How to Audit a Meta Ads Account: The Complete Agency Checklist (2026)

Why Most Meta Ads Audits Fail

Most audit templates are flat checklists: 40 items with equal weight. The problem is that several layers depend on each other. Auditing creative quality on an account whose pixel is feeding 56% wrong data produces conclusions that look right but are mathematically meaningless [1]. Baker’s 8-Layer Meta Ads Audit Framework runs the checks in dependency order, so each layer’s fixes compound rather than mask each other.

The three principles behind the framework:

1. Data quality before optimisation. If the algorithm is training on the wrong conversions, every other lever is downstream of a broken signal [1].
2. Andromeda-aware diagnostics. The 2024-2025 algorithm shift means audit thresholds for campaign count, audience targeting, and creative variety changed. See Baker’s Andromeda Adaptation Framework for the full breakdown [2].
3. Tier-appropriate benchmarks. A $3K/month account and a $50K/month account have different healthy values. Applying enterprise thresholds to a small account produces false negatives, and vice versa [3].

Layer 1: Pre-Audit Discovery

Before opening Ads Manager, capture context. Maximilian von Mer’s account audit framework defines five questions every audit must answer before tactical review [4]:

Question	Why It Matters
Status quo: where do we stand on hard numbers?	Establishes baseline ENCAC, ROAS, CPL, and lead-to-close rate
Goal: what specific outcome do we want in 90 days?	Audit recommendations must serve a target, not vague improvement
Historical review: what has been tried and at what scale?	Prevents recommending tactics that already failed
Industry benchmarking: how does this compare to mature peers?	Distinguishes account-level issues from category-level reality
Devil’s advocate: what current assumptions might be wrong?	Surfaces tracking, attribution, or offer issues hiding behind tactics

Discovery questions Baker uses on B2B SaaS audits:

• What is the difference between in-platform CPL and CRM-tracked sales-qualified opportunity cost?
• How are leads handed off to sales, and what is the SLA for first contact?
• What is the average sales cycle length, and is the attribution window aligned to it?
• Has CAPI been implemented? When? Has Event Match Quality been verified since?
• What is the offline conversion import cadence, if any?

If discovery surfaces unanswered questions, pause the technical audit until they are answered. According to Sylvia Perez (Ad Conversion, $3M/month managed spend), “auditing a B2B account without knowing the close rate by lead source is auditing the wrong half of the funnel” [5].

Layer 2: Pixel and CAPI Health

Data quality is the foundational layer. Every downstream audit conclusion depends on the algorithm receiving accurate signals.

Standard pixel verification

Check	Healthy	Action If Failing
Pixel firing on all key events	100%	Reinstall via Events Manager, verify with Pixel Helper
PageView fires once per page	Single fire	Remove duplicate snippets
Standard events match conversion goals	Direct match	Map custom events to standard schema
Test events show no errors in Events Manager	No errors	Resolve schema mismatches before continuing

CAPI and deduplication

According to Jon Loomer, the easiest non-Shopify CAPI setup is Conversions API Gateway via Stape.io at roughly $10/month, which automatically duplicates web pixel events through the API [6]. For Shopify accounts, the native Meta integration is a checkbox.

The non-negotiable check: deduplication. Rok Hladnik (Thumbstop) lists this as the first item in his low-hanging-fruit audit, noting that “even standard, two-click integrations like Shopify and Meta are frequently broken” [7].

CAPI Check	Healthy	Failing Symptom
Deduplication rate	70-100% browser+server matched	Duplicate conversions inflating ROAS
Event Match Quality (EMQ)	7.0+ on key events	Algorithm cannot match users, CPMs rise
Server-only events on iOS traffic	Present	Major iOS undercount
Edge-tagged CAPI for cross-device	Present	CRM imports lose fingerprint match

The 56% problem

According to John Moran (Tier 11), “Shopify’s own public attribution is only 44% accurate. Importing Shopify purchase data via the direct Shopify-to-Meta integration means feeding the algorithm 56% wrong data” [1]. The Tier 11 fix is first-click CAPI imports from the CRM with custom new-customer events, which raised attribution accuracy from 53% to 85%+ in the documented case studies [1].

A Tier 11 beauty account using this fix dropped ENCAC from $26 to $6.73 across six weeks while doubling spend [1]. For the full implementation, see Baker’s CAPI Setup Guide.

Layer 3: Campaign Architecture Review

Andromeda changed the right number of campaigns. Pre-Andromeda audits flagged accounts with too few campaigns. Post-Andromeda audits flag accounts with too many [2].

