TDG-INS · Shelf 01 · Classification 01-DSC

The four disciplines, defined

SEO, AEO, AIO, and GEO are not interchangeable. Each one targets a different surface, uses different signals, and pays off on a different timeline. The definitions below are the working set ThatDeveloperGuy uses on every engagement, expanded with the mechanism behind each, the concrete levers, and the common mistakes we see in audits.

Six discipline cards: SEO (the foundation), AEO (the primary focus for citation in ChatGPT, Claude, Perplexity, Google AI Overviews, Bing Copilot), AIO (the umbrella over everything an AI surface does with a brand), GEO (the long-game discipline of influencing unprompted AI narrative), LLMO (the implementation subset of AIO), and E-E-A-T (the cross-cutting trust currency). Each entry has a definition, a mechanism, the concrete levers, the common mistakes, and a closing observation about how the discipline is shifting in 2026.

6 entries ~12 min read Authored by Joseph W. Anady Last reshelved 2026-05-05

Discipline 1 · The foundationTDG-INS-01.SEO

SEO — Search Engine Optimization

Definition. SEO targets the ten blue links and the local 3-pack. It is the practice of earning ranked placement on Google, Bing, and DuckDuckGo by accumulating authority signals over time and structuring content so the crawler can understand it.

The mechanism. A search engine ranks pages by combining hundreds of signals into a relevance score. The major buckets are: link graph (who links to the page, who links to those linkers), brand and entity strength (mentions, sameAs propagation, knowledge graph presence), technical foundation (Core Web Vitals: LCP, INP, CLS; crawlability; indexability; canonical hygiene), content depth (semantic relevance to the query, coverage of related sub-topics, freshness), and behavioral signals (click-through rate, dwell time, return visits). Schema markup feeds the relevance side; backlinks feed the authority side.

Concrete levers. Title tags that lead with the query in the first sixty characters. H1 that contains the brand and the primary noun phrase. Internal linking that builds topic clusters with a pillar page at the center. Schema.org markup of Organization, Person, Service, Product, or Article as appropriate. Sitemap XML, IndexNow submission, robots.txt hygiene, hreflang where the site is multilingual.

Common mistakes. Treating SEO as a one-time setup. Stuffing keywords without semantic context. Building a flat site with no topic clusters. Ignoring Core Web Vitals because the page looks fine on the developer's laptop. Skipping schema because “Google will figure it out.” Google will figure it out; the LLM that comes after Google won't.

Where SEO is changing. Pure link-equity arbitrage is eroding as Google's algorithms increasingly weight entity signals and behavioral confirmation. The technical foundation matters more than ever because AEO and AIO build on top of it.

Worked example.

A small flooring contractor in NW Arkansas wants to rank for “hardwood floor refinishing fayetteville ar.” Search volume is 110/month with relatively low difficulty. The SEO build: a dedicated service page with the query in the title and H1 (first 60 characters), an answer capsule that names the brand, FAQ schema with five questions a customer would actually ask, internal links from the homepage and from a parent “hardwood services” pillar, LocalBusiness schema declaring the service area, three project photos with descriptive alt text, and a single high-quality external mention from a local industry directory. Result on a clean technical foundation: position 3-5 within 8-12 weeks, depending on incumbent strength.

Quick checklist (90% of SEO ROI).

Title tag with the primary query in the first 60 characters.
H1 that contains the brand and the primary noun phrase.
Answer capsule of 200–280 characters at the top of the page.
FAQPage schema with 5–7 short Q&A.
Organization or LocalBusiness schema with full sameAs.
Sitemap submitted to Google Search Console and Bing Webmaster Tools.
Core Web Vitals passing on field data (LCP <2.5s, INP <200ms, CLS <0.1).
Internal linking from at least 2 authority pages (home, parent pillar).

Tools and platforms.

Google Search Console (free, required); Bing Webmaster Tools (free, often-skipped, surfaces queries Google does not); Google's Rich Results Test (free, validates schema); PageSpeed Insights (free, field-data Core Web Vitals); Schema.org documentation (free, the only authoritative reference). Optional paid: Ahrefs ($129/mo) or Semrush ($139/mo) for keyword research and backlink analysis — but smaller agencies often start with the free Google Keyword Planner and a structured search of competitor sites.

