meta-performance-troubleshooter

Performance Troubleshooter

Diagnostic framework for identifying and resolving Meta Ads performance problems.

Quick Diagnostic

WHAT IS THE PRIMARY SYMPTOM?
│
├─► No or low delivery
│   └─► Go to [Delivery Issues]
│
├─► High CPA / Low ROAS
│   └─► Go to [Efficiency Problems]
│
├─► Low CTR
│   └─► Go to [Engagement Issues]
│
├─► High CPM
│   └─► Go to [Auction Competition]
│
├─► Performance declining over time
│   └─► Go to [Performance Decay]
│
└─► Conversions not tracked
    └─► Go to [Tracking Issues]

Diagnostic Framework

Step 1: Check Fundamentals

FUNDAMENTALS CHECKLIST:
□ Pixel firing? (meta_query(account_id, "adspixels") shows last_fired_time; event diagnostics are in Events Manager)
□ CAPI active? (Server events visible?)
□ Event Match Quality? (meta_query(account_id, "dataset_quality", entity_id=<pixel_id>): composite_score per web event)
□ Attribution window correct? (Match buying cycle)
□ Budget sufficient? (Enough for about 50 results a week per ad set: daily budget of about 50 x CPA / 7)
□ Audience size adequate? (Rule of thumb, not a Meta figure: 1M+ for prospecting)
□ Creative variety? (Count distinct concepts, not ad objects: Meta treats look-alike ads as variations of the same creative)

The 50 is Meta's learning phase figure: an ad set usually exits learning after about 50 results in the week after its last significant edit, and a budget too low for that keeps it learning limited (Meta Business Help Center, "About the learning phase", and "About learning limited", checked 2026-10-10). Meta scores EMQ out of 10 for website events sent through the Conversions API ("About event match quality").

Step 2: Ad Relevance Diagnostics

Meta's 3 ranking metrics at ad level:

Meta does not show these rankings for ads with fewer than 500 impressions, and they are not inputs into the auction (Meta Business Help Center, "About ad relevance diagnostics", checked 2026-10-10).

Step 3: Identify Root Cause

Use diagnostic matrix below per symptom.

Delivery Issues

Rules of thumb, not Meta figures: check them against the account's own history.

Symptom: No or Very Low Impressions

DIAGNOSE: Why no delivery?
│
├─► Ad Account Issues
│   ├── Account restricted/disabled?
│   ├── Payment method issue?
│   └── Policy violation?
│
├─► Campaign Settings
│   ├── Budget too low?
│   ├── Bid/Cost cap too restrictive?
│   ├── Schedule issues? (Ads not active?)
│   └── Campaign paused?
│
├─► Audience Issues
│   ├── Audience too small? (<50K)
│   ├── Too many exclusions?
│   ├── Geo targeting too narrow?
│   └── Overlap with other campaigns?
│
├─► Ad Issues
│   ├── Ads rejected/in review?
│   ├── Low Quality Ranking?
│   └── Policy flags?
│
└─► Competition Issues
    ├── Auction outbid?
    └── CPM spike in market?

Solutions Matrix: Delivery

When two or more of your ads enter the same auction, Meta lets only the one with the highest total value compete, so your ads do not bid against each other; overlap can stop an ad set from spending its budget or leaving learning (Meta Business Help Center, "Understand auction overlap", checked 2026-10-10).

Efficiency Problems

Symptom: High CPA / Low ROAS

DIAGNOSE: Why high CPA?
│
├─► Funnel Analysis
│   ├── CPM normal, CTR low → Creative issue
│   ├── CPM normal, CTR normal, CVR low → Landing page issue
│   ├── CPM high, CTR normal → Audience/competition issue
│   └── All metrics degraded → Multiple issues
│
├─► Funnel Metrics Checklist (benchmarks: rules of thumb, not Meta figures; compare with the account's history)
│   ├── CPM: €[X] (Benchmark: €15-25)
│   ├── CTR: [X]% (Benchmark: 1-2%)
│   ├── CPC: €[X] (Benchmark: €0.50-1.00)
│   ├── LP → ATC: [X]% (Benchmark: 15%)
│   ├── ATC → Purchase: [X]% (Benchmark: 30-50%)
│   └── Overall CVR: [X]% (Benchmark: 2-5%)
│
└─► Identify Bottleneck
    └─► Focus optimization effort there

Solutions Matrix: Efficiency

Funnel Optimization Priority

PRIORITY ORDER:
1. Fix tracking first (no data = optimizing blind)
2. Landing page
3. Creative
4. Targeting (let AI help)
5. Bid strategy (fine-tuning)

Engagement Issues

Symptom: Low CTR

CTR BENCHMARK (rules of thumb, not Meta figures: check them against the account's own history):
├── Poor: <0.8%
├── Average: 0.8-1.5%
├── Good: 1.5-2.5%
└── Excellent: >2.5%