Monthly Spend	Healthy Campaign Count	Common Audit Finding
Under $3K	1 ABO campaign, broad targeting	5+ campaigns split too thin to exit Learning
$3K-$10K	2 campaigns (Test ABO + Scale CBO)	8-10 campaigns with overlapping audiences
$10K-$30K	3 campaigns (Test, Scale, Retargeting)	Persona-by-persona ABO splits over-segmented
$30K-$100K	3-5 portfolio CBOs by awareness level	Legacy structure from 2022 still running
$100K+	5-8 CBOs, full-funnel sequencing	Insufficient retargeting investment

Multi-expert consensus (Charley T at Disruptor School, Manel Gomez, Dara Denney) confirms that consolidation from 15+ campaigns to 3-5 typically improves CPA and accelerates Learning Phase exit [2] [3] [8].

Architecture audit checklist:

• Campaign count matches budget tier (table above)
• No campaign in Learning Limited for more than 14 days
• No ad set spending under $50/day with conversion-objective optimisation
• CBO budget weighted toward proven creative, not split equally
• Manual placement restrictions justified by regulation, not preference
• No duplicate ad sets in the same campaign with similar audiences

For the full structure-by-budget breakdown, see Baker’s Meta Ads Campaign Structure guide.

Layer 4: Creative Scoring

Creative is the targeting layer in Andromeda. Audit it accordingly.

Entity ID diversity

According to Ryan from Leadbase ($1B+ in Meta spend analysed), Andromeda measures variation by Entity ID (the underlying visual concept), not Creative ID. Six is the minimum for the algorithm to optimise effectively, and 6-10 is the target for scaled accounts [9].

Account State	Entity IDs	Diagnosis
1-3 distinct concepts	Below threshold	Andromeda cannot match diverse audience pockets
4-5 distinct concepts	Marginal	Performance plateau likely
6-10 distinct concepts	Healthy	Algorithm has room to optimise
10+ distinct concepts	Scale-ready	Provided each has 3-5 variations

Hook and hold rate benchmarks

Tier 11’s Ralph Burns publishes the operative benchmarks for video creative [10]:

Metric	Definition	Minimum	Strong	Exceptional
Hook rate	3-sec views ÷ impressions	25%	30%	40-60%
Hold rate	15-sec views ÷ 3-sec views	30%	40%	50%+
Thumbstop ratio	3-sec views ÷ video plays	30%	40%	50%+

Diagnostic logic: Healthy hook + low hold = fix the body of the ad (introduce the offer earlier, tighten the middle). Low hook + healthy hold = the audience is wrong for the visual, or the opening is generic [10].

Format mix audit

Lattice rewards format diversity. Dara Denney’s recommended mix [8]:

Format	Share of Active Creative	Common Audit Gap
Static images	30-40%	Often under-invested for video-led brands
Video (15-60s)	30-40%	Single format dominance
Carousel	10-15%	Frequently 0% on B2B accounts
Experimental (UGC, text, AI)	10-15%	Skipped entirely

Creative fatigue signals

According to Sylvia Perez, the operational fatigue rule is frequency above tier-specific cap (7 for retargeting, lower for prospecting) combined with CTR below the trailing 90-day average [5]. Manel Gomez adds the 30-day frequency band of 2.7 to 3.3 as the most common silent CPM-spike trigger [3].

Hook type audit (Dara Denney’s 10 dominant formats in 2026):

1. Going-against-tradition hooks
2. Conversational openers (“no because”)
3. Viral product hooks
4. Reaction hooks
5. Controversy hooks
6. Trigger-word emotional hooks
7. Demographic hooks
8. Multiple stacked hooks
9. “How do I know if X” mirror hooks
10. Speed and transformation (before/after with urgency) [8]

A B2B SaaS account running only one or two of these in active rotation is undertesting hook variety, regardless of total ad count.

Layer 5: Audience Overlap Diagnostics

Use Meta’s Audience Overlap tool, then triangulate with first-click attribution.

Overlap Type	Healthy	Audit Action
Two ad sets in same campaign	Under 25%	Above 25% = consolidate
Prospecting vs retargeting	Under 10%	Higher = exclusion list missing
Lookalike 1% vs lookalike 2%	Under 50%	Use seed-quality scoring instead of stacking
Cross-campaign overlap	Under 25%	Drop the lower-CTR campaign

The cross-channel overlap problem: John Moran’s Tier 11 framework documents that without first-click CAPI, increasing Meta spend lifts in-platform numbers across Google, email, and organic, because Meta creates the first touch but those channels take last-click credit. The fix is first-click CAPI imports, which restore Meta’s true contribution [1].

