27 October 2026
Discovery
Discovery is where everything at SaidTrue begins, and it begins with one question: what do AI systems already say about your business?
Discovery
How SaidTrue approaches Discovery
Discovery is where everything at SaidTrue begins, and it begins with one question: what do AI systems already say about your business? Not what your website says, and not what your rankings say — what ChatGPT, Gemini, Perplexity and Claude tell someone who asks. We look at that first because it is the version of a business that people increasingly meet before they reach the site at all.
We start with the questions customers ask, not the pages
We do not open with keywords or with an inventory of pages. We start from the questions customers already ask — who to use, whether a business is reputable, what it is like to deal with. Those become the prompts we put to each engine. The order matters: the customer's question comes first and the website comes second. An AI answer is assembled from what is said about a business across the web, so the useful unit of discovery is the question, not the page.
Then we record what was said and what was sourced
For each question we capture three things: what was said, what was sourced, and where the truth diverges from the business's own account of itself. The sourcing is where discovery earns its keep — an answer that recommends a business, or misreads it, is drawing on something, and seeing that something shows which parts of a public footprint are doing the talking. We then score the answers, and our method for that is set out in How SaidTrue Scores AI Answers. Engines need not agree with each other, and they need not agree with themselves from one scan to the next, which is why Discovery is not a one-off exercise; monthly AI monitoring keeps the picture current.
Why we work in this order
Atlan's documentation on Discovery puts the underlying problem plainly: "Most data teams don't have a search problem. They have a context problem: assets exist, but nobody knows what they mean, who owns them, or whether they can be trusted" (Discovery | Atlan Documentation, Atlan). Reputation in AI search has the same shape. The material exists; the open question is what the engines make of it. Discovery answers that before anything else is attempted, because a plain account of what was actually said is the only honest place to start.
What Discovery leaves alone
SaidTrue gives visibility reports. Discovery does not rewrite a website or restructure a page; it produces a reading of what AI systems say about a business, where that came from, and how it scores — something a business can act on with a clear view of what it is acting against.
References
- Discovery | Atlan Documentation — docs.atlan.com
Sources and supporting material
Further reading: