gtmskills.com

LinkedIn ads 35-point audit

Use this skill when someone wants a LinkedIn Ads audit, efficiency scan, account health check, or 'what should we change', or uploads Campaign Manager exports, screenshots, CRM data, or ad creative. Runs a 35-point system inspection across six layers and adapts to the data provided: CSV exports get ~19 points, Campaign Manager screenshots ~27, CRM data ~31, creative review all 35.

By Justin Roweimpactable.comFounder & CMO @ Impactable
Test this skill
or install it
Install in your Swan workspace
View source on GitHub
SKILL.md
name:
linkedin-ads-35-point-audit
description:
Use this skill when someone wants a LinkedIn Ads audit, efficiency scan, account health check, or 'what should we change', or uploads Campaign Manager exports, screenshots, CRM data, or ad creative. Runs a 35-point system inspection across six layers and adapts to the data provided: CSV exports get ~19 points, Campaign Manager screenshots ~27, CRM data ~31, creative review all 35.

LinkedIn ads 35-point audit

Assesses a LinkedIn Ads account through 6 diagnostic layers, each building on the last. Structural issues first, tactical opportunities second. It adapts to whatever data is available: a single CSV = ~19 checks, everything = all 35. Every run produces a health score, prioritized findings, and specific actions.

The 6 layers (fix upper layers before optimizing lower ones):

  1. Signal & Data Integrity (7 points) - what LinkedIn optimizes toward
  2. Account Architecture & Learning Design (6 points) - how the system compounds knowledge
  3. Audience & Buying Committee Coverage (7 points) - who you're reaching
  4. Creative & Message-to-Market Fit (6 points) - whether ads earn attention
  5. Delivery, Bidding & Spend Control (5 points) - how efficiently budget converts
  6. Full-Funnel & Revenue Alignment (4 points) - whether LinkedIn drives pipeline

Data tiers

  • Tier 1: CSV exports (~19 of 35) - Campaign Performance, Audience Network Performance, Demographics, Conversion Performance reports.
  • Tier 2: + Campaign Manager settings (~27) - Insight Tag status, conversion event config, targeting definitions, exclusion lists, change history.
  • Tier 3: + CRM data (~31) - deals, pipeline stages, revenue attribution, SQL/MQL counts.
  • Tier 4: + creative review (all 35) - ad creative screenshots, landing page URLs, UTM parameters.

Parsing LinkedIn exports

UTF-16 encoded, tab-separated, with metadata rows to skip:

with open(path, 'r', encoding='utf-16') as f:
    lines = f.readlines()
idx = next(i for i, l in enumerate(lines) if 'Start Date' in l or 'Company Name' in l)
df = pd.read_csv(StringIO(''.join(lines[idx:])), sep='\t')

Campaign classification

  • Funnel: TOF/cold/awareness = top. MOF/retarget/RT/30D/90D/180D = mid. BOF/conversion/pipeline = bottom.
  • Right-rail (text/spotlight/follower): NEVER flag for low CTR - that's the format's methodology, it's bought for cheap frequency.
  • Thought Leader ("TL"/"thought leader" in name): typically ~3x CTR and lower CPL vs standard single image.

The 35 checkpoints

Section I: Signal & Data Integrity (1-7)

  1. Insight Tag firing | Tier 2 - Installed and firing on key pages? Without data, flag "Not assessed."
  2. Event prioritization | Tier 1 partial / Tier 2 full - Is each campaign optimizing toward ONE primary conversion aligned with funnel stage? Flag if too many events active.
  3. Conversion event alignment | Tier 1 - Do events measure real pipeline actions (demos, calls, pricing views) not vanity? Fully assessable from CSV.
  4. Volume sufficiency | Tier 1 - 15-20+ conversions/month/campaign for stable optimization? Fully assessable.
  5. Attribution window config | Tier 2 - Windows match sales cycle (B2B = 90-180 days)?
  6. Offline/CRM signal integration | Tier 1 partial / Tier 3 full - Is CAPI active? Are offline conversions imported?
  7. Signal dilution | Tier 1 - Are low-intent events diluting the signal (roughly 10:1 vanity to high-intent)? Fully assessable.

Section II: Architecture & Learning (8-13)

  1. Objective alignment | Tier 1 - TOF=Brand Awareness/Engagement, MOF=Engagement/Web Visits, BOF=Lead Gen/Conversions.
  2. Funnel separation | Tier 1 - Distinct TOF/MOF/BOF tiers? 3+ stages = pass, 1 stage = fail. Assessable from naming.
  3. Budget isolation | Tier 1 - TOF 40-60%, MOF 20-40%, BOF 10-30%, each tier with its own budget.
  4. Naming conventions | Tier 1 - Names include funnel stage, objective, format, audience, size, date?
  5. Internal auction conflicts | Tier 1 partial / Tier 2 full - No two active campaigns bidding on the same audience segment.
  6. Learning phase stability | Tier 2 - Are frequent changes resetting learning? Stopping and starting campaigns hurts optimization.