B2B audit nuance: Sylvia Perez’s rule for broad targeting on B2B is that it works only when (1) ABM is not required, (2) TAM is large enough that broad makes sense, and (3) creative explicitly calls out the ICP and repels non-buyers [5]. If any condition fails, narrow audiences with high creative discipline outperform broad.

Layer 6: Frequency Analysis

Frequency must be evaluated on a fixed window (7-day or 30-day), never lifetime [6].

Funnel Stage	Healthy 7-Day Frequency	Cap Threshold
Cold prospecting	1.5-2.5	3 per 7 days
Warm engagement	2.0-3.5	4 per 7 days
Retargeting	3.0-5.0	5 per 7 days
Brand-aware retargeting	4.0-6.0	6 per 7 days

Irina Buht’s diagnostic method: plot conversion rate by frequency bucket, and cap at the bucket where conversion rate peaks before declining [11]. For prospecting, that is typically frequency 2.3 (peak) to 4.0 (decline), so the cap goes at 3. For retargeting, the curve runs higher, peaking around 6 before drop-off [11].

Manel Gomez’s silent killer: 30-day frequency between 2.7 and 3.3 produces unexplained CPM increases and inconsistent ROAS in roughly two-thirds of audits Baker runs. Always check 30-day frequency before blaming the algorithm [3].

Jon Loomer’s context rule: Frequency 4.0 over four days is intrusive; frequency 4.0 over twelve months is invisible. Trust Meta’s native fatigue alerts in Ads Manager over arbitrary numerical thresholds [6].

Layer 7: Post-Click Funnel Audit

A clean ad audit means nothing if the landing page leaks 90% of clicks.

Landing page conversion rate floors

According to Aaron Young’s benchmarks [12]:

Vertical	LP Conversion Rate Floor	Below Floor = Audit Required
E-commerce	3%	Test offer, urgency, social proof
Lead generation	7%	Test headline, form, hero copy
B2B SaaS demo request	5-8%	Test ICP-specificity of LP copy

Form audit checklist

Arsh Sanwarwala and Market Lead’s combined form framework [13]:

• Under 4 fields where possible
• “Full name” instead of separate first/last
• Phone number removed unless essential to the offer
• Single-column layout (never multi-column)
• Mental commitment: low-friction fields first, qualifying fields after
• Sales-team qualifying questions added on Step 2 of multi-step forms
• Thank-you page states exactly what happens next and when

Michael Nadlin’s progressive disclosure data shows that roughly 80% of users who complete a low-friction Step 1 (name, email, phone) will complete a higher-friction Step 2 if Step 2 is paired with a value escalation (custom pricing, custom plan, premium asset) [13].

Lead quality filtering

For local services and high-ticket B2B, Jack Hepp’s framework: ask the sales team for the first 3-5 qualification questions they use on prospects, and put those on the form. Total lead volume drops; lead quality rises significantly. The ENCAC almost always improves [14]. For the full lead-quality stack, see Baker’s Meta Ads Lead Quality guide.

Layer 8: Competitive Intelligence

The Meta Ad Library is the audit’s external benchmark.

Check	What To Capture
Top 3 competitors’ active ad count	Volume signal
Longest-running ads (90+ days)	Proven concepts to study
Format mix used	Where competitors over- or under-invest
Hook patterns	Recurring opening structures
Landing page destinations	Funnel architecture insights

Baker’s rule: if the audited account has fewer than half the active ads of the median competitor at the same scale, creative production is the bottleneck regardless of every other layer.

Growth-Phase Audit Calibration

Audit thresholds shift by spend tier. Adapted from the 99 Ads $0 to $1M/month playbook and Baker’s client work [15]:

Phase	Monthly Spend	Audit Priority
Offer-Market Fit	$0-$10K	Test offer presentations, not channels. Audit creative volume and offer mix, not architecture
Iteration Muscle	$10K-$100K	100-200 ads/month target. Audit creative production cadence and winner rate (1-5/week)
Angle Expansion	$100K-$500K	1 audience, up to 4 angles, 400 ads/month. Audit angle diversification (EPIC: Emotional, Practical, Identity, Critical)
Audience Expansion	$500K-$1M	2-3 adjacent audiences, 80-90% on core, 10-20% horizontal tests. Audit concentration risk
Beyond $1M	$1M+	TAM, LTV, AOV. Audit international rollout, upsell architecture, subscription LTV

A $3K/month account audited against $100K/month thresholds will look broken on every layer. A $100K/month account audited against $3K thresholds will look fine when it is leaving 7 figures of incremental revenue on the table.

