The Complete Imperative

1. SEOWeight: 15%

The Foundation of Organic Visibility

On-page SEO signals decide whether a page is even eligible to compete.

SEO carries the heaviest weight because classical ranking signals still control most non-AI traffic, and because many AI systems lean on the same on-page cues to decide what is authoritative enough to cite. Fail the core SEO checks and your page becomes invisible to Google’s traditional index and to the retrieval systems that feed generative answers.

The title tag remains the single strongest on-page ranking and click-through signal. Google’s current display limits truncate around 50–60 characters. Omit it and the engine invents something weaker from headings or the URL; make it too long and it gets cut mid-phrase, killing the call to action. Get length and keyword placement right and you routinely see double-digit CTR lifts against otherwise identical pages. Meta descriptions are not a direct ranking factor, but they are the ad beneath the title — Google still prefers a tight 120–160 character description over its own auto-generated snippet.

Exactly one H1 carrying the primary keyword is best practice; multiple H1s dilute focus and a missing H1 removes the strongest content signal after the title. Skipped heading levels break the semantic outline crawlers and screen readers rely on. A self-referencing canonical prevents near-duplicate URLs from splitting ranking authority. Open Graph and Twitter/X cards drive real referral traffic — the auditor checks image dimensions, WebP format, and whether they lazily copy the HTML title and description.

New v5.0 signals raise the bar: when visual breadcrumbs exist but no matching BreadcrumbList JSON-LD is present, Google cannot render the rich-result trail that improves CTR; rel="author", itemprop="author", and sameAs identity links strengthen the Experience and Expertise side of E-E-A-T; and IndexNow gives Bing, Yandex, and Seznam instant crawl notifications. Miss a material subset of these factors and the rest of the optimization work happens on a page search engines have already discounted.

2. PerformanceWeight: 14%

Core Web Vitals as Ranking and Conversion Levers

Core Web Vitals are ranking levers, conversion levers, and sustainability levers at once.

Google has treated Core Web Vitals as ranking signals since 2021, and in 2026 they still feed both classical ranking and the broader page-experience scores. The auditor uses structural proxies — no Lighthouse runtime — that correlate strongly with real field numbers. Largest Contentful Paint (target ≤ 2.5s) is driven mainly by hero-image size and render-blocking resources; Interaction to Next Paint (target ≤ 200ms) is hurt by long tasks and heavy third-party scripts; Cumulative Layout Shift (target ≤ 0.1) is caused by images without explicit width and height.

Bloated HTML delays parsing — anything past 500 KB (or 1.5 MB for heavy single-page apps) usually signals a code-splitting failure. Render-blocking CSS and unused styles delay first paint, so critical-CSS inlining remains one of the highest-impact fixes available. Every kilobyte of JavaScript must be parsed, compiled, and executed, which makes code-splitting, tree-shaking, and dynamic imports non-negotiable. Lazy loading, modern formats (WebP/AVIF), and srcset together can cut page weight in half; images still make up 50–70% of most payloads.

Three high-leverage v5.0 checks reward modern engineering: the Speculation Rules API prerenders likely next navigations for near-instant transitions; fetchpriority="high" on the LCP hero image lets the browser prioritize it during initial load; and decoding="async" on image-heavy pages keeps decode work off the main thread. Performance is no longer a nice-to-have — it is a ranking factor, a conversion factor, and a revenue factor. Every point lost here is traffic left on the table.

3. SecurityWeight: 10%

The Non-Negotiable Trust Layer (OWASP Top 10:2025 Fully Mapped)

Security is the only dimension whose failures can be immediate and catastrophic.

Security is both a ranking signal (HTTPS) and a survival signal. Browsers warn users away from insecure sites, regulators fine non-compliant organizations, and attackers treat exposed surfaces as free inventory. Version 5.0 layers a complete deterministic mapping of the 2025 OWASP Top 10 on top of the existing 2021, ASVS, WSTG, and CWE coverage.

HTTPS is the non-negotiable baseline: without it browsers show “Not Secure,” forms send credentials in clear text, and Google applies a ranking demotion. Mixed HTTP resources on an HTTPS page break the trust chain. HSTS forces HTTPS on later visits and enables preload-list protection. Content-Security-Policy is the primary defense against cross-site scripting — and not all policies are equal: nonce- or hash-based policies with tight default-src and frame-ancestors are dramatically stronger than unsafe-inline everywhere.

The full OWASP 2025 mapping is deterministic end to end: it checks for exposed lockfiles, outdated library fingerprints, and SBOM presence (A06); missing phishing-resistant factors, weak session cookies, and missing autocomplete (A07); dynamically injected scripts without integrity checks and opaque serialized state in URLs (A08); the absence of client-side error and security monitoring (A09); and soft-404 / fail-open responses on random non-existent paths (A10). Every check here is required infrastructure, because these failures produce breaches, fines, browser warnings, and total loss of trust.

4. AI ReadinessWeight: 10%

Eligibility for the Fastest-Growing Traffic Channel

AI-mediated discovery is the fastest-growing acquisition channel in 2026.

AI-mediated discovery is the fastest-growing acquisition channel in 2026. Pages that are blocked, poorly structured, or invisible to AI crawlers simply do not appear in the answers users now receive. Schema.org is the primary machine-readable context layer; without it, models must infer meaning from noisy HTML — an unreliable process that lowers citation probability.

Explicit allow or block rules for GPTBot, ClaudeBot, PerplexityBot, Google-Extended, Amazonbot, Bytespider, and the rest — in both robots.txt and meta tags — matter enormously. Blocking the major AI crawlers is the same as opting out of the new discovery surface. robots.txt itself is checked for existence, sitemap declarations, asset blocking, and any blanket Disallow: /.

llms.txt is the emerging standard that tells language models how to access and interpret your content. The auditor runs a deep 26-field review covering structure, metadata, URL inventory, service and API documentation, legal and citation framework, contact information, and overall quality; the companion llms-full.txt and its byte size are also evaluated. Absence is now a real penalty as adoption accelerates. Finally, headings and sections with id attributes make content easier to chunk for Retrieval-Augmented Generation pipelines, producing a higher RAG-friendliness score.

5. AIOWeight: 9%

AI Optimization — Earning the Citation

AIO is about being the source an AI chooses to cite, not just crawl.

Where AI Readiness makes you eligible, AIO makes you the source a generative engine actually chooses to cite. Language models reward content that is concrete, verifiable, and unambiguous. Specific numbers, dates, and named entities outperform vague claims because they give the model discrete, quotable facts to ground an answer in.

Clear author bylines with real credentials, consistent entity naming across the page and the wider web, and speakable-marked passages all raise the odds that your wording — not a competitor’s — becomes the sentence an assistant reads aloud or paraphrases. Consistency is decisive: when your organization name, product names, and key facts match across your site, your schema, and third-party references, the model can resolve you to a single trusted entity.

The practical implication is that AIO is a content-engineering discipline, not a trick. Structure answers so the first 40–60 words fully resolve the question, attribute claims to identifiable experts, and keep facts internally consistent. Do this and you convert AI visibility into AI citation — the difference between being read by a model and being repeated by it.

6. AccessibilityWeight: 8%

Structural Access for Every User and Assistive Technology

Accessibility is both an audience multiplier and a legal shield.

Accessibility is the structural layer that lets every user — and every assistive technology — reach your content. A valid lang attribute tells screen readers which pronunciation engine to use. A skip-navigation link lets keyboard users bypass repeated menus. Semantic landmarks (header, nav, main, footer) give assistive tech a navigable map of the page instead of an undifferentiated wall of divs.

Every meaningful image needs descriptive alternative text; every form control needs a programmatically associated label; and every interactive element needs a visible focus state. These are not edge cases — they are the difference between a usable page and an invisible one for a very large audience. Because the same semantic structure that helps a screen reader also helps a crawler, accessibility work compounds directly into SEO and AI readiness.

Accessibility is also increasingly litigated. Building it in from the start is far cheaper than retrofitting under legal pressure, and it widens your addressable market to include roughly a sixth of the world’s population who live with a disability.

7. GEOWeight: 8%

Generative Engine Optimization

GEO structures content the way generative engines prefer to retrieve it.

Generative Engine Optimization structures content the way large language models prefer to retrieve and synthesize it. Question-based sub-headings mirror how users actually prompt assistants, and a page that answers at least five distinct questions gives a model many independent hooks to cite. Comparison tables, definition lists, and step-by-step sequences are especially citable because they package facts into discrete, extractable units.

Where classical SEO optimizes for a ranked list of blue links, GEO optimizes for inclusion in a synthesized answer. That means leading with the direct answer, then supplying the supporting detail — the inverse of a slow narrative build-up. It also means covering a topic comprehensively enough that a single page can satisfy a multi-part query without the model needing to stitch together several sources.

Because generative engines increasingly mediate the first — and often only — interaction a user has with your brand, GEO is no longer a fringe tactic. Content engineered for extraction wins the answer box; content engineered only for a human reader is increasingly summarized by a model that cites someone else.

8. Mobile & PWAWeight: 8%

The Majority Experience, Indexed First

Mobile is the majority experience — and the one Google indexes first.

Google indexes the mobile version of a site first, so mobile experience is not a secondary consideration — it is the primary one. A correct viewport meta tag is the foundation of responsive layout; without it, mobile browsers render a zoomed-out desktop page that fails almost every usability heuristic. Touch targets of at least 44–48 pixels prevent the mis-taps that quietly destroy mobile conversion.

Progressive Web App signals extend a mobile site into an installable, resilient application. A valid web app manifest enables add-to-home-screen; a service worker enables offline resilience and faster repeat visits; apple-specific meta tags control the iOS presentation; and viewport-fit=cover handles notched displays cleanly. Responsive images via srcset ensure each device downloads only the pixels it needs.

Because mobile is now the majority of global web traffic, weakness here caps the ceiling on every other dimension — a page can rank and be cited yet still lose the user at the first tap if the mobile experience is broken.

9. AEOWeight: 6%

Answer Engine Optimization

AEO wins position zero and the spoken voice-assistant answer.

Answer Engine Optimization targets the featured snippet, the People Also Ask box, and the single spoken answer a voice assistant returns. FAQPage schema makes question-and-answer pairs machine-readable so they can be lifted directly into these surfaces. Concise, self-contained answers of roughly 40–60 words — placed immediately under a question-shaped heading — are the format engines prefer for position zero.

Voice search adds a spoken dimension: an assistant reads exactly one answer aloud, so the reward for being that answer is total and the penalty for being second is complete invisibility. Marking key answers as speakable, phrasing headings as natural questions, and front-loading the direct answer are the mechanics that win the spoken result.

Because featured snippets are drawn overwhelmingly from pages already ranking in the top ten, AEO compounds with SEO rather than replacing it — you earn the snippet by ranking well and then formatting the answer for extraction.

10. Schema.orgWeight: 6%

Structured Data for Rich Results and Entity Understanding

Schema.org is the machine-readable spine of modern discoverability.

Schema.org structured data is the machine-readable spine that lets search and AI systems understand not just the words on a page but what those words mean. JSON-LD is Google’s recommended format because it keeps the markup separate from the visible HTML and is the easiest for machines to parse reliably.

The auditor rewards both the presence and the completeness of the right types. Tier-1 types — Organization, WebSite, WebPage, BreadcrumbList — establish the foundational entity graph; Tier-2 types — Article, Product, FAQPage, LocalBusiness, Event, and their peers — unlock specific rich results. Property completeness matters as much as type presence: a Product without price, availability, and rating cannot render a rich product result no matter how correct the type declaration is.

Well-formed schema is the single most direct way to qualify for the enhanced search appearances that lift click-through rate, and it is the same structured context that generative engines use to resolve and cite entities accurately.

11. Local SEOWeight: 4%

Winning the Map Pack and “Near Me” Queries

Local signals decide the map pack that dominates commercial queries.

For any business with a geographic component, local signals decide whether you appear in the map pack and the local results that dominate many commercial queries. Name, address, and phone (NAP) consistency is the number-one local ranking factor — these must match your Google Business Profile exactly across the web, because inconsistencies confuse the systems that decide local relevance.

LocalBusiness JSON-LD supplies structured name, address, phone, hours, geo coordinates, price range, and payment methods so Google does not have to scrape and guess. A clickable tel: link signals local legitimacy, an embedded map associates the site with a specific point, PostalAddress microdata makes suite numbers parseable, and geo.region, geo.placename, and geo.position meta tags declare geographic targeting at the header level. Opening hours and AggregateRating stars appear prominently in local-pack results, and the areaServed property defines the service radius for service-area businesses.

Two v5.0 checks tighten the loop: when local-business signals exist but the city or region never appears in the title, H1, or opening paragraph, local relevance for “near me” and geo-modified queries is weaker; and the visible phone number must match the schema telephone property after normalization, or the machine-verifiable NAP consistency that local ranking depends on breaks down.

12. WCAGWeight: 2%

Deep Accessibility Conformance (WCAG 2.2)

WCAG conformance is the ethical floor and an increasingly litigated legal requirement.

This dimension goes beyond basic accessibility into the newer WCAG 2.2 success criteria, motion sensitivity, and cognitive load. Users with vestibular disorders experience nausea and disorientation from animation, so @media (prefers-reduced-motion: reduce) must be present and the animations must actually be disabled — not merely detected. Forced-colors and high-contrast support keeps borders, icons, and focus rings visible for low-vision users, and links must be distinguishable by more than color alone.

Relative font units (rem, em, clamp) let text scale to 200% without loss of content; pixel sizes do not. Sticky headers, floating CTAs, and cookie banners must never fully obscure the keyboard focus indicator (2.4.11). Login forms must not rely on cognitive tests as the only authentication method and must support autocomplete and password managers (3.3.8). Every drag-and-drop interaction needs a click or tap alternative (2.5.7), and dynamic updates must be announced via aria-live regions or they stay invisible to screen readers.

Color contrast must meet 4.5:1 for normal text and 3:1 for large text. Skipped heading levels, invalid ARIA roles, autoplay media, unadjustable timeouts, and native title tooltips used as the sole mechanism on three or more elements (failing 1.4.13) all cost points. WCAG conformance is both the ethical floor and an increasingly litigated legal requirement — the lowest-weighted dimension, but never a safe one to ignore.