01 · Measure
Start with the real questions.
Define the buyer questions before running them, then record what each engine says and which sources it uses.
Patrick Robinson Consulting
I study where businesses are cited, compare named competitors under the same rule, and publish the evidence, tools and practical methods that follow. One operator, clear measurement and no promised placement.
The method
This site is the public side of my AI-search work. It holds the measurements, explanations and tools. I use the same system on selected, tightly scoped commissioned projects when the question and the useful outcome are clear.
01 · Measure
Define the buyer questions before running them, then record what each engine says and which sources it uses.
02 · Compare
Compare named businesses at query level. Separate a brand mention from a cited source, and keep observation apart from inference.
03 · Act
Turn the findings into clear content, better structure or a more useful source surface. Re-measure when the scope calls for it.
The evidence
The strongest evidence here is the published research: 649 classified citation appearances, frozen method, downloadable dataset.
Original research
I analysed 649 classified citation appearances across seven AEO questions, four engines and repeated runs. Vendor-owned pages dominated; independent comparisons were rare. Full dataset and methodology downloadable.
Read the report →Essays
Building-in-public writing on how AI search decides who to cite, including the one where I measured which AI SDR tools the answer engines actually recommend, and who has the citation gap.
Read the essays →Questions
It means using a fixed set of real buyer questions to see which businesses and sources appear across major AI answers, then using those findings to decide what content, structure or off-site evidence deserves attention. It is measurement and editorial judgement, not a claim that any engine will cite a particular page.
The original research, essays, downloadable datasets and AI Visibility Check are public. Together they show how I define buyer questions, compare answer engines, classify cited sources and separate measured findings from inference.
I am not currently selling a standard Citation Engine audit or subscription. I consider selected commissioned or partner work when there is a specific business, a defined question and a bounded scope. The contact form is the place to set that out.
The first report classified 649 citation appearances across seven AEO questions, four AI engines and repeated runs. Vendor-owned pages accounted for 61.9% of those appearances, while independent comparisons accounted for 0.6%. That describes the observed source mix; it does not establish that a mention or link caused an engine to cite a source.
No. AI engines control their own crawling, grounding and answers, and those answers vary. Measurement can establish the observed baseline and show what changes later, but it cannot promise a citation rate, ranking, traffic level, lead volume or revenue result.
Patrick Robinson has run an independent digital consulting practice since 2014, with SEO as one of its main pillars. His current work focuses on measuring AI-search visibility, publishing original research and turning evidence into clear digital content and tools.
Free · 60 seconds
Before you spend a penny on visibility, check whether the engines can fetch your pages at all. CDN settings hide more sites from AI search than bad content does.
Get in touch
I am not currently selling a standard audit or subscription. The form is open for specific commissioned work, partnerships, press and peer questions. Set out what you are trying to measure or build, and I will reply in writing if it fits.