GEO Audit Case Study: kanon.agency 12 → 58.7 in 5 Days
How Kanon Agency improved its own GEO score from 12 to 58.7 in five days by fixing AI crawler access, structured data, technical baseline, and entity signals.
When we built the GEO-AAFO audit tool, we used kanon.agency as our first test subject. The initial score was 12 out of 100.
Five days later, after systematically working through the remediation roadmap the tool generated, the score reached 58.7.
This is a record of what we found, what we fixed, and what moved the needle — and what didn’t, yet.
Starting Point: GEO Score 12 / 100
The baseline audit ran 30 prompts across ChatGPT, Perplexity, Claude, and Gemini on August 6, 2026. The results were clear:
- Perplexity: 10/30 cited (33%) — all on branded T1 prompts only
- Claude: 0/30 cited (0%)
- Gemini: 0/30 cited (0%)
- ChatGPT: 0/30 cited (0%)
Every category and intent prompt — “best GEO agency,” “AI search optimization agencies,” “I need an audit of my AI search presence” — returned 0 citations across all four platforms.
The tool’s component breakdown showed where the score was lost:
| Component | Score | Max | Problem |
|---|---|---|---|
| AI Access | 4 | 10 | Missing crawler directives in robots.txt, no llms.txt |
| Structured Data | 8 | 25 | Organization schema present but incomplete sameAs |
| Technical | 9 | 15 | No issues |
| Content Quality | 6 | 25 | No buying-journey content, no case studies |
| Entity & Authority | 3 | 10 | No Wikidata entry, severe name ambiguity |
| Freshness | 2 | 15 | Single-page site, no dated content |
Total: 12 / 100
The Core Problem: Name Ambiguity
Before fixing any technical signal, we needed to solve a structural problem.
When any of the four AI platforms received a branded query — “What does Kanon Agency do?” — they returned results for Kanon Gaming, KANON Digital Marketing (kanontheagency.com), Kanon Services, Kanon Contractors, and several unrelated international businesses.
kanon.agency didn’t surface at all on Claude, Gemini, or ChatGPT. Perplexity found it only because it aggressively fetches live pages — it could read the site directly. The other three platforms rely on what’s already indexed and corroborated externally, and for kanon.agency that external footprint was empty.
There was no Wikidata entity to anchor “Kanon Agency” = kanon.agency. No third-party directory listing. No press mention. No external corroboration of any kind.
What We Fixed
Day 1: Wikidata Entity
Created wikidata.org/wiki/Q140906187:
- Label:
Kanon Agency - Description:
American AI visibility agency specializing in generative engine optimization services - P31 (instance of): marketing agency
- P856 (official website): https://kanon.agency
- P17 (country): United States
- P571 (inception): 2026
- P4264 (LinkedIn company ID): kanonagency
Then added the Wikidata URL to the Organization schema sameAs on the homepage:
"sameAs": [
"https://www.wikidata.org/wiki/Q140906187",
"https://www.linkedin.com/company/kanonagency",
"https://www.crunchbase.com/organization/kanon-agency"
]
This is the single signal AI systems use to disambiguate an entity. Without it, “Kanon Agency” was permanently ambiguous. With it, there’s a ground-truth reference that connects the name to the domain.
Time: ~1 hour. Expected impact: entity disambiguation across all platforms within 3–7 days of indexing.
Day 1: robots.txt and llms.txt
Added explicit AI crawler permissions to robots.txt:
User-agent: GPTBot
Allow: /
User-agent: OAI-SearchBot
Allow: /
User-agent: ClaudeBot
Allow: /
User-agent: Claude-SearchBot
Allow: /
User-agent: PerplexityBot
Allow: /
User-agent: Google-Extended
Allow: /
Created /llms.txt:
# Kanon Agency
> American AI visibility agency offering GEO and AAFO audits for B2B companies
## Key pages
- /: Homepage — GEO + AAFO audit services, pricing overview, contact
- /articles: Articles on GEO, AI visibility, and B2B lead generation
- /articles/what-is-geo-and-why-b2b-companies-need-it-now: Introduction to GEO
Time: 30 minutes. Impact: AI Access component 4 → 10 (+6 pts).
