About CiteSite

Search is splitting in two. People still type queries into Google, and they increasingly ask ChatGPT, Gemini or Perplexity a question and act on the answer. Those answers name brands, and the brands they name are not always the ones ranking first in Google.

CiteSite measures both halves for one website: where it ranks for the queries that matter, and whether AI assistants mention, recommend and cite it when asked the questions its customers ask. It then ranks the gaps by what would move most, and drafts the content to close them.

SaaS founders, in-house marketers, agencies and SEO professionals — anyone accountable for whether a brand gets found, who has noticed that a rank tracker no longer covers the question.

Two scores, weighted equally, and a blend of the pair called the CiteSite Score. Equal weighting is a claim in itself: that neither half is a footnote to the other.

• Search visibility — positions for the queries a site should be winning, weighted by how much attention each position actually receives.
• AI visibility — across a set of prompts written for the brand’s market, whether an assistant mentions it, recommends it, cites its pages, and where it lands relative to competitors.

Scores run 0–100 and are banded rather than read to the point: barely visible below 25, losing ground below 50, competitive below 75, and winning above it. A four-point move inside a band is noise, and the interface is built not to dramatise it.

CiteSite crawls the site itself, obeying robots.txt, to establish what the brand is, what it sells and who it competes with. From that it derives the queries and the prompts, which are visible and editable — nothing is measured against a question you cannot see.

Google positions come from a search-results provider rather than from scraping. AI answers come from asking the assistants directly: OpenAI for ChatGPT, Google for Gemini, and Perplexity, each with web search enabled where the model supports it.

That last detail is the one worth understanding. An answer produced with live web search is evidence about visibility today. An answer produced from a model’s memory is evidence about what it absorbed in training, which is a different and much weaker claim. CiteSite records which of the two it got for every single answer, shows it in the interface, and carries it into every export — so a recalled answer can never be filed as a measurement.

No tool can guarantee that an AI system will cite a page. Generated answers vary between users, sessions and days, and the systems producing them are neither documented nor stable. Anyone selling certainty about them is selling something they cannot deliver.

What can honestly be done is this: measure what the assistants say now, repeat it on a schedule so the direction is visible, and improve the signals that make a site easier for those systems to find, understand and reference. That is the whole of the promise.

CiteSite is operated as an independent product. It is not owned by, affiliated with, or endorsed by OpenAI, Google, Perplexity or any other search or assistant provider; those are measured, not partnered with.

For anything at all — how a number was produced, a correction, a request to see or delete your data — write to contact@citetech.site. The privacy policy lists every processor that touches your data and what each one receives.