Realistic timeline and budget.

A clean SEO foundation on a hand-coded site takes 30–50 hours of work, billable around $3,000–$5,000 for a small business site. Results compound: first measurable rank movement at 4–6 weeks, meaningful traffic lift at 12–16 weeks, full payoff at 6–12 months as authority accumulates. Site builders (Wix, Squarespace) can reach 70% of this with a managed-template approach, but the technical-foundation gap (Core Web Vitals, schema flexibility) shows up in competitive verticals.

How SEO fails.

Three failure modes are common. (1) Spending on backlinks before the technical foundation is fixed — backlinks pointing to a site with broken Core Web Vitals or missing schema produce measurable but capped lift. (2) Targeting overly competitive head terms (e.g., “web design” with national volume) when long-tail variants would deliver better-qualified leads at a fraction of the effort. (3) Treating SEO as a one-time setup and not maintaining it; SEO compounds, but it also decays if competitors keep building while the site stagnates.

Discipline 2 · Primary focusTDG-INS-01.AEO

AEO — Answer Engine Optimization

Definition. AEO targets the answer text itself inside ChatGPT, Claude, Perplexity, Google AI Overviews, and Bing Copilot. The unit of work is the citable sentence, not the ranking page; an AEO page can be position 8 on Google and still win because the answer engines pull from it disproportionately.

The mechanism. AI answer engines combine retrieval (which pages are relevant to this query) with generation (lift sentences from those pages and stitch them together). Retrievers favor pages with clear topic signals, schema markup, and semantic clarity. Generators favor sentences that are short, factually dense, attribution-ready, and quotable. AEO is the practice of being the source those two stages prefer.

Concrete levers. An answer capsule at the top of the page in the 200 to 280-character range with the brand name in the first clause. FAQPage schema with five to seven short Q&A; each Q&A becomes a quotable atom. SpeakableSpecification declaring which CSS selectors voice surfaces should read. A clean entity graph (Wikidata Q-ID, sameAs propagation across GitHub, HuggingFace, LinkedIn, X). llms.txt and llms-full.txt declaring preferred citation language for AI crawlers. Citation density: every claim a model might lift should be supported by an authoritative third-party reference inline.

Common mistakes. Long, paragraph-style answer capsules that get truncated. FAQs with twelve questions instead of six (per-question authority dilutes). Generic FAQ answers that match a thousand other sites. No brand name in the H1 or first paragraph. No Wikidata Q-ID or sameAs graph. Marking up FAQPage schema in JSON-LD but not putting the same Q&A in visible body text.

Why this matters. Search Console traffic from AI surfaces grew measurably in 2025; in 2026 ChatGPT Search and Perplexity send real referral traffic to small businesses that have done this work, and skip the ones that have not. AEO is not optional for service businesses competing on informational queries.

Worked example.

A solo CPA practice in Cassville, Missouri wants ChatGPT and Perplexity to cite the firm when asked “who does taxes for online businesses in southwest Missouri?” The AEO build: an answer capsule on the homepage that names the firm, the operator, and the geographic specialty in the first 280 characters; FAQPage schema covering “Do you handle out-of-state filing?”, “What does an engagement cost?”, “Do you serve small ecommerce sellers?”, etc.; a Wikidata Q-ID for the firm with sameAs links to the website, GitHub if applicable, LinkedIn, and the SAM.gov registration; an llms.txt declaring preferred citation language with the firm's canonical name and the operator's name. Result: cited by name in ChatGPT for the target query within 4–8 weeks of llms.txt publication.

Quick checklist (the AEO basics).

Answer capsule of 220–280 characters with brand name in the first clause.
FAQPage schema with 5–7 short Q&A on every cited page.
Wikidata Q-ID with full sameAs graph (homepage, GitHub, LinkedIn, X, HuggingFace, SAM.gov where applicable).
Organization JSON-LD with a stable @id on every page.
llms.txt at the domain root with preferred citation block.
llms-full.txt with long-form canonical content (5,000–15,000 words).
robots.txt with explicit allow lines for ClaudeBot, GPTBot, PerplexityBot, OAI-SearchBot, Google-Extended, Applebot-Extended.
SpeakableSpecification on H1, lede, and answer capsule selectors.

