· AI-SEO· GEO· Case Study

We Are Our Own Case Study: Documenting Raptor’s AI-SEO Journey

A public log of how we apply our own AI-SEO methodology to raptorai.dev: baseline metrics, weekly experiments, and what actually moves the needle in LLM visibility.

Most agencies sell AI-SEO without ever showing it work. We are taking the opposite approach. Raptor AI is going to apply our own methodology to raptorai.dev in public. Every experiment, every result, every miss, so prospective clients can judge the work before they hire us.

Why we are doing this in public

New agencies hit a chicken-and-egg problem: you need case studies to win clients, but you need clients to build case studies. The substitute proof that actually compounds is to be your own case study. If our methodology works on a four-page agency site with zero domain authority, it will work on yours.

Baseline (Day 0, 2026-05-06)

Honest starting point, recorded so we cannot revise it later:

  • Indexable pages: 4 (/, /seo-for-ai, /privacy, /terms)
  • Google Search Console: not yet verified
  • Bing Webmaster Tools: not yet verified
  • Backlinks (rough): minimal
  • LLM brand mentions: unmeasured, need to establish baseline this week
  • Schema in place: Organization. A prior generic SoftwareApplication declaration wrongly described the agency as a free software product and was removed later. Tool schema now belongs only on actual tools.

The methodology we are testing

Three pillars, applied in order. Each gets its own pillar post over the next four weeks.

  1. Trust signals (E-E-A-T). Named authors with Person schema, real bios, public artifacts. LLMs cite authors more readily than anonymous content.
  2. Topical authority. A pillar/cluster model around AI-SEO and GEO: pillar posts deep, supporting posts narrow, every internal link pointing back to the canonical hub at /seo-for-ai.
  3. AI-structured content. Question-led headings, direct answers in the first paragraph, FAQ blocks with FAQPage schema, Article schema on every post, citation-friendly formatting (lists, definitions, claims with sources).

What we are publishing in the next four weeks

  • How to optimize for ChatGPT search
  • Measuring brand visibility in LLMs (with a free tool)
  • GEO vs SEO: what is actually different
  • Schema patterns LLMs actually consume
  • A teardown of a recognizable brand’s AI visibility

How we will report

This post is updated weekly. New rows added to the metrics table below. No metric removed once published. Wins and losses both stay on the page.

Weekly log

Week 0 (2026-05-06): baseline recorded above. Schema and on-page SEO setup shipped.

Week 4 (2026-06-08), first traffic baseline. Vercel Analytics, trailing 30 days. These are small numbers, and we are publishing them anyway. That is the entire point.

  • Visitors: 53 · Page views: 137 (trailing 30 days)
  • Top pages: / (41), /tools/ai-brand-check (15), /insights (11), /seo-for-ai (10)
  • Where visitors come from: X, Facebook, Instagram, Google, Bing
  • Referrals from AI surfaces (ChatGPT, Perplexity, AI Overviews): zero, so far. For an AEO case study that is not a failure. It is the number we exist to move. Watch this line.
  • Audience skew: Mexico 43%, US 21%, Germany 13%. The bilingual, cross-border footprint we build for.

Week 5 (2026-06-09), Semrush AI-search cleanup. We used the audit warnings as a forcing function, but not as the strategy. The fixes shipped around authority, crawlability, and page usefulness instead of adding empty text.

  • Added /llms.txt as a concise AI-readable map of Raptor AI, core services, and the most important source pages.
  • Expanded /, /insights, and /tools/ai-brand-check so each page explains its role in AI rankings, audits, AI-ready sites, consulting, and training.
  • Replaced generic anchor text with destination-specific links and added /consulting to the sitemap.
  • Added structured data for the home service, insights collection, AI Brand Check tool, global organization, and website entity.

Week 6 (2026-06-15), trailing-30 snapshot. Same measurement window as Week 4, shifted one week forward. Directional growth, not a clean delta: these windows overlap by roughly 23 days.

