AgentSkillsCN

tiktok-ads-optimization

针对移动应用安装推广活动,提供 TikTok 广告的最佳实践。在创建、优化或分析 TikTok 广告活动时使用。提供预算分配、活动结构、定向投放、出价策略、创意轮播,以及 KPI 指标等方面的指导。

SKILL.md
--- frontmatter
name: tiktok-ads-optimization
description: TikTok Ads best practices for mobile app install campaigns. Use when creating, optimizing, or analyzing TikTok ad campaigns. Provides guidance on budget allocation, campaign structure, targeting, bidding, creative rotation, and KPI benchmarks.

TikTok Ads Optimization Skill

This skill provides comprehensive guidance for running TikTok ads for mobile app installs, especially with limited budgets ($20-100/day).

When to Use This Skill

  • Creating new TikTok ad campaigns
  • Analyzing campaign performance
  • Optimizing underperforming campaigns
  • Setting up targeting and bidding strategies
  • Deciding on creative rotation schedules

Core Principles

1. Budget Concentration > Distribution

Never split limited budgets across multiple campaigns/ad groups.

Budget/DayStructure
< $301 Campaign, 1 Ad Group (mandatory)
$30-501 Campaign, 1-2 Ad Groups
$50-1001 Campaign, 2-3 Ad Groups
$100+Test CBO with 3-5 Ad Groups

Why: TikTok requires 50 conversions in 7 days to exit learning phase. Splitting budget = "zombie campaigns" that never learn.

2. Learning Phase Math

code
Minimum Daily Budget = 10 × Target CPI

Example:
- Target CPI: $4
- Minimum Budget: $40/day
- Time to 50 conversions: 12-13 days

3. Low Budget Strategy: Click → Install Progression

For budgets under $50/day:

PhaseDurationOptimization GoalPurpose
Phase 1Week 1-2ClickBuild pixel data, find winning creatives
Phase 2Week 3+App InstallUse learned data for efficient installs

Rationale: Click optimization collects 10x more data at same budget, teaching the algorithm who your audience is.

Campaign Settings Reference

Campaign Level

SettingRecommended ValueNotes
ObjectiveApp Promotion → App Install-
Budget TypeDaily BudgetNot Lifetime
Campaign StructureManual (not Smart+)Smart+ needs $200+/day

Ad Group Level

SettingRecommended ValueNotes
Optimization GoalClick (Phase 1) or Install (Phase 2)See progression strategy
Bidding StrategyMaximum Delivery (Lowest Cost)For discovery phase
TargetingAutomatic (Broad)Narrow targeting fails with limited budget
Age18-44 (or custom)Let algorithm optimize within range
GenderAllDon't restrict
LocationSingle country or regionDon't split by country
PlacementTikTok OnlyExclude Pangle for quality
ScheduleAlways OnLet algorithm optimize delivery time

Why Broad Targeting?

TikTok official data shows:

  • CPA 15% lower with broad targeting
  • CVR 20% higher than interest-based targeting

Limited budget + narrow targeting = insufficient data = poor optimization.

Bidding Strategy Progression

code
Week 1-2: Maximum Delivery (Lowest Cost)
    ↓ Gather 50+ conversions
Week 3+: Cost Cap (1.2× baseline CPA)
    ↓ If delivery drops
Increase Cost Cap by 10-20%
StrategyWhen to Use
Maximum DeliveryDiscovery phase, unknown CPA
Cost CapAfter baseline established
Bid CapStrict budget control (limits delivery)

Creative Management

TikTok Fatigue is 4x Faster Than Meta

MetricTikTokMeta
Peak PerformanceDays 1-7Days 1-14
Fatigue OnsetWeek 2Week 3-4

Creative Rotation Schedule

Budget/DayActive CreativesRotation FrequencyMonthly Production
< $503-5Every 7 days4-6 new concepts
$50-1005-8Every 5-7 days8-12 new concepts
$100+8-15Every 3-5 days15-20 new concepts

Kill Criteria (48-72 hours)

