Most agencies publish case studies about their best clients. We're going to do something more useful: publish the audit of ourselves — including the part where we fail.
On August 5, 2026, we ran getseenco.com through the same AI Visibility Audit we sell. Same engine, same scoring, same live citation testing. No edits, no cherry-picking. Here is exactly what it found.
The scores
- GEO (AI visibility): 47/100. Partial visibility — most AI engines could find us but had little reason to cite us.
- AEO (answer engine readiness): 42/100. Our content wasn't structured to be quoted as the answer.
- SEO (traditional search health): 84/100. Strong foundation — clean indexability, healthy robots.txt with every major AI crawler explicitly allowed, sitemap, semantic HTML.
- Core Web Vitals (Google Lighthouse, mobile): 92/100 performance, LCP 3.2s, CLS 0.
That spread is the story. Technically, the site was nearly flawless. Commercially, it was invisible where it matters most.
The live test: 0 for 3
The most important part of our audit doesn't infer anything — it asks an AI assistant, with live web search, the actual questions our buyers ask, and then checks the cited sources deterministically. We ran three:
- "How do I get my business to show up in ChatGPT answers?" — we were not cited. The AI cited eseospace.com, HubSpot, and other agencies instead.
- "What is GEO and how is it different from SEO?" — our literal core offering — not cited. Contentful, Jasper, and within.co took the citations.
- "Why isn't my website showing up in AI search results?" — not cited.
A GEO agency that AI engines don't cite for "what is GEO" has no business hiding that finding. It's also the single most common pattern we see in client audits: a technically healthy site that gives AI engines nothing citable.
The diagnosis
The audit's findings converged on one root cause with three symptoms:
- No named human authorship. Every author signal on the site read "GET SEEN" — a brand, not a person. AI engines and Google's quality systems weight named expertise (E-E-A-T) heavily, and a company name carries none of it. This despite our founders having genuinely citable backgrounds — two decades of technology leadership, a NASDAQ-listed CTO role, a Microsoft Partner of the Year consultancy — that the site never surfaced in machine-readable form.
- No visible freshness. No publish or update dates on core pages. For engines that weight recency — Google AI Overviews and Perplexity in particular — undated content is a reason to prefer someone else's.
- No first-hand evidence. No case studies, no published data, no outcomes an AI could quote. Everything citable that we knew, we hadn't written down.
What we shipped the same day
- Named authorship, everywhere: every blog post now carries a real byline with Person schema (not Organization), linked to the author's credentials and LinkedIn profile.
- datePublished and dateModified on all content — in the page and in structured data — with a rule we'd give any client: only set dateModified when you actually modified something. Fake freshness is worse than no freshness.
- This case study — first-hand data, specific numbers, a named author, and a falsifiable claim. Exactly the content type every AI engine's citation logic favors.
- A platform-by-platform guide to citation mechanics (see our companion post), because generic "optimize for AI" advice is exactly the commodity content that never gets cited.
What happens next
We'll re-run the audit on a schedule and publish the deltas here — score by score, question by question, including the live citation tests. If the number doesn't move, you'll see that too. That's the standard we hold our clients' work to, so it's the standard we'll hold ours to.
If you want to know your own number: the same audit that produced every figure in this post is what we run for clients. The free version takes about two minutes.