Client Zero, November 2026: Where We Started and Where We Are

At a Glance

Every visibility figure here is a dated single run, not a guarantee. AI answers change from day to day.

Why We Test on Ourselves

Before selling answer engine optimisation (AEO) to anyone, we apply the same method to our own site and publish the result, including the parts that did not work. AEO is the work of making a business easy for AI answer tools to find, understand and name. The June case study set the starting point. The August update showed the structural score reaching its ceiling while AI visibility moved only a little. This update puts all the dated readings on one timeline, and adds the first runs where we asked AI tools the questions ourselves.

Two questions run through the whole page, and they need different measures:

Can AI read the site? We control this. It is the structural score from our free AEO Score Calculator. Does AI name us? AI tools control this. We measure it by asking AI tools real questions and counting the answers. A perfect score on the first does not make the second happen, which is the main lesson of the last six months.

Where We Started and Where We Are

Graph 1: Can AI Read the Site?

Graph 2: Does AI Name Us?

The Full Timeline

1. 20 May 2026Client Zero beginsDay-one structural score 19 of 100. A new brand with no history in any AI tool.
2. 13 Jun 2026First AI perception reading: 36, 30 and 41From a free grader. We later found it allows one run per brand and re-serves it, and our exports of 13 June, 20 June and 3 August were identical. So it is one reading taken once, not a trend.
3. 20 Jun 2026Structural score 91The site is now readable by machines. We said then that this did not show what AI knew.
4. 24 Jul 2026Continuous monitoring starts: 9.4%A baseline from only 14 sampled answers.
5. 2 Aug 2026Structural score 100; monitored visibility 16.0%Average citation position improved from 7.0 to 1.5. We also found a similarly named organisation in the United Kingdom was distorting some early results, which made the 9.4% baseline less reliable.
6. 22 Oct 2026First readings from outside tools, and our own first reading for the periodOur plain question: named by 1 of 5 tools. A ChatGPT-only outside report: named in 3 of 10 prompts, with no negative sentiment.
7. 6 Nov 2026A wider outside test: 19 of 27 answersRank Prompt ran 9 buyer prompts across Google AI Overviews and AI Mode, ChatGPT and Claude: named in 89% of audit searches, 33% of consultant searches, and the most-mentioned brand at 9.84% share of voice.
8. 7 Nov 2026An outside audit of one pageOur About page scored well on readability for AI but only 6 of 100 on answer targeting: few question headings and no FAQPage markup.
9. 8 Nov 2026Our own runs, tool by toolPlain question: named by 3 of 5. Solo-founder question: named by 4 of 4 clean runs, first on three of them. The same day: structural score 100, PageSpeed 97 mobile and 100 desktop.

What the Tools Said on 8 November

We asked each tool two questions in a new private or temporary chat, with one counted run per tool, and saved every answer in full. We use the same method each time so the readings can be compared. On 22 October we ran five tools and three questions, and the plain question is worded identically on both dates. The solo-founder question was reworded between them: on 22 October 3 of 5 tools named us (ChatGPT and Perplexity listed us first), and on 8 November 4 of 4 clean runs did. Client Zero is ongoing, and each new reading will be added. The plain question, “Who is the best AEO consultancy in Australia?”, was named by Grok, Perplexity and DeepSeek, and not by ChatGPT or Gemini. We left out Claude and Google on that question because their answers used a signed-in profile, so they were not clean tests.

The second question describes what we do: “Who is the best AEO and GEO consultant for solo founders in Australia?” Every clean run named us. ChatGPT, Grok and DeepSeek listed us first, and Perplexity named us as the closest fit. Google AI Mode listed us third, but its sign-in state was unclear, so we show it separately. Gemini’s two runs disagreed and neither was saved, so it is not counted. The pattern is plain: AI names us when the question describes what we do, and less when it is the general question.

The cropped screenshots, tool by tool, are on the November dashboard. They show only our own entry.

The Full Results, Metric by Metric

Every result from the November 2026 dashboard is set out below in plain text. Our own runs, the three outside tests and our own site checks are kept separate and labelled. All visibility figures are single runs on the dates shown, so they are a snapshot and not a trend.

Our own runs, 8 November

3 of 5 named on the plain question.

On the plain question, “Who is the best AEO consultancy in Australia?”, three of five tools named us, up from one of five on 22 October. When the question describes what we do, “Who is the best AEO and GEO consultant for solo founders in Australia?”, every tool with a clean test named us, and three put us first. ChatGPT and Gemini did not name us on the plain question.

These are single runs, one per tool on one day, so they show a snapshot, not a trend. The same question can return a different answer on another day: Gemini’s two solo-founder runs disagreed and neither was saved, so it is left out. The solo-founder wording matches our own offer, so it is an easier test than the plain question. Several tools said their picks rest largely on our own published pages and are not an independent ranking, and Perplexity weighed our own monitoring (named in none of six unbranded prompts) against us. Screenshots are cropped to our own entry; other firms are removed, not blurred. Outside tools’ results sit on the Independent tab and are never blended with these.

Independent analysis: three outside reports

70% named in outside tests.

Everything else on this dashboard is our own measurement. This tab is different: outside tools wrote their own prompts, chose their own platforms and reported the result. We have not re-run them. In the main test AI named us in 19 of 27 answers but linked to our own site in only 4 of 165 sources. The ChatGPT-only report is lower because its prompts are broader: named in 3 of 10, and always on branded or specialist-consultant questions.

