SEO Believer: The Build Blueprint

Built to be the most citable SEO and GEO consultancy site in Colorado, and to convert the Boulder business owner who lands on it.

---

1. THE THESIS

Most Colorado SEO sites do one of two things. They run keyword-stuffed agency homepages that promise "proven results" and prove nothing, or they sit on top of third-party listicles that quietly own the search results and feed the AI engines. There is real open ground here, but it's narrower than it looks, and it has a clock on it.

SEO Believer wins on one thing the incumbents structurally cannot copy: rigor. Annette has a Medical Technology and Biochemistry background. She ran SEO and operations for a nonprofit that climbed from a domain rating of 0.8 to 62 with no paid advertising, and adoption demand driven entirely by organic search placed more than 4,000 dogs. The brand is "the SEO consultant in Colorado who runs your campaign like a lab." Falsifiable predictions. Published methodology. Inconvenient findings stated out loud. That positioning survives the day GEO becomes a commodity. "GEO consultant" does not.

GEO is the credential layer, not the headline. It proves Annette is ahead of the curve. But a Boulder dentist does not search for "generative engine optimization." They search for "why isn't my business showing up on Google." So the site speaks the buyer's language at the front and carries the expertise underneath.

The honest "best SEO consultants in Boulder" comparison page is a strong move, but it's a borrowed one and it carries reputational risk, so it comes later, after the expertise pages have earned some signal. The site is the proof of concept. It ranks itself, and it shows the receipts. But the receipts that matter most are client receipts, and those get built in public as they arrive.

---

2. WHAT WE BUILD AT LAUNCH, AND WHAT WAITS

The original plan specced more than fifty URLs. That's a content factory built for a team of fifteen, run by one person who also takes every sales call, runs HARO, posts on LinkedIn, and delivers actual work for a paying client. Sequencing doesn't reduce that load. It only reorders it. So we cut launch to a tight spine and gate everything else behind a real traffic signal.

Rule for every speculative asset: it ships only if the asset before it earned a measurable signal in 30 days. No signal, no expansion. The kill criteria are in Section 7. A stale glossary or an abandoned tracker is worse than one that never existed, because it tells a prospect the consultant quit.

Legend: [EXISTS] already built · [NEW] to build · [REBUILD] exists but needs rework

Launch set (the only things we build first)

/                                  [REBUILD]  Home, rebuilt around buyer outcomes + proof
        /about                             [REBUILD]  About Annette, the E-E-A-T and rigor hub
        /contact                           [EXISTS]   Contact + booking
        /404                               [EXISTS]   404
        
        SERVICES (the conversion spine)
        /services                          [EXISTS]   Services overview
        /services/ai-search-optimization   [NEW]      GEO service
        /services/local-seo                [NEW]      Local SEO service
        /services/google-business-profile  [NEW]      GBP optimization
        /services/technical-seo-audit      [NEW]      Technical SEO + schema
        
        BUYER-LANGUAGE PAGES (these come BEFORE the GEO pillar)
        /seo/boulder                       [NEW]      Boulder, written in owner outcomes
        /seo/denver                        [NEW]      Denver, written in owner outcomes
        
        EXPERTISE / CITATION LAYER
        /guides/generative-engine-optimization  [NEW]  GEO pillar (the credential piece)
        /glossary                          [NEW]      Glossary index
        /glossary/[term] (6 to 8 terms)    [NEW]      Single-definition pages, launch set only
        
        PROOF ENGINE
        /case-studies                      [EXISTS]   Index
        /case-studies/bone-voyage          [EXISTS]   DR 0.8 to 62 (with honest framing, see below)
        /case-studies/rebuilding-seminars  [REBUILD]  The live prediction scoreboard (the key asset)
        
        CONVERSION
        /audit                             [NEW]      Free audit lead magnet (gated, email)
        /review                            [NEW]      Review request landing
        
        TECHNICAL ROOT FILES
        /robots.txt  /sitemap.xml  /llms.txt  /llms-full.txt   [NEW]