LOW CTR CAUSES:
│
├─► Creative Issues
│   ├── Weak hook (first 3 sec video)
│   ├── Unclear value proposition
│   ├── Poor visual quality
│   ├── Wrong format for placement
│   └── Creative fatigue
│
├─► Targeting Issues
│   ├── Irrelevant audience
│   ├── Audience too broad
│   └── Message-audience mismatch
│
└─► Ad Copy Issues
    ├── Weak headline
    ├── No clear CTA
    └── Benefits unclear

Solutions: Low CTR

CREATIVE FIXES:
├── Test several new hook variations
├── A/B test headlines
├── Try different formats (video vs static)
├── Update visuals/thumbnails
└── Add motion/animation

TARGETING FIXES:
├── Narrow to higher-intent segments
├── Test different interest combinations
├── Create lookalikes from purchasers
└── Exclude low-engagement segments

COPY FIXES:
├── Lead with benefit, not feature
├── Add social proof
├── Create urgency
└── Clearer, action-oriented CTA

Auction Competition

Symptom: High CPM

CPM BENCHMARK (rules of thumb, not Meta figures: check them against the account's own history):
├── Low: <€10
├── Average: €15-25
├── High: €25-40
└── Very High: >€40

HIGH CPM CAUSES:
│
├─► Market Factors
│   ├── Peak season (Q4, Black Friday)
│   ├── Industry competition spike
│   └── Major events/elections
│
├─► Targeting Factors
│   ├── Very competitive audience
│   ├── Small audience (premium pricing)
│   └── Overlapping with own campaigns
│
└─► Quality Factors
    ├── Below-average ad relevance diagnostics
    ├── Poor engagement history
    └── New ad account (no history)

Ad quality and estimated action rates are part of the auction's total value, next to the bid, so a more relevant ad could win against higher bids (Meta Business Help Center, "About ad auctions", checked 2026-10-10). The old relevance score was replaced by ad relevance diagnostics.

Solutions: High CPM

IMMEDIATE ACTIONS:
├── Expand audience
├── Remove restrictive targeting
├── Test different placements (including Threads)
├── Check audience overlap with Meta's overlap tool in Ads Manager (the MCP audienceoverlap route returned a Meta error in our tests; see audience-overlap-detector)

STRATEGIC ACTIONS:
├── Improve ad quality (better auction position)
├── Test off-peak timing
├── Build lookalikes from best customers
└── Diversify to less competitive channels

SEASONAL STRATEGY:
├── Pre-peak: Lock in audiences, test creatives
├── Peak: Accept higher CPM, focus on ROAS
└── Post-peak: Capitalize on lower competition

Performance Decay

Symptom: Metrics Declining Over Time

DECAY DIAGNOSIS (thresholds are rules of thumb, not Meta figures: check them against the account's own history):
│
├─► Creative Fatigue
│   ├── Symptoms: Rising frequency, falling CTR
│   ├── Threshold: Frequency >3-4
│   └── Solution: Fresh creative rotation
│
├─► Audience Saturation
│   ├── Symptoms: Shrinking reach, high frequency
│   ├── Threshold: Reached >70% of audience
│   └── Solution: Expand targeting, new audiences
│
├─► Seasonal Effects
│   ├── Symptoms: Industry-wide decline
│   └── Solution: Adjust expectations, test new offers
│
├─► Competitor Activity
│   ├── Symptoms: CPM up, CTR down
│   └── Solution: Differentiate, refresh positioning
│
└─► Algorithm Changes
    ├── Symptoms: Sudden performance shift
    └── Solution: Adapt to new best practices

Creative Fatigue Detection

META'S OWN SIGNAL:
□ Delivery column shows "Creative limited" or "Creative fatigue"

FATIGUE INDICATORS (rules of thumb, not Meta figures: check them against the account's own history):
□ CTR dropping week-over-week
□ Frequency >4 in past 7 days
□ Same creative running >14-21 days
□ Engagement Rate Ranking declining
□ CPM increasing without market shift

REFRESH CADENCE (rules of thumb, not Meta figures: check them against the account's own history):
├── High-spend campaigns: Every 7-14 days
├── Medium-spend: Every 14-21 days
├── Low-spend: Every 21-30 days
└── Evergreen/testimonials: Can run longer

Meta shows "Creative limited" when an ad's cost per result is above that of ads you ran in the past but less than twice as much, and "Creative fatigue" at twice as much or more. Meta offers this only for ad sets with one creative, and not for Advantage+ catalog ads, dynamic creative or Advantage+ app campaigns (Meta Business Help Center, "About creative fatigue recommendations in Meta Ads Manager", checked 2026-10-10). Check the status in Ads Manager.

