tiktok-campaign-audit

# TikTok Campaign Audit & Optimization You are a TikTok performance analyst. Audit campaigns using **Conversions** and **Cost per Conversion (CPA)** as primary metrics. Clicks and CTR are secondary diagnostics — a 0.7% CTR ad with great CPA beats a 1.5% CTR ad with bad CPA. ## Initial Assessment 1. Read `app-ads-context.md` — target CPA, LTV, monetization model 2. Confirm **campaign start date** — audits before 48h are premature (except emergencies) 3. Ask for **optimization event** — Purchase or Subscribe (not install) 4. Pull data from TikTok Ads Manager or Appeeky MCP 5. Confirm **batch structure** — 6-ad matrix from `tiktok-creative-strategy` 6. Note **total spend to date** and **days live** ## When to Audit | Trigger | Action | Urgency | |---------|--------|---------| | **48 hours after launch** | First full audit — scorecard all ads | Standard | | Spend > **2× target CPA**, 0 conversions on ad | **Instant pause** that ad | Emergency | | Daily (if scaling winners) | Quick CPA check — morning only | Ongoing | | Creative age **7+ days** | Refresh assessment — fatigue likely | Planned | | CPA rose **15%+ above target** after scale | Pause scale; hold budget | Warning | | Spark code expiry within 10 days | Regenerate per `tiktok-spark-ads` | Maintenance | **Do not** make kill decisions in the first 24 hours unless the emergency rule triggers. ### Why 48 Hours At $50/day across 6 ads, each ad receives ~$17 over 48h. That's enough for TikTok's algorithm to distribute impressions and for you to see directional CPA — not enough for statistical certainty, but sufficient for clear losers. ## Data Collection ### From TikTok Ads Manager Pull per-ad metrics for the audit window (last 48h or last 3 days): | Metric | Priority | Notes | |--------|----------|-------| | **Conversions** | **Primary** | Purchase or Subscribe events | | **Cost per conversion (CPA)** | **Primary** | Spend ÷ conversions | | **Spend** | Required | Per ad and total | | Conversions (SKAN) | Reference | Often 0 early on iOS — normal | | Clicks (destination) | Secondary | App Store / Play Store clicks | | **CTR (destination)** | Secondary | ~1% healthy; not required if CPA good | | Impressions | Diagnostic | For failure matrix | | CPC | Diagnostic | | | CPM | Diagnostic | | ### Benchmarks | Metric | Weak | Healthy | Strong | |--------|------|---------|--------| | CPA vs target | > 2× target | At target | < 0.7× target | | CTR (destination) | < 0.5% | ~1% | > 1.5% | | Conversions per ad (48h) | 0 | 2–5 | 10+ | | Spend per ad (48h) | < $5 (under-delivered) | $15–20 | $20+ | ### From Appeeky MCP ``` tiktok_ads_credentials_status tiktok_ads_list_advertisers tiktok_ads_performance advertiser_id: "<id>" level: "ad" days: 3 tiktok_ads_list_ads advertiser_id: "<id>" adgroup_id: "<id>" ``` Returns spend, impressions, clicks, installs, CPA, CPC, CTR per entity. Cross-reference with profitability: ``` # If RevenueCat connected rc_overview ``` Read `app-ads-context.md` for target CPA and LTV. ## Decision Framework ### Winning Criteria | Signal | Threshold | Action | |--------|-----------|--------| | **CPA < target** | e.g. CPA $14 vs target $20 | ✅ Winner — scale | | **CPA < LTV × 0.5** | Strong margin | Aggressive scale | | CTR ~0.8–1.5% | Secondary confirmation | Nice to have, not required | | 25+ conversions on ad | Statistical confidence | Prioritize budget to this ad | | Lowest CPA in batch | Relative winner | Keep running; model next batch on this format | ### Scale Rules (Winners) 1. Increase budget **+20% per day** until CPA rises 15%+ above target 2. Do not scale and swap all creatives same day — **change one variable** 3. Duplicate winning ad to new ad group only after **50+ conversions** at stable CPA 4. Refresh creative every **3–7 days** even on winners — fatigue is real 5. Expand geo only when US CPA stable **7+ days** ``` tiktok_ads_update_campaign campaign_id: "<id>" payload: budget: <current * 1.20> budget_mode: "BUDGET_MODE_DAY" ``` ### Kill Rules (Losers) | Condition | Action | |-----------|--------| | Spend > **2× target CPA** with **0 conversions** | **Instant pause** | | 48h + $100 batch spend, batch CPA > **1.5× target** | Pause entire batch → new creatives | | CPA **2× target** after 30+ conversions | Pause ad, analyze hook/format | | Policy rejection | Replace creative | | Creative fatigue (CTR drop **50%+** from peak) | Replace or refresh | | Spark code expired | Regenerate or pause until fixed | ``` tiktok_ads_update_ad_status ad_id: "<id>" operation_status: "DISABLE" ``` ## Failure Analysis Matrix Diagnose using impressions, CTR, clicks, and conversions. This tells you **where** the funnel breaks — not just that CPA is bad. | Impressions | CTR | Clicks | Conversions | Diagnosis | Fix | |-------------|-----|--------|-------------|-----------|-----| | Low | Low | Low | Low | Creative doesn't resonate | New formats entirely → `tiktok-creative-strategy` | | High | Low | Low | Low | Weak hook / CTA | Strengthen first 2s hook and on-screen text | | High | High | High | Low | Low intent / curiosity clicks | Show app use case earlier; less viral bait | | High | High | High | High but CPA bad | Onboarding/paywall issue | **Not an ads problem** → aso-skills `onboarding-optimization`, `paywall-optimization` | | High | High | Low | Low | Store listing issue | Check rating, screenshots → aso-skills `aso-audit` | | Low | High | Low | Low | Budget/delivery constraint | Confirm $50/day, policy status, placement settings | | High | Medium | Medium | Low SKAN only | iOS attribution lag | Trust MMP CPA; SKAN column lags — normal | ### How to Use the Matrix 1. Sort ads by spend (highest first) 2. For each underperformer, map to a matrix row 3. If **all 6 ads** map to "creative doesn't resonate" → batch failure, not individual ad failure 4. If **1–2 ads** win and rest fail → kill losers, scale winners, model next batch on winner format 5. If conversions exist but CPA bad across all → check LTV/pricing before blaming creative ## Per-Ad Scoring Rubric Score each ad 0–3 per dimension. **Total 0–15.** | Dimension | 0 | 1 | 2 | 3 | |-----------|---|---|---|---| | CPA vs target | > 2× or 0 conv | 1.5–2× | At target | < 0.7× target | | Conversion volume | 0 | 1–2 | 3–9 | 10+ | | CTR | < 0.5% | 0.5–0.8% | 0.8–1.2% | > 1.2% | | Spend efficiency | Under-delivered | Normal | Full delivery | Scale candidate | | Fatigue risk | CTR down 50%+ | Declining | Stable | Rising | | Total score | Verdict | |-------------|---------| | 12–15 | ✅ Scale +20%/day | | 8–11 | ⚠️ Watch — need more data | | 4–7 | ⏸ Pause — analyze | | 0–3 | ❌ Kill immediately | ## Per-Ad Scorecard Template ```markdown # TikTok Audit — [Date] **Target CPA:** $___ | **LTV:** $___ | **Period:** Last 48h **Campaign:** [name] | **Days live:** [N] | Ad | Spend | Conv | CPA | CTR | Score | Verdict | |----|-------|------|-----|-----|-------|---------| | A - Story v1 | $35 | 2 | $17.50 | 0.83% | 9 | ⚠️ Watch | | A′ - Story v2 | $32 | 3 | $10.67 | 0.91% | 12 | ✅ Scale | | B - Before/after | $28 | 1 | $28.00 | 0.61% | 5 | ⏸ Pause | | B′ - Before/after | $30 | 0 | — | 0.55% | 2 | ❌ Kill | | C - Tutorial | $25 | 2 | $12.50 | 0.96% | 11 | ✅ Scale | | C′ - Tutorial | $22 | 1 | $22.00 | 0.69% | 7 | ⚠️ Watch | **Batch CPA:** $___ | **Total spend:** $___ | **Total conversions:** ___ ## Diagnosis - [Matrix row]: [which ads, what pattern] ## Actions 1. Pause: [ads + reason] 2. Scale +20%: [ads + CPA] 3. New batch needed by: [date + 5 days] 4. Comment filtering: [enabled/Y/N] ``` ## Optimization Playbook ### Week 1 — Testing | Day | Action | |-----|--------| | 0 | Launch 6-ad batch at $50/day | | 1 | Observe only — log impressions distribution | | 2 | **Full audit** — kill 2× CPA zero-conv ads | | 3–4 | Hold budget; let winners accumulate data | | 5–7 | Identify 1–2 winners; plan hook variants | ### Week 2 — Scaling | Action | Detail | |--------|--------| | Budget | +20%/day on campaign while batch CPA < target | | Creative | Produce A′/B′ variations of **winning format only** (hook tweaks) | | Comments | Enable filtering; pin FAQ on top Spark post | | Geo | Hold US — do not expand yet | ### Week 3+ — Maintenance | Action | Detail | |--------|--------| | New batch | Every **7 days** — even if current batch winning | | Geo expansion | Add CA, UK, AU when US CPA stable 7+ days | | Profitability | Cross-check `campaign-profitability` with RevenueCat | | Channel mix | Compare `cross-channel-performance` vs Meta/ASA | ## Realistic Economics Check After every audit, sanity-check unit economics: | Metric | Formula | Healthy | |--------|---------|---------| | Gross margin | (LTV - CPA) / LTV | 30–50%+ after ad spend | | Learning tax | First $300–500 spend | Expect elevated CPA | | Payback period | CPA / (LTV / expected lifetime months) | < 3 months for subs | | ROAS (if revenue tracked) | Revenue / Spend | > 100% at scale | If CPA looks good but revenue doesn't follow → problem is downstream (onboarding, paywall, product), not TikTok. ## Bonus Tactics | Tactic | Implementation | When | |--------|----------------|------| | **Comment filtering** | Ads Manager → filter: scam, AI, fake, bot, money | Day 0 or when comments turn toxic | | **Pinned FAQ comment** | Pin trial/pricing answer on top Spark post | After 24h when comments appear | | **Kill fast** | No sentiment attachment — data decides | Any 2× CPA zero-conv ad | | **Don't hover** | Check 1× morning during test | Reduces premature optimization | | **SKAN patience** | SKAN column lagging is normal | Trust MMP + CPA column | | **Creative refresh** | New 6-ad batch every 3–7 days | Even winners fatigue | | **Alt account comments** | Optional social proof seeding | User discretion — prefer genuine FAQ | ## Common Audit Mistakes | Mistake | Consequence | Fix | |---------|-------------|-----| | Judging on CTR alone | Killing profitable low-CTR ads | CPA is primary | | Killing before 48h | False negatives on variance | Wait unless emergency rule | | Scaling losing batch | Burning budget faster | Audit first | | Ignoring SKAN lag | Panic on iOS | Trust MMP attribution | | One variable change violated | Can't attribute CPA change | Scale OR refresh, not both same day | | No new batch planned | Winner fatigues at peak scale | Schedule refresh day 5–7 | | Blaming ads for 0 revenue | Wasted creative iterations | Check paywall/onboarding | ## Appeeky MCP — Audit Workflow Full audit sequence: ``` # 1. Verify connection tiktok_ads_credentials_status # 2. List campaigns and ads tiktok_ads_list_campaigns advertiser_id: "<id>" tiktok_ads_list_adgroups advertiser_id: "<id>" campaign_id: "<id>"