Phase 2, only if launch earns a signal

/best-seo-consultants-boulder       Honest comparison listicle (deferred on purpose, see §4)
        /best-seo-consultants-denver        Honest comparison listicle
        /services/apple-maps-optimization   Apple Maps / Business Connect
        /guides/get-cited-by-ai             How to get cited by ChatGPT/Claude/Gemini/Perplexity
        /guides/google-ai-overviews         AI Overviews mechanics
        /guides/google-business-profile-front-range
        /guides/apple-business-connect
        /glossary/[term]                    Expand toward ~20 only if the first 6 to 8 rank
        /seo/longmont /seo/louisville /seo/lafayette /seo/broomfield  Only if a city earns a real page
        /tools/llms-txt-generator           The one safe free tool (static, no API cost)
        /research/front-range-site-teardown A qualitative teardown, NOT a "dataset" (see §4)
        /experiments                        The GEO experiment log (the real authority play, see §4)
        /blog/[post]                        Bi-weekly cadence only if the spine is earning

The Ajijic case study and a self-tracking "AI citation tracker" are cut. Ajijic is off-positioning for a Boulder SEO buyer. A tracker that tracks only ourselves has an audience of one until we have a reputation that makes anyone care, and it adds a weekly maintenance tax for a vanity signal. The client prediction scoreboard does that job better and honestly.

---

3. PAGE-BY-PAGE SPECS (launch set)

/ (Home), rebuilt

Purpose: speak to the buyer first, prove second.

Hero in buyer language: the headline answers "will this person get me more customers," not "what is GEO." Something like: "I help Front Range service businesses show up when their customers are searching, on Google, on Maps, and now inside AI answers." Subhead names the rigor: a scientist's method, published, falsifiable.

Proof block, framed honestly: the Bone Voyage climb stated plainly, with the transfer-of-credibility gap addressed rather than hidden. A nonprofit with adoptable dogs earns links a drain-cleaning company never will, and a sophisticated owner knows that. So the proof point is the method, not the metric: "Here's the system that took a nonprofit from invisible to a domain rating of 62. Here's how the same method applies to a local service business, and here's a client running it live right now." Then a link to the Rebuilding Seminars scoreboard.

Dual CTA: "Book a strategy call" (high intent) and "Get the free audit" (low intent email capture).

Schema: Organization + ProfessionalService / LocalBusiness (full entity from entity/seo-believer-entity.md), WebSite, BreadcrumbList.

/about, rebuilt into the rigor and E-E-A-T hub

Purpose: make scientific rigor the brand, not a footnote.

Content: Annette's credentials up front (Medical Technology and Biochemistry, former Director of Bone Voyage, Adoption.com, the ESL recruiting platform). The methodology stated as a creed: hypotheses, controlled tests, published methodology boxes, willingness to report a result that doesn't flatter the pitch. A disqualification statement, which an agency that needs volume cannot write: "I take a limited number of clients. Here's who I turn away and why." Author archive aggregating her bylined work, with the full sameAs identity web.

Schema: Person (jobTitle, alumniOf, knowsAbout, sameAs).

Service pages (4): /services/*

Purpose: turn intent into qualified calls.

Each (800 to 1,500 words): a plain problem statement, the specific method, what's included, what's not included (rare and trust-building), 5 to 8 FAQs, a reference to relevant proof, and a price band ("$2,000 to $5,000 per month, three-month minimum"). The price band is a qualification tool that filters tire-kickers. It is not a citation hook, and we won't pretend it is.

Schema: Service + Offer (price band) + FAQPage + BreadcrumbList. Mark up only FAQs that are visible on the page.

Buyer-language pages (2): /seo/boulder, /seo/denver

Purpose: meet the owner's actual search vocabulary, and do local SEO without thin pages.

We spec two, not six. Annette can write a genuinely useful, differentiated page about the Boulder and Denver markets. She probably cannot write four more honest ones about Broomfield or Lafayette competitive structure under deadline, and a thin location page hurts more than it helps. The other four wait until there's something real to say.

Each (600 to 1,000 words, unique): written in outcomes the owner cares about (calls, booked jobs, foot traffic), a genuine observation about that market, named neighborhoods and landmarks, distinct title, meta, and H1.