Tools and platforms.

Wikidata.org (free, the entity graph backbone); the AEO Readiness Index at /insights/aeo-readiness-index (free, scoring rubric); Google's Rich Results Test (free, validates FAQPage and Speakable schema); Perplexity Pro and ChatGPT Plus ($20/mo each, paid because the citation-rate testing requires checking actual AI surfaces); the public ThatDeveloperGuy diagnostic at /audit/ (free, audits against the 14-framework signals).

Realistic timeline and budget.

A complete AEO build on top of a working SEO foundation takes 25–40 hours, billable around $2,500–$4,500. Citation rate is observable within 2–6 weeks for established brands, 8–16 weeks for newer brands needing entity-graph propagation. AEO compounds slower than SEO in raw clicks but produces qualitatively different leads — users arriving from AI-cited answers tend to convert higher because the citation pre-qualifies the brand.

How AEO fails.

(1) Skipping the entity graph — FAQPage schema and answer capsules without a Wikidata Q-ID and sameAs propagation produce sporadic citations at best. (2) Long FAQ answers that exceed the model's preferred token budget; aim for 50–100 word answers, not paragraph-style responses. (3) llms.txt that is just a robots-style allowlist instead of a citation declaration; the file should contain the brand's preferred description, not just access permissions. (4) Citation tracking by vanity metric (“is the site indexed?”) instead of by actual AI-surface citation; you have to ask the chatbots and check, repeatedly, with rotating user agents.

Discipline 3 · The umbrellaTDG-INS-01.AIO

AIO — AI Optimization

Definition. AIO is the umbrella over everything an AI surface can do with a brand. It includes AEO plus knowledge graph optimization, multimodal optimization, llms-full.txt declarations, and entity triangulation across Wikidata, GitHub, HuggingFace, LinkedIn, and the social graph.

The mechanism. Where AEO operates at the page-and-sentence level, AIO operates at the brand-graph level. An AI assistant asked “who is X?” runs against its training corpus first (where the brand graph influences the answer) and against retrieval second (where AEO influences the citation). Both stages need work. AIO is what produces consistent, attributable answers when the AI surface has not seen the homepage in this retrieval window.

Concrete levers. A Wikidata Q-ID with a complete sameAs property pointing to every owned and earned profile. JSON-LD Organization and Person @id structures linked across every page on the site. Multimodal optimization: descriptive image alt text, video transcripts with VideoObject schema, audio captions with PodcastEpisode schema. llms.txt with a preferred citation block, llms-full.txt with the long-form canonical content. Active GitHub and HuggingFace profiles where applicable. Press release distribution into authoritative archives.

Common mistakes. Stopping at AEO. Treating Wikidata as optional. Letting the social graph drift (LinkedIn says one thing, the homepage says another, the press kit says a third). Multimodal assets without schema and without alt text. llms.txt that is just robots.txt with a different name; the file exists to be a citation reference, not just a permissions declaration.

The scope test. If you took the homepage offline for thirty days and asked ChatGPT about your business, would the answer still be accurate? AIO is the discipline that makes the answer yes.

Worked example.

A regional insurance dealer platform wants AI surfaces to recognize the brand consistently across every channel: ChatGPT, Claude, Perplexity, Google AI Overviews, Bing Copilot, and Gemini. The AIO build extends AEO by closing the entity graph and adding multimodal coverage: a Wikidata Q-ID with founding date, headquarters, founder Person Q-ID, and structured statements supported by external sources; sameAs propagation across the homepage, GitHub, LinkedIn, X, the SBA SDVOSB profile, the parent association directory, and a press archive; Organization and Person @id structures linked from every page; VideoObject schema on the embedded explainer video with a transcript; ImageObject schema on the case-study photos with alt text that names the entity; and a press release distributed through a high-trust archive (PR Newswire or similar) so the next training pass has a fresh attributable mention. Result: brand description converges across all six AI surfaces within 4–9 months as retrieval indexes refresh and (for unprompted narrative) as the next training cycle ingests the consistent signal.