The Baker Audit Output: From Findings to Fixes

A finished audit produces three artefacts:

1. The 8-Layer scorecard (red, amber, green per layer)
2. The dependency-ordered fix list (Layer 2 issues fixed before Layer 4, etc.)
3. The 30/60/90 implementation plan mapping each fix to a measurable target

Multi-expert consensus (Tier 11, Sylvia Perez, Manel Gomez) is that fixing one tracking issue in Layer 2 typically delivers more performance lift than rebuilding 50 ads in Layer 4 [1] [3] [5]. The audit’s job is to prevent the team from spending two months on Layer 4 when Layer 2 is broken.

Common Audit Mistakes

Mistake 1: Auditing creative on broken pixels

Conclusions about which creative converts depend on conversion data being accurate. Always verify Layer 2 first.

Mistake 2: Applying enterprise thresholds to small accounts

A $3K/month account is not failing because it has only 4 active ads. It is in the offer-fit phase. Audit the offer, not the campaign architecture.

Mistake 3: Treating frequency as a single number

Frequency 4.0 across 7 days is fatigue. Frequency 4.0 across 90 days is healthy. Always specify the window.

Mistake 4: Ignoring cross-channel cannibalisation

Meta often creates the first touch that Google, email, or organic claim. Without first-click CAPI, the audit will recommend cutting Meta spend that is actually driving the entire funnel [1].

Mistake 5: Stopping at in-platform metrics

The CRM is the source of truth. CPL is a vanity metric without close rate, ENCAC, and lifetime value behind it. Audit to the closed deal, not the lead form submission.

---

Sources

1. John Moran, Tier 11. “First-Click CAPI Imports, Custom New-Customer Events, and the 56% Shopify Attribution Problem.” Beauty Account ENCAC Case Study ($26 to $6.73), 2026.
2. Manel Gomez, Crece sin Limite. “Andromeda Architecture and Campaign Consolidation.” $3.3M Ad Spend Analysis, 2026.
3. Manel Gomez, Crece sin Limite. “30-Day Frequency Band 2.7-3.3 as Silent CPM Trigger.” Meta Scaling Diagnostic Framework, 2026.
4. Maximilian von Mer. “Five-Question Account Audit Framework.” Strategic Audit Methodology, 2026.
5. Sylvia Perez, Ad Conversion ($3M/month managed spend). “B2B Meta Ads Lead Quality and Broad Targeting Prerequisites.” B2B Meta Playbook, 2026.
6. Jon Loomer. “Conversions API Gateway, Frequency Context, and Native Fatigue Alerts.” Meta CAPI and Frequency Guides, 2026.
7. Rok Hladnik, Thumbstop. “Low-Hanging-Fruit Audit Checklist: Pixel Deduplication First.” Meta Ads Audit Methodology, 2026.
8. Dara Denney. “Top 10 Hook Types and Format Mix Recommendations Under Andromeda.” Meta Creative Strategy 2026.
9. Ryan, Leadbase. “Entity ID Diversity Requirements: 6-10 Distinct Visual Concepts.” $1B+ Meta Spend Analysis, 2026.
10. Ralph Burns, Tier 11. “Hook Rate, Hold Rate, and Thumbstop Benchmarks for Video Creative.” Tier 11 Creative Diagnostic, 2026.
11. Irina Buht. “Frequency Cap Methodology: Conversion Rate by Frequency Bucket.” Meta Frequency Cap Recommendations, 2026.
12. Aaron Young, Scale Bottom-Up. “Landing Page Conversion Rate Floors by Vertical.” Benchmarks and KPIs, 2026.
13. Arsh Sanwarwala, Michael Nadlin (Market Lead). “Form Field Strategy and Progressive Disclosure (80% Step-2 Completion).” High-Ticket Lead Gen Forms, 2026.
14. Jack Hepp, Local Ads. “Lead Quality Filtering via Sales-Team Qualifying Questions.” Local Lead Generation Framework, 2026.
15. Founder of 99 Ads. “Five-Stage Scaling Playbook: $0 to $1M/Month.” Scaling Strategy, 2026.