AI Visibility Work: Where I Stand in November 2026

AI visibility is whether tools such as ChatGPT, Gemini and Google AI Mode mention and cite your brand when someone asks a question. I have been measuring it for client sites since spring 2026. This is where that work stands today, including the first audit of my own site, where Ktdigital was mentioned in 3 of 30 test runs.

What I mean by AEO, GEO and AI visibility

AEO (Answer Engine Optimization) is making your content the answer that AI tools give. GEO (Generative Engine Optimization) is making your brand a source those tools cite. SEO still matters, because ranking and indexing feed both. The difference is the unit of measurement: for AI visibility I track mentions, citations and the wording used about a brand for a fixed set of questions, not only keyword rankings.

Evidence from client work (anonymised)

Client names are withheld. Two results I can share, both dated:

• 10,700+ AI-feature impressions in 15 weeks. Source: Google Search Console, one Vietnamese events and entertainment website. These are impressions from AI Overviews and AI Mode, not clicks.
• #1 in Google AI Mode for one tracked commercial query. Baseline check on 12 November 2026, events niche. One query, one date: a starting signal, not a benchmark. In the same check the site was not mentioned for several other tracked queries, and that gap is now the work plan.

Measuring my own site first

A consultant who sells AI visibility audits should audit themselves. On 13 November 2026 I ran a fixed bank of 10 prompts, in Vietnamese and English, on Google AI Mode, ChatGPT and Gemini: 30 runs in total.

1. Result: Ktdigital was mentioned in 3 of 30 runs (10%): once in each platform, for different prompts.
2. Best signal: Google AI Mode named Ktdigital first for an English query about GEO and AEO consultants for small businesses in Vietnam, and cited the site.
3. Where it is absent: general questions such as which freelance SEO services are reliable in Vietnam, or how to measure AI visibility.

What these numbers do not prove

• Each prompt ran once. AI answers change from run to run.
• I was logged in to my own accounts, so personalisation can inflate results. One ChatGPT answer included wording that suggested it.
• Perplexity was not measured: it did not answer without an account. I skipped it for now.
• Referral data is small: GA4 shows 16 sessions from chatgpt.com between 14 September and 11 October 2026.

What I changed on the site

• Added AEO to the homepage title, H1 and structured data, so the site describes what I actually do. See About Aiden Hoang Nguyen.
• Published an “AI Visibility Results” section with the dated figures above.
• Launched a free AI Visibility audit with a short request form.
• Kept the paid GEO audit and GEO services pages as the next step after the free audit.

What happens next

1. Re-measure in mid-December 2026: the same 10 prompts, same platforms, so the comparison is fair.
2. Extend the bank: add nine prompts that come from real client enquiries.
3. Run a logged-out pass: reduce personalisation in the results.
4. Fix, then measure again: pages that should be answers, schema, and mentions outside the site, in that order.