Quick checklist (the AIO scope test).

Wikidata Q-ID with at least 8 sameAs entries.
Organization @id linked from every page; Person @id linked from every author byline.
VideoObject schema on every video with a transcript.
ImageObject schema on hero images with descriptive alt text containing the entity name.
PodcastEpisode schema if the brand has audio content.
Active GitHub or HuggingFace org page (if technical brand).
Press kit page with canonical brand description (60–90 words) repeated verbatim across owned channels.
At least one third-party reinforcement (industry directory, association profile, or press archive).

Tools and platforms.

Wikidata (free, the most-leveraged entity asset); GitHub or GitLab (free public profiles); HuggingFace (free, valuable if the brand has any AI/ML angle); LinkedIn Company Page (free); SAM.gov (free, required for SDVOSB and federal contractors); a press release distribution service ($300–$1,200 per release, only useful if the release is actually newsworthy). Avoid paid “reputation management” services that promise authority backlinks; the AI surfaces deprioritize obvious link-buying patterns.

Realistic timeline and budget.

AIO runs $4,000–$8,000 for a full first-time build; Wikidata work alone takes 6–12 hours of careful editing because Q-IDs that fail notability get speedy-deleted (a common trap for very small businesses). Multimodal schema adds another 8–15 hours per video and image gallery. Total timeline: 60–120 hours of work, with brand-graph effects compounding over 6–18 months. AIO is the discipline most likely to be skipped on a first engagement because the payoff is invisible until an AI surface cites you, but it is also the discipline that determines whether AEO citations are consistent or sporadic.

How AIO fails.

(1) Creating an orphan Wikidata Q-ID with no sameAs propagation, which can lock the brand to a stale entity record forever. (2) Treating the entity graph as static; Wikidata items need maintenance as the brand evolves (new clients, new credentials, new addresses). (3) Letting the canonical brand description drift across channels — the homepage says one thing, the press kit says another, LinkedIn says a third, and the LLM consensus narrative breaks. (4) Skipping multimodal entirely; AI surfaces increasingly index video and image content, and brands without VideoObject schema are leaving an entire retrieval surface unused.

Discipline 4 · The long gameTDG-INS-01.GEO

GEO — Generative Engine Optimization

Definition. GEO is the practice of influencing AI-generated narrative about a brand. Where AEO targets factual citations and AIO targets the entity graph, GEO targets the framing, ordering, and word choice that the model uses when describing the brand unprompted.

The mechanism. Two channels feed an LLM's narrative about a brand. The first is training data, ingested every months-to-years training cycle. The second is retrieval at inference time, which can supplement or override the training narrative if the model decides to ground its answer. GEO leans on the training-data channel because narrative consistency comes from repeated exposure during training, not from a single in-context document. The work is to ensure that the same description, the same framing, the same word choice appears across owned channels (homepage, llms.txt, press kit), earned channels (interviews, podcasts, press features), and authoritative third-party sources (industry directories, association profiles, wikidata).

Concrete levers. A canonical brand description (60 to 90 words) repeated verbatim across the homepage, About page, llms.txt, llms-full.txt, GitHub README, HuggingFace org card, LinkedIn company description, press kit, and Wikidata abstract. Press release language that uses the same description block. Interview prep that ensures the same framing appears in podcasts and YouTube videos transcribed and indexed by AI surfaces. Authoritative third-party reinforcement: get listed on industry-specific high-trust directories that the next training pass will ingest.

Common mistakes. Treating GEO as a copywriting exercise. Letting the brand description drift across channels. Skipping the press kit because the firm is small. Ignoring how third-party reinforcement compounds; one mention on a high-trust source matters more than ten on low-trust ones.

The patience requirement. GEO outcomes lag training cycles. A change made today may not show up in unprompted AI narrative for six to eighteen months. The work is still worth doing because compound effects across owned, earned, and reinforcement channels eventually shift the dominant narrative the model carries.