Schema: LocalBusiness / ProfessionalService with areaServed, FAQPage, BreadcrumbList.

/guides/generative-engine-optimization (the credential pillar)

Purpose: own the GEO topic no Colorado incumbent owns, and prove Annette is ahead of the curve. This is the lead expertise asset because it carries no competitor risk, unlike the comparison listicle. Lead with pure expertise, earn signal, then take on the riskier pages.

Sections: a clean one-sentence definition an AI can quote directly, up top. Then SEO versus GEO, how AI engines select sources, the tactics that matter, llms.txt, schema for AI, freshness signals, and a Front Range playbook. Reference the Princeton GEO research and named current practitioners.

Example of the target definition shape (write the real one, don't just instruct it): "Generative engine optimization is the practice of structuring content so AI search tools like ChatGPT, Perplexity, and Google AI Overviews quote it directly when answering a question, rather than ranking it on a page of blue links." That's the unit an AI lifts. Every guide and glossary entry needs one written like this, not a placeholder.

Schema: Article (author by URL, datePublished, dateModified) + FAQPage + BreadcrumbList.

/glossary and /glossary/[term] (6 to 8 terms at launch)

Purpose: capture "what is X" queries with the cheapest-to-earn citations, at a sane scale. We launch 6 to 8, not 25. We expand only if these rank.

Each term page: one crisp single-sentence definition first, then 150 to 300 words of context, then related terms. Bylined.

Schema: DefinedTermSet on the index, DefinedTerm on each. No Article schema on a 200-word definition, Google won't treat it as one and it's the wrong type.

/case-studies/rebuilding-seminars (the live prediction scoreboard): the most important new asset

Purpose: solve the cold-start problem and build the one differentiator no incumbent has the nerve to run. Every other proof point is the site ranking the site, or a nonprofit ranking a nonprofit. This is client proof, in a client vertical, scored in public.

How it works: a dated, public, falsifiable scoreboard. Day-one baseline (current rankings, traffic, leads), Annette's written predictions at 30, 60, and 90 days, then the actual results posted next to the predictions as they land, including the misses. Skin in the game. No Colorado competitor exposes themselves this way, and a busy skeptical owner trusts a public bet far more than a testimonial.

Honesty rail: while results are thin, the page leads with the prediction and the method, not with claimed wins. Once real numbers arrive, they go in exactly as they happened.

Schema: Article with author, datePublished, dateModified. Not Dataset, this is a narrative log, not a published structured dataset.

/audit and /review

/audit: free audit lead magnet gated by email, delivers an automated report. Top-of-funnel capture.

/review: routes happy clients to Google, Yelp, and LinkedIn. Useful only once there are clients to route, so it ships but stays quiet until then.

---

4. THE CITATION ENGINE, REBUILT FOR ONE PERSON

The original named four citation magnets. Two were strong, two were liabilities. Here's the honest version.

1. The GEO pillar plus a small glossary. Single-sentence definitions are the kind of clean, quotable unit an AI lifts directly into an answer, and no Colorado incumbent owns the GEO topic yet. This is the highest citation leverage per hour of work. One caveat we state plainly rather than asserting as fact: the claim that comparison pages are "the single most retrieved shape" is a reasonable hypothesis, not a proven law, and the GEO whitespace is a land-grab with a clock on it, maybe six to twelve months before a national GEO authority adds a Colorado page. We verify both assumptions against live query data before we lock the roadmap.

2. The GEO experiment log (/experiments), the real authority play. This is the asset the positioning actually requires and the original plan was missing. The whole brand is "AI search expert," but the only proof is a traditional, pre-GEO SEO win. The gap closes with live, dated experiments on our own properties showing cause and effect: "I added llms.txt to this page on this date. Here's the citation delta two weeks later." That's harder and more honest than a glossary, and it's what makes Annette the credible Colorado GEO authority rather than someone reciting the Princeton paper. It also doubles as the freshness instrument the cut "tracker" was reaching for.