These are three single reports from outside tools, so they show a pattern, not a trend. The pattern is consistent: we are named when buyers ask for an audit or use our name, we are named far less when they ask for a consultant or a general how-to, and AI rarely links to our own site. The page audit points at the same cause from the other side: the About page is easy to read, but it is not yet written as direct answers to questions. It covers one page, so we will audit more pages. These are the tools’ results as reported; we have not re-run the prompts. One report also flagged that two answers described our audit as including ongoing tracking, which it does not.

AEO Score: can AI read the site

100 /100 structural.

This is the part we control: can AI read the site. It scored 100 of 100 (A+). A perfect structural score does not mean AI recommends us, which is why the other tabs measure that separately.

Structure is solved: 100 of 100. The other tabs show where the work is now, which is being named and cited.

Brand Recognition: does AI name us

3 of 5 named on the plain question.

Named by three of five tools on the plain question we ran ourselves (up from one of five on 22 October), and by every clean run on the solo-founder question. Outside tests found us named in 19 of 27 answers, but only 3 of 10 prompts in the ChatGPT-only report.

Recognition depends on how the question is phrased. We are named when the question describes what we do or names us, and far less on the generic consultant question. All figures are single runs, so treat them as a snapshot.

Share of Voice: how often we are named against others

9.8% of brand mentions.

We were the most-mentioned brand: 19 mentions, a 9.8% share. The next brand was 3.6%. The market is wide open, so even the leader holds under a tenth of the mentions.

We lead on share of voice in this test, but it is one run by an outside tool on one date, and the prompts were ones we chose to track. It is a snapshot, not a ranking of the market.

Brand Sentiment: how AI described us

5 of 5 answers not negative.

None of the five full answers spoke badly of us. Four read as a straightforward recommendation or listing. One, DeepSeek, named us but said the conclusion rests heavily on our own description and should be checked independently.

The tone was favourable, but five answers read by us is thin evidence. We treat sentiment as unmeasured until a scored audit is run.

Presence Quality: where in the answer we appear

5 of 27 answers in the top three.

Of 27 answers, we were first once and in the top three five times. Most of the time we were named mid-list, positions 4 to 10. In our own solo-founder runs we were first on three of four clean tools.

We are usually in the answer, but seldom at the top of it in the outside test. Position is where the next gain is. Both sets are single runs.

GEO Visibility: engine by engine

3 of 3 tested engines name us.

In the outside test all three engines named us, Google AI Mode most often at 89%, Claude least at 56%. In our own runs, Grok, Perplexity and DeepSeek named us on both questions. Local AI search (maps and nearby results) is not measured here.

We appear across engines, strongest in Google’s and weakest in Claude’s. Different tools answered differently on different days, so the ranking of engines is indicative only.

Analysis Summary: strengths and gaps

↗ up since 22 October.

The site is built correctly and we are named more often than we were on 22 October. The weak points are the generic consultant question, how often AI cites our own pages, and answer-ready headings. Every figure is a single run, so this is a first reading.

Direction is positive, but the evidence is single runs by us and by outside tools. The honest read: structure is done, recognition is improving, and citation of our own pages is the main thing to fix.

Narrative Themes: what AI says about us

5 of 5 quote the $1,500 flat fee.

Every answer gave the flat $1,500 Founder Audit price. Most described it as one-off with no retainer. Some placed us in Sydney, and two mentioned the do-it-yourself fix list. One said the case rests on our own description.

AI describes the offer consistently and accurately. The one caveat worth acting on is that independent sources confirming our own description would give AI more reason to trust it.

Source Analysis: what AI cites

2.4% of citations are our site.

AI cited our own site 4 times in 165 (2.4%), from the homepage and the Founder Audit page. Most citations went to Google itself (17.6%) and to competitor pages. Our buyer’s guide page was not cited in this test.

AI names us more than it cites us. Getting our own pages, such as the buyer’s guide and case studies, cited is the clearest source gap. Single run, one outside tool.

Site Performance: homepage speed

97 /100 mobile performance.

The homepage scores 100 on desktop and 97 on mobile for performance, and 100 for accessibility, best practices and SEO on both. Desktop paints in 0.3 seconds and mobile in about 1.1 seconds. One mobile measure sits just in the amber band: layout shift (0.111, against a 0.1 target). Total blocking time is 40 ms on mobile and 50 ms on desktop.

Speed is not what holds us back. Desktop scores 100 in all four categories and mobile performance is 97. The mobile layout shift and render-blocking requests are the only open work. One lab run per device, so scores can move a few points between runs.

Technical Foundation: nine technical areas

87 /100 average of nine areas.

Nine technical areas averaged 87 out of 100. Redirects and HTTPS scored 100. Structured data scored lowest at 71: the schema is present and rich, but the audit lists missing recommended properties. A separate area, answer targeting, scored 6 and is covered under Independent analysis.

The technical base is sound. The schema warnings are worth working through, but they are one tool’s list for one page, not a sitewide audit.

What Is Still Missing

Being cited. In the outside test AI cited our own site 4 times in 165 sources, 2.4%. Most citations went to Google itself and to other firms’ pages. AI names us more than it links to us.

The consultant question. We were named in only 33% of consultant-style answers in the outside test, against 89% for audit-style questions.

Answer-ready pages. The outside page audit found few question headings and no FAQPage markup on the About page. This is fixable and is next on the list.

Not yet measured. Scored brand sentiment, local AI search, and real-visitor speed data. The dashboard marks each of these as not measured.

How Far to Trust These Numbers

We published the limits with the figures, because an AEO consultancy that overstates its own results is not worth hiring.

What This Means for Your Business

What We Do Next

We will freeze the prompts and repeat the same runs on a schedule, so the next update can show a trend and not another snapshot. We will add clean signed-out runs for Claude and Google, a saved Gemini run, and answer-ready headings and FAQ markup on our key pages. We will also use several outside audits, not one, so our own results are checked.