  • Visitors: 59 · Page views: 219 (trailing 30 days)
  • Top pages: / (48), /tools/ai-brand-check (12), /insights (11), /seo-for-ai (10), /consulting (6)
  • New in top pages: /for/lawyers (3) and /for/real-estate (3), the vertical pages are now indexed and getting visits.
  • Referrers: t.co (5), Instagram (4), WhatsApp (3), Facebook (2), Google (2), Bing (1)
  • Referrals from AI surfaces (ChatGPT, Perplexity, AI Overviews): zero, again. Six weeks in. That line is still the one we exist to move.
  • Audience skew: Mexico 53%, US 12%, Germany 10%.

Week 7 (2026-06-22), trailing-30 snapshot. Visitors held flat week over week. Page views grew on the same visitor base, so sessions went deeper rather than wider. We log the flat number as readily as a rising one.

  • Visitors: 59 · Page views: 259 (trailing 30 days)
  • Top pages: / (54), /insights (12), /tools/ai-brand-check (12), /seo-for-ai (10), /consulting (9), /insights/raptorai-ai-seo-journey (4)
  • New in top pages: /for/dental (3), another vertical page now indexed and getting visits. /consulting climbed from 6 to 9.
  • Referrers: t.co (5), Instagram (4), WhatsApp (3), Facebook (3), Bing (1), Google (1).
  • Referrals from AI surfaces (ChatGPT, Perplexity, AI Overviews): zero, still. Seven weeks in. That line is the one we exist to move, and it has not moved.
  • Audience skew: Mexico 59%, US 17%, Germany 7%.

Week 8 (2026-06-29), trailing-30 snapshot. Best raw-traffic reading yet, and still nothing from AI surfaces. These trailing-30 windows overlap heavily, so read direction, not a clean delta.

  • Visitors: 75 · Page views: 264 (trailing 30 days, up from 59 / 259)
  • Top pages: / (70), /tools/ai-brand-check (12), /consulting (11), /insights (10), /seo-for-ai (10), /for/lawyers (6), /for/dental (4)
  • /for/lawyers doubled (3 to 6) and /consulting nearly doubled (6 to 11). Both vertical pages, lawyers and dental, now draw steady visits.
  • Referrers: m.facebook.com (6), facebook.com (5), l.instagram.com (5), WhatsApp (3), google.com (2), l.threads.com (2), linkedin.com (1), search.brave.com (1). Meta properties drive most referred traffic.
  • Referrals from AI surfaces (ChatGPT, Perplexity, AI Overviews): zero, again. Eight weeks in. The two search referrers we did get, Google and Brave, were classic web search, not AI answers. This is the line the whole experiment exists to move, and it has not moved.
  • Audience skew: Mexico 52%, US 20%, Germany 7%, with Singapore (4%) and the UK (3%) in the mix.

Week 9 (2026-07-06), trailing-30 snapshot. Highest raw readings so far, and the service pages entered the top-page list for the first time. Still nothing from AI surfaces.

  • Visitors: 97 · Page views: 330 (trailing 30 days, up from 75 / 264)
  • Top pages: / (86), /tools/ai-brand-check (14), /consulting (13), /seo-for-ai (12), /insights (10), /for/real-estate (9), /for/lawyers (7), /for/dental (6)
  • All five service pages now draw visits: /services/ai-memory-systems (6), /services/ai-receptionist (6), /services/web-builds (6), /services/ai-visibility (4), /services/qa-testing (4). /for/hospitality, published July 1, was indexed and receiving visits within the week (3).
  • Referrers: l.instagram.com (9), m.facebook.com (6), facebook.com (5), google.com (3), WhatsApp (3), l.threads.com (2), linkedin.com (1), search.brave.com (1). Meta properties still drive most referred traffic.
  • Referrals from AI surfaces (ChatGPT, Perplexity, AI Overviews): zero, again. Nine weeks in. Google and Brave sent classic web search, not AI answers. The line this experiment exists to move has not moved.
  • Audience skew: Mexico 46%, US 28%, Germany 3%. The US share is the highest we have logged.

