How SourceNode manages its own AI fingerprint
Written by
Odin Lund
Category
Playbook
Date
Share this article
Why we publish our own fingerprint
A company that manages how AI describes businesses should be describable itself. When someone asks ChatGPT, Gemini or Google what SourceNode is, the answer should be assembled from sources that agree, and it should be checkable. So we treat sourcenode.co as a client: the same read, the same verified facts, the same structured data, the same earned sources, measured on the same tracker.
The verified facts, stated once here so every other source can match them: SourceNode is an AI search identity company. It runs AI agents that read how AI models describe a business, correct the sources behind those answers, publish the pages customers’ questions need, earn placement in the third-party sources the models cite, and measure the result in share of voice, visibility, citations and customers. It was founded by Odin Lund and Alex Merutka. Its first published engagement took William Mattar Law Offices from 4.6% to 47.2% share of voice in AI answers in three months. Its measurement runs at aisearchmonitor.com.
What we do to our own site
One page per question. The product page answers “what does SourceNode do”; the case study answers “does it work”; the About page answers “who is behind it”; each research post answers one question a customer or a model might ask about AI search. Every page states the same facts in the same words, because a model that finds two versions of a founding story will hedge or pick one.
Structured data in the head of every page: Organization (name, URL, logo, founders, sameAs links) and WebSite. Plain statements of fact near the top of each page, before any argument. Descriptive titles that match the question, not a slogan. And the category terms we want associated with the business, used consistently: AI search, GEO (generative engine optimization), AEO (answer engine optimization), AI visibility, AI fingerprint. We do not stuff them; the GEO study found that keyword stuffing does nothing for generative engines, while clear facts and citations do (Aggarwal et al., KDD 2024).
Earned sources and the read
The sources that describe SourceNode outside our site: the founders’ LinkedIn profiles, our partner Node Media, the research posts here, and the directories and publications where AI search companies get listed. Each is checked against the verified facts above and corrected where it drifts. The list is short today and will grow; the discipline is that nothing on it contradicts the site.
Then the read. We run our own prompt set (“what is SourceNode”, “who manages AI search visibility for businesses”, “alternatives to AI visibility dashboards”, and the questions our customers actually ask) on each model, record how we are described and which source produced each claim, and fix the source when the claim is wrong. The results are ours to publish, and we will, in this series, as they move.
What you can copy, and sources
You do not need our agents to do the first pass. Write down the ten facts you need AI to repeat. Check them across your site, listings, reviews and press, and fix every conflict. Give each customer question one page. Add Organization schema. Then ask the models about yourself, every week, and write down what they said and where it came from. That is the loop. Our agents make it continuous and measured; the discipline is the same by hand.
Sources
SourceNode verified facts, maintained on sourcenode.co, this page being the canonical statement.
Aggarwal et al., “GEO: Generative Engine Optimization,” KDD 2024 (arXiv:2311.09735).
Schema.org Organization and WebSite types.
SourceNode tracking, aisearchmonitor.com.
Share Blog