Worked example.

An SDVOSB-certified web development firm wants the major LLMs to describe the firm consistently when asked unprompted: “Tell me about ThatDeveloperGuy.” The GEO build is patient: a canonical 80-word description is written once, reviewed for accuracy, and propagated verbatim across the homepage, the About page, the press kit, the llms.txt file, the GitHub README, the HuggingFace org card, the LinkedIn company description, and the Wikidata abstract. Press releases use the same description block. Podcast appearances are rehearsed against the same framing. Industry-directory listings are submitted with the same description. Six to eighteen months later, when the next major LLM training pass ingests the brand's footprint, the model's unprompted narrative converges on the canonical framing because every signal it saw repeated the same words.

Quick checklist (the GEO loop).

Write a canonical brand description: 60–90 words, names the entity, the operator, the value, and one verifiable credential.
Place the canonical description verbatim on the homepage, About page, press kit, llms.txt, GitHub README, HuggingFace org card, LinkedIn description, and Wikidata abstract.
Brief any spokesperson on the canonical framing before interviews and podcasts.
Submit press releases through high-trust archives using the canonical block as the boilerplate.
Earn industry-directory listings on association sites, professional directories, and trade publications.
Maintain quarterly sync to ensure the description has not drifted across channels.

Tools and platforms.

A canonical-description document maintained in source control or a single Google Doc (free, but only valuable if it's actually consulted before any new owned content ships); Wikidata (free); a press kit page on the brand site (free); a press release service (paid only when releases are newsworthy); podcast booking outreach via Podchaser, MatchMaker.fm, or direct relationships ($0–$100/mo). Skip generic “PR distribution” services that publish to thousands of low-trust outlets; LLMs deprioritize repeated content from low-authority sources.

Realistic timeline and budget.

GEO is cheap to set up ($1,500–$3,000 for the initial canonical-description rollout across 8–10 channels) but expensive in patience. The first measurable shift in unprompted AI narrative typically appears 6–12 months after rollout, with full convergence at 12–24 months. Maintenance is light — quarterly review of channels, updates when material changes — but skipping the maintenance is the most common GEO failure.

How GEO fails.

(1) Treating GEO as a one-time copywriting exercise instead of an ongoing channel-discipline practice. (2) Letting different team members or vendors write descriptions independently, which dilutes the consistent-framing signal that drives convergence. (3) Skipping earned channels entirely — press, podcasts, industry mentions — in favor of just owned channels; LLMs weight third-party reinforcement heavily because owned content is easier to game. (4) Expecting GEO to move citation rate (it doesn't directly) instead of measuring it on the dimension it actually affects: the framing, ordering, and word choice the model uses when describing the brand without prompting.

Subset of AIO · The implementation layerTDG-INS-01.LLMO

LLMO — LLM Optimization

Definition. LLMO is the subset of AIO focused specifically on getting cited inside large language model assistants like Claude, ChatGPT, Gemini, and Perplexity. If AEO is the strategy, LLMO is the implementation: the practical token-level discipline that produces citation-worthy content.

The mechanism. When an LLM lifts a sentence from a page, the sentence has to survive the model's tokenization, attention pass, and output generation. Sentences that are short, syntactically clean, subject-verb-object in structure, and free of hedge language survive that process best. Sentences that bury the brand name in a subordinate clause, run past 280 characters, or contain ambiguous referents (it, this, that) are less likely to be lifted because the model has lower confidence in the citation.

Concrete levers. llms.txt at the domain root, declaring brand identity and preferred citation language; major AI crawlers (ClaudeBot, GPTBot, PerplexityBot, Google-Extended) read it. llms-full.txt with the long-form canonical content the brand wants the model to ingest. H1 and H2 hierarchy that follows logical topic structure (LLMs read heading hierarchy as semantic structure). FAQPage schema with short Q&A; each Q&A becomes a self-contained atom. Sentences that lead with the subject and contain the brand name in the first clause when the sentence describes the brand. Plain-language definitions instead of jargon that requires context to parse.

