + creative
Bi-Weekly
D14, D30 ROAS
Cohort view
Monthly
LTV Curve Shape
Phase assessment
Escalation Triggers
Immediate Escalation (Same Day)
D7 ROAS drops below 40% for 3+ consecutive days | Fraud detected (TTC anomaly) | Channel paused by platform
Weekly Review Flag
D7 ROAS between 40-50% | CPI rising >20% week-over-week | IPM declining on top creatives
Scale Signal
D7 ROAS >60% for 2+ weeks | New creative IPM >20 | D30 ROAS trajectory toward 100%+
RACI Matrix
| Decision | Fairy Dragon UA | Fairy Dragon CEO | Transcend |
|---|---|---|---|
| Daily bid/budget adjustments | R/A | I | I |
| New creative launch | R/A | I | I |
| Channel pause/resume | R | C | A |
| New channel test ($5K+) | R | C | A |
| Budget increase >20% | R | A | C |
| Phase transition assessment | I | C | R/A |
R = Responsible (does the work) | A = Accountable (final decision) | C = Consulted | I = Informed