3. The honest comparison listicles, deferred and de-risked. This is genuinely a strong move, and it's also the riskiest and most copyable one, so it does not lead. Three honest problems the original glossed: it's a known tactic anyone can copy the moment we rank, so it's no moat; ranking yourself is a visible conflict of interest that AI engines increasingly down-weight; and the brief tells Annette to go have coffee with these exact Boulder consultants in her first 30 days, which sits in direct tension with publishing a page that ranks them. So: we build the relationships first, publish the listicle only after the GEO pillar has earned signal, and we lead the page with the method for evaluating consultants, not with our own slot.

4. One qualitative teardown, honestly labeled. The original called an audit of ten sites a "dataset" with Dataset schema and a journalist-citation fantasy. Ten is an anecdote, not a dataset, and no journalist cites n=10. So we either commit to n=50 to 100 (real work, genuinely citable) or we reframe it honestly as "a teardown of ten Front Range sites," drop the Dataset schema, drop the journalist fantasy, and treat it as a qualitative trust-builder. For a solo operator at launch, the honest teardown is the right size.

The free tools mostly get cut. An llms.txt generator is a static, safe, weekend build, so it survives. But gbp-audit and ai-citation-check call paid APIs on every visitor request with no auth, which is unbounded cost and a real liability: someone scripts it and we get a surprise bill. A "free AI-citation check" that conveniently always finds the prospect isn't cited and we are is also a trust trap. Those are cut unless and until they get auth, rate limits, and prominent disclosure.

---

5. SCHEMA AND TECHNICAL PLAN

Declare Organization and Person once with @id and reference them everywhere. Never mark up an FAQ that isn't visible on the page.

Validate every block with the Schema.org validator and Google's Rich Results Test before shipping. One honest note from Ahrefs' own findings: schema is table-stakes machine readability, not the citation lever. The lever is content structure, the definitions, the stats, the clean headers. Schema just annotates it.

llms.txt and llms-full.txt: /llms.txt points to the about page, the GEO service, local SEO, the Bone Voyage case study, and the scoreboard. /llms-full.txt concatenates the full text of the GEO pillar, the scoreboard, and the case study.

robots.txt: allow GPTBot, ClaudeBot, PerplexityBot, Google-Extended, Applebot-Extended, and all reputable crawlers. Never block AI crawlers. The tradeoff is asymmetric in favor of being citable.

E-E-A-T and author signals: every guide, glossary entry, listicle, and teardown carries Annette's named byline with photo, title, and a credentialed bio. Every study and the comparison pages ship with an explicit, falsifiable methodology box and a visible dateModified. Publish at least one inconvenient finding.

Internal linking, because 20+ pages go orphan without it. Hub-and-spoke. The GEO pillar is a hub, each glossary term and supporting guide links up to it and the pillar links down to them. Each service page links to its matching buyer-language page and to the relevant proof. Every page is reachable within two clicks from the home page or a hub. We map this before building, not after.

Freshness, done honestly. "Refresh dateModified monthly" without changing content is the exact fake-freshness signal the brief's own Appendix B warns against, and AI engines increasingly weight real recency over a self-declared date field. So the monthly refresh is a substantive update (new finding, new data, corrected claim) or it doesn't happen. We prune and consolidate underperformers rather than only adding.

Performance: Astro static export. LCP under 2.0s, CLS under 0.05, INP under 200ms, content pages under 500KB, images WebP or AVIF with width and height. Google Search Console (all four property variants), GA4, Bing Webmaster Tools, Plausible as a clean secondary.

---

6. CONVERSION ARCHITECTURE

The buyer is a Boulder or Denver service business, $150K to $5M, an owner who reads but is too busy to execute.

Dual CTA on every page: primary "Book a strategy call," secondary "Get the free audit." Present on guides and glossary too, not just service pages.

Lead capture: the /audit magnet gated by email delivering an automated report, plus the one static free tool (llms-txt-generator) as a no-friction entry, plus a Cal.com or SavvyCal scheduler on /contact and inline on service pages.

Qualification: a short pre-call form (industry, revenue band, current SEO state, budget range, biggest pain point) so Annette takes calls with qualified prospects only, then a five-email welcome sequence: what to expect, the Bone Voyage method in long form, the state of AI search, a Front Range teardown, and a direct CTA.