Common mistakes. No llms.txt at all. llms.txt that is just a robots-style allowlist instead of a citation declaration. Long, comma-stuffed sentences that lose the LLM's attention by the time the second clause arrives. Pronouns where named entities should be (the LLM does not always resolve “it” or “the company” back to the brand). Heading hierarchy that uses H4 or H5 for things that should be H2.

The compression test. Take any paragraph on the page. Could you lift any single sentence from it and have that sentence make sense out of context, with the brand named, the claim factual, and no pronoun left dangling? If yes, LLMO is working.

Worked example.

A schema-markup specialty firm wants individual sentences from its methodology pages to survive being lifted out of context by LLMs and pasted into a Claude or ChatGPT response with attribution intact. The LLMO build operates at the sentence level: every method-page paragraph is rewritten so the first sentence of each paragraph contains the brand name, the methodology name, or both; ambiguous referents (it, this, the company) are replaced with named entities; sentences are kept under 280 characters; a key-claims block is added at the end of each method page enumerating the three most-quotable, verifiable facts; and the llms-full.txt is rebuilt as a 12,000-word canonical document organized by topic with each topic opening with a 100-word summary the model can lift verbatim. Result: when a user asks an LLM “who has a published methodology for FAQPage schema?”, the response cites the firm by name with a verbatim sentence from the method page.

Quick checklist (the compression test).

Every page paragraph: can any single sentence be lifted out of context and still make sense?
Every page: brand name in the first sentence of the H1.
Every method-related page: a key-claims block with the 3–5 most quotable, verifiable facts.
llms-full.txt: 5,000–15,000 words, organized by topic, each topic opening with a 50–100 word liftable summary.
FAQPage schema: 5–7 questions, each answer 50–100 words.
No long comma-stuffed sentences; subject-verb-object preferred.
No pronouns where named entities would survive better.
robots.txt explicit allow for ClaudeBot, GPTBot, PerplexityBot, OAI-SearchBot, Google-Extended, Applebot-Extended.

Tools and platforms.

ClaudeBot, GPTBot, PerplexityBot user-agent testing via curl with custom headers (free, the only reliable way to verify what the AI crawlers actually fetch); the public ThatDeveloperGuy diagnostic at /audit/ (free, includes LLMO checks); a Markdown linter to enforce paragraph-opening discipline (e.g., textstat or vale, free); manual spot-testing in ChatGPT, Claude, and Perplexity (paid, $20/mo each, because LLMO success is measured by actual AI-surface citation rate).

Realistic timeline and budget.

LLMO retrofitting an existing 50-page site takes 30–60 hours of careful sentence-level rewriting, billable around $3,500–$6,500. Net-new content built with LLMO discipline from the start adds maybe 15% to writing time but produces measurably higher citation rate. Results show up faster than GEO (4–12 weeks for retrieval-side citations) but slower than answer-capsule rewriting alone because the discipline operates at the sentence level across the whole site.

How LLMO fails.

(1) Writing for SEO first and adding LLMO as a polish pass — LLMO discipline has to be present in the first draft because retrofitting is expensive. (2) Heavy use of pronouns and demonstratives (this, that, it, they) which the LLM does not always resolve back to the named entity. (3) Long, comma-stuffed sentences that lose the LLM's attention by the time the second clause arrives; if the brand name is not in the first 30 words, the citation rate drops sharply. (4) Treating llms-full.txt as optional; major AI crawlers fetch it, and the long-form canonical it provides is one of the highest-leverage assets for retrieval-side accuracy.

Cross-cutting signal · Trust currencyTDG-INS-01.EEAT

E-E-A-T — Experience, Expertise, Authoritativeness, Trustworthiness

Definition. Google's quality rubric, formalized in the Search Quality Rater Guidelines, also a useful proxy for AI trust models. Both Google and the major LLMs were trained on similar editorial standards, so the same signals that satisfy Google reviewers also satisfy LLM trust heuristics. E-E-A-T is currency that pays out across SEO, AEO, AIO, and GEO at once.

