RankX AI documentation
How RankX AI measures whether AI assistants name your brand, what every number means, and how to drive the whole platform from an AI assistant over MCP.
RankX AI measures whether AI assistants name your brand when someone asks them a buying question, and turns what it finds into a prioritised fix list. It tracks five assistants, Google AI Overviews, two AI shopping surfaces and ordinary Google positions, from one account.
These pages are written from the product's own code by the engineers who build it, and they live in the same repository, so a change to the product and a change to its documentation are the same commit.
Where to start
What RankX AI is
The five surfaces RankX AI watches, and the questions each one answers.
Quickstart
From signing up to your first visibility numbers, in order, with what happens at each step.
Your first week
What appears when. Most of the platform is empty on day one, and that is correct.
Credits and metering
One wallet, reserved before the work and released if it fails. The part buyers most often misread.
What RankX AI measures
RankX AI watches four kinds of surface, and they are genuinely different measurements that must not be averaged together.
- AI assistant answers. RankX AI asks tracked prompts on ChatGPT, Gemini, Claude, Perplexity and Grok, then records whether your brand was named, where it sat, and which sources the answer leaned on. See how RankX AI measures visibility.
- Google AI Overviews. A separate surface with separate data, collected on your tracked keywords by the rank tracker rather than by the scheduled prompt run. The two kinds of citation are the most common confusion in the product, and citations, two kinds exists to settle it.
- AI shopping answers. Whether your products are recommended on ChatGPT Shopping and Google AI Mode, which merchants take the slots instead, and where ads are observable.
- Ordinary Google positions, to depth 20, plus Search Console and Analytics once those are connected.
What RankX AI does with it
Measurement without a next step is a dashboard nobody opens. RankX AI turns findings into a Tasks board, scores how machine-readable your site is with AI Readiness, audits the site itself, researches keywords, and drafts content against briefs.
Every screen, and where it is documented
These docs are organised the way the product's own navigation is, so you can read with the app open beside you. This is the full map.
| In RankX AI | Documented at |
|---|---|
| Dashboard | The section pages below. The dashboard is a summary of them, and every tile on it links to the screen it summarises |
| Website Audit | Website Audit |
| Tasks | Tasks |
| Visibility Overview | AI Visibility |
| AI Prompts | AI Prompts |
| AI Chat Feed | AI Chat Feed |
| AI Answer Citations | AI Answer Citations |
| AI Readiness | AI Readiness |
| Shopping Overview | AI Shopping |
| Products | Products and the catalogue |
| Merchants | Merchants |
| Ads | Ads |
| Rankings Overview | Google Results |
| Rank Tracking | Rank tracking |
| AI Overviews | AI Overviews |
| AI Overview Citations | AI Overview Citations |
| Traffic Overview | Traffic |
| Search Console | Search Console |
| Google Analytics | Google Analytics |
| Indexing | Indexing |
| Keyword Research | Keyword research |
| Topic Clusters | Topic clusters |
| Content Studio | Content briefs, generating articles, quality scores and publishing. Content Studio is the workspace those four steps happen in |
| Campaigns | Campaigns |
| Agency Admin | Running an agency |
| Settings | Integrations, plans and limits and security |
Super Admin is staff-only and cross-tenant, so it is deliberately not documented here.
Driving RankX AI from an AI assistant
RankX AI's programmatic interface is its MCP server: a Model Context Protocol endpoint with a documented tool surface, scoped permissions, an OAuth 2.1 authorisation server and a set of ready-made Agent Skills. Connect Claude, ChatGPT, Cursor, VS Code or a script, and ask for the work in words.
There is no public REST API. That is a deliberate statement rather than an omission: MCP is where the effort has gone, and the MCP section documents every tool, its scope and what it costs to call.
For agents
Every page here is available as clean Markdown by appending .md to its URL, or
by sending Accept: text/markdown. The documentation index is at
/docs/llms.txt, the curated brand file at
/llms.txt, and the full site corpus at
/llms-full.txt.
Reference tables under /docs/reference are generated from the product itself
rather than typed by hand, and each carries the source commit it was read from.
Terminology is fixed and used consistently: a Website is one site you track, a Client Workspace is one of an agency's clients, AI Overviews is Google's summary above the results, and AI Answer Citations are the domains cited inside an AI assistant's answer. The last two are different data from different engines.