Week 10 (2026-07-13), trailing-30 snapshot. First decline since this log began, though still the second highest visitor reading on record. Published late, together with Weeks 11 and 12, on 2026-07-27. Still nothing from AI surfaces.

  • Visitors: 89 · Page views: 234 (trailing 30 days, down from 97 / 330)
  • Top pages: / (79), /tools/ai-brand-check (10), /consulting (9), /insights (8), /seo-for-ai (8), /for/real-estate (7), /for/dental (5), /for/lawyers (5), /for/hospitality (3). All five service pages drew 3 to 5 visits each.
  • First appearances in the log: /terms (4), /for/contractors (3), /privacy (2). The contractors page had been live since June 10 and only now cracked the list.
  • Referrers: l.instagram.com (8), m.facebook.com (6), facebook.com (4), google.com (4), WhatsApp (3), l.threads.com (2), linkedin.com (1). Meta properties still lead.
  • Referrals from AI surfaces (ChatGPT, Perplexity, AI Overviews): zero, again. Ten weeks in.
  • Audience skew: Mexico 42%, US 31%, with Germany, Norway, Sweden, and Singapore at 3 visitors each. Fifth straight window of US share growth.

Week 11 (2026-07-20), trailing-30 snapshot. Visitors held but sessions thinned: the lowest page-view reading since this log began. Google reached the top of the referrer list for the first time, tied with Instagram. Published late. Still nothing from AI surfaces.

  • Visitors: 92 · Page views: 177 (trailing 30 days; visitors up from 89, views down from 234)
  • Top pages: / (77), /for/real-estate (7, the number-two page for the first time), /consulting (6), /services/ai-receptionist (6), /services/qa-testing (6), /tools/ai-brand-check (6), /for/lawyers (5), /insights (5), /seo-for-ai (5)
  • Referrers: google.com (7) and l.instagram.com (7) tied at the top, then m.facebook.com (6), facebook.com (5), l.threads.com (2), WhatsApp (1), linkedin.com (1).
  • Referrals from AI surfaces (ChatGPT, Perplexity, AI Overviews): zero, again. Eleven weeks in. Google arrived as classic web search, not AI answers.
  • Audience skew: US 35%, Mexico 29%, Germany 4%. The first window where the US leads Mexico.

Week 12 (2026-07-27), trailing-30 snapshot. Visitors held near ninety while page views stay well below the Week 9 peak. Google is now the top referrer on its own for the first time in this log. Still nothing from AI surfaces.

  • Visitors: 91 · Page views: 190 (trailing 30 days; 92 / 177 last week)
  • Top pages: / (72), /tools/ai-brand-check (9), /for/real-estate (8), /services/ai-receptionist (8), /services/qa-testing (8), /consulting (7), /seo-for-ai (7), /services/ai-memory-systems (7), /services/web-builds (7), /insights (6)
  • Vertical pages hold a steady floor: /for/contractors (5), /for/dental (5), /for/lawyers (5), /for/hospitality (3). This page itself drew 2.
  • Referrers: google.com (8), facebook.com (7), l.instagram.com (5), m.facebook.com (3), WhatsApp (1), linkedin.com (1). Google alone at the top; Meta properties no longer lead the list.
  • Referrals from AI surfaces (ChatGPT, Perplexity, AI Overviews): zero, again. Twelve weeks in. Google is sending classic web search, not AI answers. The line this experiment exists to move has not moved.
  • Audience skew: US 34%, Mexico 23%, Germany 7%, Sweden 7%. The US now clearly leads Mexico.

No metric removed once published. AI-referral line remains zero, the number this whole experiment is built to move.

Want this run on your brand?

We are taking on a small number of brands as the first paid case studies for the same methodology. Contact Raptor AI about an AI visibility audit and tell us what you sell. If you want to see how AI describes your brand right now, run our free check first.