A public-signal-only AI-visibility audit run against a real, independent SaaS marketing site using ai-visibility-skills — the site is anonymized here since this was an unsolicited technical audit, not a client engagement.

Disclaimer: This audit was run against a real, independent SaaS product using only publicly available data — crawl responses, robots.txt, sitemaps, and page HTML/JSON-LD. No client relationship exists between this audit and the audited site, and no private access (analytics, Search Console, etc.) was used or granted. The site’s name and domain are withheld here by choice: this is offered as proof of what the ai-visibility-skills audit produces, not as commentary on a specific company.

Overview

A public-signal AI-visibility audit run with ai-visibility-skills against a real SaaS marketing + docs site, using the pack’s current V3 six-pillar scoring methodology. The target is a typical SaaS product site: marketing pages, a docs/resources section, pricing, and a FAQ.

Audit result

Overall Readiness Score: 93.5/100 — READY

Pillar (weight)	Score	Status
Discovery (20%)	100/100	Ready
Technical Access (20%)	100/100	Ready
Machine Understanding (20%)	75/100	Partial
Answer Ready (20%)	100/100	Ready
Trust (15%)	90/100	Partial
Agent Ready (5%)	100/100	Ready

What’s working well

robots.txt declares two sitemaps; both return 200 with well-formed XML, and sampled sitemap URLs all resolve.
Docs/resource pages are server-rendered — page text and application/ld+json (Organization, SoftwareApplication, FAQPage, Article) are present in the raw HTML, not hidden behind client-side JS.
Pricing is explicit both in visible text and in Offer.price/priceCurrency schema — no mismatch between what a person sees and what a machine can extract.
A dedicated FAQ page has substantive, non-placeholder answers, directly extractable as citable Q/A pairs.
Action endpoints (/signup, /contact, /support) are all live and return 200, giving agents a machine-discoverable conversion path.

What’s limiting visibility

Organization JSON-LD is missing contactPoint — agents and answer engines rely on this to surface a verified contact route; without it, an agent can’t confidently populate “how to reach this company.” (−15, P0)
Organization JSON-LD is missing sameAs — no LinkedIn/X/GitHub authority links on the organization entity (only the founder’s personal profile has one), which weakens disambiguation confidence for citation. (−10, P0)
No public case studies or testimonials — a SaaS site of this kind commonly benefits from at least one citation-ready case study or client quote; none was found anywhere in the sitemap or homepage content. (−10, P1)
Non-scoring but real: HSTS, X-Content-Type-Options, X-Frame-Options, and CSP headers are all absent (the site is behind Cloudflare, so these are a low-effort edge-config fix). No /llms.txt or other emerging agent-discovery files exist yet — experimental, doesn’t affect the score.

Priority actions, as the audit itself would ticket them

P0 — Add a contactPoint block to the site-wide Organization JSON-LD (low effort, high impact for trust and agent actions).
P0 — Add sameAs authority links to the Organization JSON-LD.
P1 — Publish one short case study or testimonial page, marked with CreativeWork/Review structured data.

Why this is here

This is the same skill pack and scoring rubric used in the Jameel Fragrance case study — shown here anonymized because that engagement was consented and named by the client, while this one is an unsolicited technical audit of a public site with no client relationship. The findings and evidence are real; the identity isn’t published. Full raw audit PDFs (before/after a separate infra pilot, unrelated to the findings above) are retained as evidence and available on request.