How the diagnostic works
The methodology, in the open
Ten pillars. One perception gap. One composite score. Five tiers. We publish the conceptual model so you can decide whether the framework is sound before you trust the output. We reserve the algorithmic specifics — weights, signal extraction, model prompts.
What we publish, what we reserve
A diagnostic earns trust through methodological transparency. It earns durability through algorithmic protection. We do both deliberately.
Published
- The ten pillars — named, defined, scoped
- The four-layer marketing stack model we sit at the top of
- The perception-gap mechanic (math + philosophy)
- The MIDAS composite (four sub-components + weights)
- The five MIDAS tiers and what each one means
- The signal sources the analysis draws from
Reserved
- Per-pillar scoring weights and rubrics
- Signal extraction logic and routing rules
- Model prompt structures and consensus formulas
- MIDAS sub-component weighting formula across business types
- Detected vs declared tie-breaking when signals disagree
The ten pillars
Each pillar is a surface of your marketing infrastructure that produces signal independently. We score them separately so a weak pillar can't be hidden behind a strong one.
Attribution
Do you measure what marketing dollars actually produce?
Tracking accuracy, event hygiene, attribution-model alignment, server-side vs client-side coverage, consent-blocking analysis. Cross-checks claims (analytics dashboards) against detected behaviour (tag firing order, pixel state, conversion-count parity across tools).
Tech Stack
Are the tools you pay for doing the work you bought them for?
Tool inventory by category, redundancy detection, agentic-readiness sub-score (Conventional 0.7 / Agentic 0.3), signal-of-life checks (is this tool actually writing to your stack), and integration coherence between detected tools.
Conversion
Can a visitor actually become a customer without friction killing the path?
Form submission tracking, CTA event firing, checkout flow analysis, required-field hygiene, mobile vs desktop parity. Tests whether the conversion measurement matches the conversion experience.
Trust & Security
Does your stated posture match your detected behaviour?
DMARC/SPF/DKIM posture, consent banner enforcement, privacy policy vs detected tag analysis, SSL certificate state, third-party script consent gating, cookie scope vs declared use.
Brand
Is your voice the same person across every surface a prospect sees?
Tone consistency across website / social / ads / email (LLM-evaluated), NAP (Name/Address/Phone) consistency, brand sentiment from public review surfaces, citation alignment across knowledge graphs.
Marketing
Are the channels you run actually a system, or just running in parallel?
CRM detection + lifecycle automation signals, email sequence detection, social cadence and channel coherence, audience segmentation evidence across detected platforms.
Advertising
Do your ad-platform claims match what your site can actually receive?
Ad-platform tag detection, UTM hygiene, landing-page-to-ad alignment, attribution inflation checks across platforms, ad-account-state signals where externally observable.
Competitors
Do you know which competitors AI systems compare you against?
AI-powered competitor discovery via Gemini + Perplexity, URL validation and de-hallucination, positioning gap analysis, AEO surface comparison across LLM engines, brand mention frequency in AI responses.
Product
Can a first-time visitor tell what you do in under six seconds?
Above-the-fold messaging audit, pricing presentation, offer clarity scoring, product/marketing-story coherence, value proposition specificity.
Web / SEO / GEO
Are you discoverable by both Google and the systems replacing it?
Core Web Vitals (LCP / FID / CLS), schema markup coverage, canonical hygiene, indexation health, robots.txt AI-crawler directives, llms.txt presence, FAQPage / Article / Organization schema completeness, structured-data validity.
Where the diagnostic sits in your stack
Most marketing stacks have three layers — execution, analytics, automation. Almost none have the fourth: infrastructure management. The diagnostic operates at that fourth layer, testing whether the assumptions the first three layers depend on are actually true.
The four-layer model is published in full on a separate resource page, with the layer-by-layer breakdown and what each layer assumes about the layer beneath it.
Read the four-layer model →The perception-gap mechanic
The diagnostic is two halves running in parallel. The gap between them is the most revealing thing it produces.
Side A
Your self-assessment
You rate your own marketing infrastructure across the ten pillars. One slider per pillar, takes about a minute. We assume you know your business — this is the operator's view.
Side B
Independently detected score
In parallel with your self-assessment, our system scrapes public signals, queries multiple LLMs, validates URLs, and produces a score per pillar from no opinion at all. This is the outside-observer view.
The math
Gap = Side A − Side B
Computed per pillar. A positive gap means you rated yourself higher than detected reality (the more common direction); a negative gap means detected reality is ahead of your perception (you're under-selling something that's working).
The philosophy
Most marketing dashboards measure outcomes (what happened) or activity (what we did). Almost none measure the gap between what you think is happening and what is actually happening. That gap is where surprise lives, where unbudgeted risk hides, and where the highest-leverage corrections sit.
The Yellowhead Diagnostic names that gap quantitatively, per pillar, and produces a roadmap from it. A pillar with a small gap and a low score is a known weakness. A pillar with a large gap is a blind spot.
The MIDAS™ composite
Ten pillar scores, plus a perception gap, plus a coherence read, plus a budget signal is a lot of numbers to hold at once. MIDAS rolls them into a single 0–100 score that lets you compare reports over time or against an industry baseline.
Pillar Performance
60%Measures
Your detected pillar scores, weighted by business impact.
Why it's in the composite
Detects which specific surfaces of your marketing infrastructure are healthy and which are degraded.
Perception Accuracy
15%Measures
How closely your self-assessment matches the independently detected reality.
Why it's in the composite
Detects how well you know your own infrastructure. Large gaps in either direction (over- or under-rating yourself) are themselves a finding.
Infrastructure Coherence
15%Measures
Whether related systems reinforce each other, or operate as parallel silos.
Why it's in the composite
A well-funded stack of disconnected tools scores lower than a thoughtful stack of connected ones. Coherence is what compounds.
Budget Efficiency
10%Measures
The ratio between detected investment signals and detected outcome signals.
Why it's in the composite
Detects when spend has outpaced strategy — when you're buying capability faster than you're using it.
The composite passes through a logarithmic curve so foundational gaps weigh more heavily than incremental gains.
The five MIDAS tiers
The composite score maps to one of five MIDAS tiers. Tier names communicate posture, not just rank. They tell you what state your infrastructure is in, not just how it compares to others.
Exceptional
Best-in-class infrastructure across pillars, with strong cross-system integration and accurate self-awareness. Rare — the bar this tier sets is real, and most stacks never reach it.
Strong
Well-integrated infrastructure with minor gaps. Growth-ready. The systems are talking to each other and the team understands their actual position.
Operational
Working infrastructure with optimisation headroom. Pillar scores are mixed and some blind spots exist, but the operator is broadly aware of them. The most common tier.
Exposed
Functional but blind spots create material risk. Perception gaps are large in at least one pillar. Decisions are being made on stale or incomplete information.
Critical
Foundational gaps. Infrastructure is actively leaking value — compliance, security, attribution, or brand surfaces are detectably broken. Either unknown to the operator or known and unaddressed.
Run the methodology on your own infrastructure
The diagnostic is free and takes five to ten minutes end to end. The framework above is what it runs.
Start Your Free DiagnosticNo credit card. No commitment.