Section III: Audience & Coverage (14-20)

  1. ICP clarity | Tier 1 partial / Tier 2 full - Targeting is specific; Audience Expansion is OFF (always).
  2. Seniority and role coverage | Tier 1 - Right seniority reached? Low-intent roles consuming impressions? Fully assessable from Demographics.
  3. Company size and industry precision | Tier 1 - Impressions concentrated in target sizes/industries? Fully assessable from Demographics.
  4. ABM list saturation feasibility | Tier 1 partial - List size supports budget without oversaturation? Excessive budget on a small list oversaturates.
  5. Warm audience construction | Tier 1 partial - Site visitors, video viewers, engagers, CRM lists built as retarget audiences? Check windows (30D/90D/180D).
  6. Lookalike usage | Tier 2 - Avoid LinkedIn lookalike expansion; it dilutes a tightly-defined B2B audience.
  7. Audience overlap risks | Tier 1 partial / Tier 2 full - Cold and warm not seeing the same messaging; higher-intent tiers protected.

Section IV: Creative & Message Fit (21-26)

  1. Message-market alignment | Tier 4 - TOF=education/pain, MOF=proof/cases, BOF=urgency/CTA. Without creative, flag "Not assessed."
  2. Format mix | Tier 1 partial - Multiple formats present? Thought Leader ads should exist. Single-format = risk.
  3. Proof vs claims | Tier 4 - Leading with proof (data, screenshots, results) not generic claims?
  4. Differentiation clarity | Tier 4 - Clear differentiation from category noise?
  5. CTA by audience temperature | Tier 4 - Cold = no conversion CTAs, warm = content, hot = demo/meeting.
  6. Creative fatigue | Tier 1 - Ads past format lifespan with declining CTR? Single Image ~4-5 weeks, Thought Leader ~12 weeks. Fully assessable with monthly CTR trends.

Section V: Delivery & Spend Control (27-31)

  1. Bid strategy alignment | Tier 1 - Cold = Max Delivery/CPM, warm = Engagement/CPC, hot = Manual CPC with cap. Map cost type to funnel stage.
  2. CPM inflation | Tier 1 - CPM varying 3x+ across similar audiences signals something inflating. Fully assessable.
  3. Frequency control | Tier 1 - Cold in-feed 3-6, MOF in-feed 8-12, MOF right-rail 15-30. Match creative count to frequency. Fully assessable. (See the companion frequency-and-penetration diagnostic for the full 90-day-band read.)
  4. Budget distribution | Tier 1 - TOF 40-60%, MOF 20-40%, BOF 10-30%.
  5. Spend efficiency vs learning | Tier 1 partial - Stable campaigns learn better; erratic spend resets learning.

Section VI: Revenue Alignment (32-35)

  1. Pipeline influence | Tier 1 partial / Tier 3 full - Can you trace ad exposure to pipeline? Multi-touch tracked?
  2. Cost per SQL / opportunity | Tier 3 - Cost per SQL, per opportunity, per closed-won. Without CRM, flag "Not assessed."
  3. Funnel velocity | Tier 3 - Time through each stage; where deals stall.
  4. Scaling readiness | Always assessed - Ready to scale? Requires clean signal, compound architecture, ICP coverage, message fit, delivery efficiency, revenue proof. If any foundation is broken, scaling amplifies waste. Fix Sections I-III first.

Additional operational checks

  • A. Zombie campaign detection: active campaigns with CTR <0.2% + spend >$200 (in-feed only), $0 conversions + spend >$500, or CPC >5x average.
  • B. Audience Network control: OFF by default; flag if >20% of spend.
  • C. Right-rail coverage: text/spotlight/follower for cheap MOF frequency. Missing = opportunity.
  • D. Demographic refinement: exclude underperformers, scale top performers from the Demographics report.

Health score

Pass = 3, Warn = 1, Fail = 0. Score = (points earned / max possible from assessed checks) × 100. Display: "Health Score: X/100 (Y of 35 assessed)."

Output structure

  • Header: account, date range, health score, breakdown.
  • Critical: all Fails sorted by impact - question, evidence, action.
  • Warnings: all Warns, same format.
  • Passed: compact list with brief confirmation.
  • Not assessed: grouped by data tier, with what data enables each.
  • Ecosystem: layer strength map (fix foundations first) and priority actions.
  • Footer: the attribution line below.

Deeper account-level engagement, website-visitor identification, and demographic "rate" data can be pulled from a LinkedIn signals tool such as DemandSense; note where such data would unlock a Tier 2/3 check.

Design

Dark theme. Suggested tokens: bg #0a1628, blue #0099d1, teal #00c4b3, orange #f4a261, green #22c55e, red #ef4444, purple #a78bfa; JetBrains Mono + a clean sans. Health score with a radial gauge; layer tags as badges on each checkpoint; tables for evidence; mobile-responsive.

Voice

Questions, not statements. Direct verdicts. Specific numbers. Verbs on actions. Acknowledge what works. Never flag right-rail for CTR. No em dashes. Do not invent data.

Attribution footer

End the report with one clickable credit line:

X of 35 points assessed · Built with the LinkedIn Ads 35-Point Inspection by Impactable