Proof, honestly transferred. Bone Voyage is real and impressive, but it is not self-evidently transferable to a Boulder dentist, so we never assert it is. We lead with the method and point to the live client scoreboard for proof in a real vertical with real buyer metrics (leads, calls, booked jobs), not DR. Scarcity plus honesty (the disqualification statement, the limited client count) does work an agency can't copy.

Pricing transparency: publish the band openly. It's rare among Colorado competitors and it pre-qualifies hard. Productized tiers (Foundation, Growth, Authority) come later, once the call patterns are clear.

---

7. PRIORITIZED ROADMAP, WITH KILL CRITERIA

Order is by leverage per hour, front-loaded, and every speculative asset has a stop signal. Sequencing orders the work, it does not shrink it, so the launch set is deliberately small and expansion is earned.

Tier 0, foundation (must precede everything, correctly sized and correctly first)

1. entity/seo-believer-entity.md, the canonical NAP and sameAs. Every schema block pulls from it. NAP drift wrecks local SEO before it starts. (This file already exists at C:\Users\annet\ClaudeProjects\SEOBeliever\entity\seo-believer-entity.md. Confirm it's complete before building.)
2. Sitewide Organization / WebSite / Person schema, BreadcrumbList, robots.txt (AI crawlers allowed), sitemap.xml, Search Console and Bing verification, internal-link map.
3. Rebuild /about into the rigor and E-E-A-T hub. Every future asset's author signal depends on it.

Tier 1, lead with expertise and buyer language (no competitor risk)

1. The GEO pillar plus the first 6 to 8 glossary definitions. Pure expertise, uncontested whitespace, cheapest citations to earn, no reputational exposure.
2. The two buyer-language pages (/seo/boulder, /seo/denver), written in owner outcomes.
3. The four service pages with price bands and what's-not-included, and the home rebuild around buyer outcomes plus the proof block.

Tier 2, proof and conversion (turn visibility into leads)

1. The Rebuilding Seminars prediction scoreboard. The single most differentiating asset. Build it early so it accrues a public track record over months.
2. The /audit magnet, scheduler, qualification form, and five-email sequence. Land the leads the moment traffic arrives.

Tier 3, earned expansion (only on a real signal)

1. The /experiments GEO log, started as soon as there's a property to test on. This is what makes the GEO authority claim real.
2. The comparison listicles, only after the GEO pillar has earned signal and after Annette has had coffee with the Boulder and Denver consultants she'd name.
3. Expand the glossary toward ~20, add supporting guides and additional cities, only where each earns a genuinely useful page.

Kill criteria (the part the original never had):

• Glossary: if the first 6 to 8 terms are not indexed and pulling any impressions in Search Console at day 30, stop. Do not grind to 25.
• GEO pillar: if it isn't ranking on page two or better, or getting cited by any AI engine, by day 60, the GEO-as-traffic thesis is weaker than hoped. Keep it as the credential layer, stop pouring hours into the topical web around it.
• Comparison listicle: if it can't be written without ranking ourselves dishonestly, or it threatens a peer relationship that matters more, don't publish it.
• Any free tool or teardown: if it can't ship with bounded cost, clear disclosure, and an honest label, it doesn't ship.
• The whole expansion: if the launch spine isn't producing a single qualified lead in 90 days, the problem is positioning or proof, not page count. Stop building pages and go fix the offer.

The logic: nothing earns until the entity and author foundation exist. Lead with the asset that carries pure expertise and zero competitor risk, prove with a live client bet rather than a borrowed tactic, capture leads the moment traffic arrives, and expand only where a real signal says to. Build the things that get pointed at and the things that get bought, and stop building the moment the data says it isn't working.

---

Build paths: project root C:\Users\annet\ClaudeProjects\SEOBeliever\, canonical entity source at C:\Users\annet\ClaudeProjects\SEOBeliever\entity\seo-believer-entity.md (already present, verify completeness first), site code under C:\Users\annet\ClaudeProjects\SEOBeliever\site\.