Day 2: LinkedIn Company Page
Completed the LinkedIn company page at linkedin.com/company/kanonagency:
- Logo and cover image uploaded
- About section expanded with specific ICP: “B2B SaaS and professional service companies”
- Specialties: Generative Engine Optimization, AI Agent Friendly Optimization, GEO, AAFO, AI Visibility, B2B SaaS Marketing
- Location: United States
- First post published
LinkedIn matters because it’s now the #1 most-cited domain for professional queries across AI search platforms. It also serves as one of the two sameAs anchors in the Organization schema alongside Wikidata.
Time: 1 hour. Impact: External Authority component, ChatGPT and Perplexity corroboration.
Day 3–5: Buying-Journey Content
The biggest gap in the original score was content. The site was a single scrolling homepage with no standalone pages, no dated articles, and no content addressing any of the 10 buying-journey stages.
The tool’s Coverage Map showed seven of ten stages red-flagged at 0% mention rate:
- Problem Recognition: 0%
- Available Approaches: 0%
- Provider Selection: ~17%
- Cost & Pricing: 0%
- Risk: 0%
- Industry Fit: 0%
- Implementation: 0%
We built the blog section and published two articles targeting the highest-priority gap stages:
Article: “What is GEO and Why B2B Companies Need It Now”
Targets: Problem Recognition, Available Approaches
URL: /articles/what-is-geo-and-why-b2b-companies-need-it-now
Article: “GEO Audit Case Study: kanon.agency 12 → 58.7 in 5 Days” (this article)
Targets: Expected Outcomes, Risk, Provider Selection
URL: /articles/geo-audit-case-study-kanon-agency-12-to-58
Each article page has:
- Its own canonical URL
- Unique title and meta description
- Article JSON-LD schema with
datePublishedandauthor - FAQPage schema for Q&A sections
- Direct answer-first structure for AI extractability
Time: ~6 hours. Impact: Content Quality component, Freshness component, buying-journey coverage.
Score After 5 Days: 58.7 / 100
| Component | Before | After | Change |
|---|---|---|---|
| AI Access | 4 | 10 | +6 |
| Structured Data | 8 | 14 | +6 |
| Technical | 9 | 11 | +2 |
| Content Quality | 6 | 14 | +8 |
| Entity & Authority | 3 | 7 | +4 |
| Freshness | 2 | 10 | +8 |
| Total | 12 | 58.7 | +46.7 |
The largest gains came from Content Quality (+8) and Freshness (+8) — adding standalone article pages with dated content and proper Article schema. AI Access (+6) from the crawler directives and llms.txt. Structured Data (+6) from the completed sameAs array.
What Didn’t Move Yet
T2/T3 citation rate is still 0% across all platforms. Category and intent prompts — “best GEO agency,” “AI search optimization agencies,” “what is GEO?” — still return no citations for Kanon.
This is expected. Getting into category queries requires:
- External corroboration — being named in third-party roundups, directories, and publications. We have not yet placed in any agency roundup lists. That takes 4–6 weeks of outreach.
- Indexing time — the new article pages were published today. AI platforms take 1–4 weeks to index and incorporate new content.
- Wikidata propagation — the entity was created two days ago. Google, ChatGPT, and Claude need time to incorporate it.
The GEO score measures what you’ve built. The citation rate measures what AI systems have discovered. There’s always a lag between the two.
What’s Next
The roadmap for reaching 75+ follows Phase C and D of the agency plan:
- Press placement — one mention in an indexed GEO/AI visibility publication. This is the External Authority gap (currently 3/15).
- Comparison pages — a “Kanon Agency vs. [competitor]” page for the top T2 Comparisons-stage prompts.
- More buying-journey articles — Cost & Pricing, Risk, and Industry Fit stages are still at 0%.
- Agency roundup pitches — outreach to Nutshell, Digital Elevator, Minuttia, and similar lists that rank for “best GEO agencies.”
We’ll publish a follow-up once the T2/T3 citation rate moves.
The Takeaway
A GEO score of 12 doesn’t mean the business is invisible because it published bad content. It usually means AI systems can’t find it, can’t identify it unambiguously, or can’t find enough external evidence to justify recommending it.
In kanon.agency’s case, all three were true. The fixes were straightforward — they just hadn’t been done. Wikidata, crawler access, content structure, and a LinkedIn page. Under 10 hours of work. +46.7 points.
The harder work — press, roundup placements, and building the external information ecosystem — takes months, not days. But the foundation has to come first.
Want to see where your site scores? Book a GEO + AAFO audit →