Tracking Issues

Symptom: Conversions Not Tracked

TRACKING DIAGNOSTIC:
│
├─► Pixel Issues
│   ├── Pixel not installed correctly
│   ├── Pixel blocked by ad blockers
│   ├── Event not firing on conversion
│   └── Wrong pixel ID
│
├─► CAPI Issues
│   ├── CAPI not configured
│   ├── Events not deduplicating
│   ├── Low Event Match Quality
│   └── Token expired
│
├─► Attribution Issues
│   ├── Wrong attribution window
│   ├── Cross-device not tracked
│   ├── Long purchase cycle
│   └── iOS privacy impact
│
└─► Technical Issues
    ├── Thank you page not loading
    ├── Redirect issues
    └── Tag manager conflicts

Tracking Fix Checklist

QUICK FIXES:
□ Test Pixel with Meta Ads Data Advisor (formerly Meta Pixel Helper; a Chrome extension)
□ Check Events Manager for recent events (meta_query(account_id, "adspixels") only gives last_fired_time)
□ Skip domain verification and AEM event ranking: Meta no longer requires either for web event setup
□ Confirm event parameters (value, currency)
□ Check for duplicate events

CAPI VERIFICATION (check these in Events Manager):
□ Server events visible in Events Manager?
□ Deduplication working? (Same event from 2 sources)
□ Event Match Quality 6.0 or higher where possible? (Meta's recommendation, "Verifying Your Setup", Conversions API docs)
□ Customer parameters hashed correctly?

ATTRIBUTION CHECK:
□ Attribution window matches buying cycle
□ GA4 comparison (same trends?)
□ Post-purchase survey attribution

Meta's Aggregated Event Measurement page says web events need no AEM setup any more, the 8-event prioritization is gone, and domain verification is no longer required for event configuration, though a business may still need it for other reasons (Meta Business Help Center, "About Meta's Aggregated Event Measurement", checked 2026-10-10). Meta renamed the Pixel Helper extension to Meta Ads Data Advisor (Meta for Developers, "Meta Ads Data Advisor", checked 2026-10-10).

Kill vs Scale Decision

Rules of thumb, not Meta figures: check them against the account's own history.

Decision Framework

SCALE SIGNALS (Increase Budget):
├── ROAS >target for 5+ days
├── CPA <target and stable
├── CTR above benchmark
├── Frequency <3
├── Room to grow in audience
└── Consistent daily performance

OPTIMIZE SIGNALS (Test & Tweak):
├── ROAS within 20% of target
├── CPA fluctuating but manageable
├── Some ads performing, others not
├── Creative starting to fatigue
└── Potential with adjustments

KILL SIGNALS (Pause/Stop):
├── ROAS <1x (losing money)
├── CPA >2x target for 7+ days
├── CTR <0.5% consistent
├── No improvements despite tests
├── Insufficient budget for learning
└── Audience exhausted

Kill Decision Checklist

BEFORE KILLING, VERIFY:
□ Tracking is accurate (not a tracking issue)
□ Attribution window appropriate
□ Gave sufficient time (7+ days minimum)
□ Tested multiple creatives
□ Budget was adequate
□ Not during anomaly period (holiday, etc.)

IF ALL VERIFIED, THEN:
├── Killing = correct decision
├── Document learnings
├── Archive for reference
└── Reallocate budget to winners

MCP: Pull Performance Data for Diagnosis

# Get campaign-level performance metrics
meta_get_insights(account_id="act_XXXXX", level="campaign", date_preset="last_7d", fields=["spend","impressions","clicks","ctr","cpm","cpc","actions","cost_per_action_type","frequency"])

# Get ad-level breakdown to isolate creative issues
meta_get_insights(account_id="act_XXXXX", level="ad", date_preset="last_7d", fields=["spend","impressions","clicks","ctr","frequency","actions","quality_ranking","engagement_rate_ranking","conversion_rate_ranking"])

# meta_get_insights returns one page of rows: while a response has paging.next, fetch the next page
meta_query(account_id="act_XXXXX", entity_type="ads", pagination_url="<paging.next>")

Troubleshooting Output Template

# Campaign Troubleshooting Report

## Campaign Overview
- Campaign: [name]
- Objective: [objective]
- Daily Budget: €[X]
- Running since: [date]
- Current Status: [status]

## Symptom Analysis
**Primary symptom:** [description]
**Duration:** [since when]
**Severity:** [Low/Medium/High/Critical]

## Diagnostic Results

### Fundamentals Check
- [x] Pixel: [OK/Issue]
- [x] CAPI: [OK/Issue]
- [x] EMQ: [score]
- [x] Budget: [OK/Issue]
- [x] Audience: [OK/Issue]

### Key Metrics
| Metric | Current | Benchmark | Status |
|--------|---------|-----------|--------|
| CPM | €[X] | [from this account's history] | [OK/High/Low] |
| CTR | [X]% | [from this account's history] | [OK/High/Low] |
| CPC | €[X] | [from this account's history] | [OK/High/Low] |
| CVR | [X]% | [from this account's history] | [OK/High/Low] |
| CPA | €[X] | €[target] | [OK/High] |
| ROAS | [X]x | [target]x | [OK/Low] |

### Root Cause Identified
[Description of root cause]

## Recommended Actions

### Immediate (This Week)
1. [Action 1]
2. [Action 2]

### Short-term (2-4 Weeks)
1. [Action 1]
2. [Action 2]

### Monitor
- [Metric 1]: Watch for [threshold]
- [Metric 2]: Watch for [threshold]

## Prognosis
[Metrics to re-check after the actions, with today's values as the baseline. Estimate an outcome only from this account's own data, with the assumptions stated; otherwise write "not estimated".]