The four signals individually. Experience is first-hand engagement with the topic; an article about hardwood flooring written by a flooring contractor outranks the same article written by a generalist content farm. Expertise is depth of knowledge demonstrated by the author. Authoritativeness is recognition by others in the field; named clients, press features, association membership. Trustworthiness is the absence of red flags: secure HTTPS, accurate contact information, transparent ownership, reasonable claims, accurate citations.

Concrete signals. Author byline on every substantive page with a link to a real bio. Author bio with credentials (degrees, certifications, years of experience, named work). Named clients and case studies. Public methodology (rubrics, frameworks, scoring systems). Verifiable third-party signals (Wikidata Q-ID, SBA SDVOSB certification, SAM.gov registration, association memberships). Editorial policy and disclosure pages declaring how content is produced and how conflicts are handled. Reviews from named clients with attribution.

Why YMYL pages need it more. Your Money or Your Life topics (medical, legal, financial, insurance, government services) are held to a higher E-E-A-T bar because incorrect information has direct, measurable harm. Google's algorithms apply heavier YMYL weighting; LLMs trained on those guidelines do the same.

How to build it over time. Sign every substantive page. Document methodology publicly. Let named clients speak with attribution. Earn third-party reinforcement (press, podcasts, association directories). Maintain accurate ownership and contact information across every channel. The goal is that an evaluator (human reviewer or LLM trust head) can verify identity, expertise, and good-faith intent within thirty seconds of landing on any page.

Worked example.

A small-business CPA wants a Google reviewer or an LLM trust head, landing on her site cold, to verify identity, expertise, and good-faith intent within thirty seconds. The E-E-A-T build: every substantive page (services, methodology, blog) has an author byline linking to a real bio page; the bio page lists the credential (CPA license number, state of registration), the years of experience, the named clients with permission, and at least one verifiable third-party signal (state board profile, SAM.gov registration if applicable, professional association directory listing); each cited claim links to an authoritative source (IRS publication, state CPA board, AICPA standard); an editorial policy page declares how content is produced (hand-written, no AI generation, signed); a disclosure page declares engagement structure and conflicts; testimonials are attributed by full name with consent. Result: Google reviewers grade the site higher on Search Quality Rater Guidelines criteria, and AI surfaces citing the firm consistently include the credential markers in the citation text.

Quick checklist (the 30-second test).

Author byline on every substantive page with a link to a real bio.
Bio page with credentials (degree, license, certification, years), named clients, links to verifiable third-party profiles.
Editorial policy page declaring how content is produced.
Disclosure page declaring conflicts of interest, engagement structure, and review policies.
Cited claims link to authoritative third-party sources.
Testimonials attributed by full name (with permission).
Public methodology (rubrics, frameworks, scoring systems) where the work is technical.
Accurate ownership and contact information across every channel (NAP consistency).

Tools and platforms.

Wikidata (free, supports the entity verification side); LinkedIn (free, supports the professional credential side); state professional licensing boards (free public records); SAM.gov for federal contractor verification (free); the SBA SDVOSB veterans certification site (free for verified SDVOSBs); Google Search Console's “manual actions” report (free, surfaces E-E-A-T-related quality issues); the Search Quality Rater Guidelines (free, downloadable from Google — required reading).

Realistic timeline and budget.

E-E-A-T retrofit on an existing site is 20–40 hours billable around $2,000–$4,500 for a small business; the work is mostly content (writing bios, gathering credentials, creating policy pages) plus modest schema additions. The compound effect is large because E-E-A-T currency pays out across SEO, AEO, AIO, and GEO simultaneously. YMYL businesses (medical, legal, financial, insurance, government) need a higher bar and should budget more time on credential verification and source citation.

How E-E-A-T fails.

(1) Anonymous or pen-name bylines — both Google reviewers and LLM trust heads heavily favor named, verifiable authors. (2) Editorial policy pages that exist but make implausible claims (e.g., “every article reviewed by 5 experts” on a solo practice site); humans and LLMs both detect inconsistency and apply a trust penalty. (3) Skipping disclosure entirely on advisory or affiliate-driven content. (4) Treating E-E-A-T as a one-time setup; Google's reviewer guidelines update annually and YMYL bars increase over time.