MetricKill ThresholdAction
Hook Rate (2-sec views)< 20%Kill immediately, remake hook
CTR< 1%Kill after 48hrs
CPI> 2× targetKill after 72hrs
Zero conversionsAfter 5 daysKill

Let It Run Criteria

MetricConditionAction
CPI1.2-1.5× target, improving trendRun 3 more days
High Hook + CTR, low CVRCVR < 5%Keep ad, fix App Store page

KPI Benchmarks (2025-2026)

CTR (Click-Through Rate)

Optimization GoalPoorAverageGoodExcellent
App Install< 1%1.0-1.5%1.5-2.0%> 2.0%

CPI (Cost Per Install)

App CategoryAverage CPITarget
Utilities/Tools$1.50< $2
Gaming$2.00< $2.50
Health/Fitness$2.50< $3
Subscription Apps$4.00< $5
Self-Improvement$3-5< $4

CVR (Conversion Rate)

MetricPoorAverageGood
Install Rate (click → install)< 5%6-10%> 10%

Other Metrics

MetricBenchmark
Hook Rate> 30%
Hold Rate (6-sec/2-sec)> 8%
CPC$0.30-0.80
CPM$3-10
Frequency (cold audience)< 2.5

Smart+ vs Manual Campaign

AspectSmart+Manual
Learning Period7 days3 days
Budget Requirement$200+/day$20+/day
CPI39% lowerHigher but controllable
User QualityLowerHigher
ControlMinimalFull

Decision: Use Manual for budgets under $200/day.

Geographic Strategy

Single Language Campaigns

LanguageCountriesStructure
JapaneseJapan only1 Ad Group
EnglishUS + UK + CA + AU1 Ad Group (combined)

Never split by country with limited budget - it fragments data.

Why Not Split English Countries?

StructureBudget/CountryLearning Time
Split (4 Ad Groups)$10 eachNever exits learning
Combined (1 Ad Group)$40 total~12 days to exit

Troubleshooting

Campaign Not Delivering

SymptomCauseFix
0 impressionsBudget too low, targeting too narrowIncrease budget or broaden targeting
Stuck in reviewPolicy violationReview creative/landing page
Low deliveryBid too low (Cost Cap)Increase bid 10-20%

CPI Too High

CPI vs TargetAction
1.2-1.5×Wait 3 more days
1.5-2×Test new creatives
> 2×Kill campaign, restart with different approach

CTR Too Low

CTRDiagnosisFix
< 0.5%Hook brokenRemake first 2 seconds
0.5-1%Weak CTATest different CTAs
1-1.5%AcceptableContinue testing

Weekly Review Checklist

DayCheckAction
DailySpend pace, delivery issuesFix immediately
Day 3Hook Rate, CTR by creativeKill losers
Day 7CPI, CVR, learning phase statusScaling decision
WeeklyCreative fatigue, frequencyRotate creatives

Example Campaign Structure

Phase 1: Discovery (Week 1-2)

code
Campaign: anicca-jp-click-discovery
├── Budget: 6,000 JPY/day
├── Bidding: Maximum Delivery
└── Ad Group: broad-jp-click
    ├── Optimization: Click
    ├── Targeting: Automatic (Broad)
    ├── Age: 18-44
    ├── Location: Japan
    ├── Placement: TikTok Only
    └── Creatives: 3-5 variations

Phase 2: Install Optimization (Week 3+)

code
Campaign: anicca-jp-install-scale
├── Budget: 6,000-10,000 JPY/day
├── Bidding: Cost Cap (1.2× Phase 1 CPA)
└── Ad Group: broad-jp-install
    ├── Optimization: App Install
    ├── Targeting: Automatic (Broad)
    ├── Creatives: Winning 2-3 from Phase 1 + 2 new tests

Sources

  • TikTok Ads Manager Official Documentation
  • TikTok Business Help Center
  • Industry benchmarks from Varos, AdBacklog, Digital Eagle (2025)
  • TikAdTools best practices
  • Thesis A/B testing data

Last updated: 2026-01-28