AgenticSEO platform terms

A published item is the unit of delivered work in AgenticSEO. It is one of exactly three things: an approved fix pushed live (title, meta description, JSON-LD, llms.txt), a published answer-ready FAQ, or a full page published to the customer's site.

The distinction that matters is deployment. A draft, a downloaded .docx, or a recommendation waiting for approval is not a published item — only work that reached the live site counts. That makes the measure honest: it cannot be inflated by generating more suggestions.

Published items are what the first-month delivery guarantee is counted in, and every one is recorded in the delivery ledger at the moment it goes live.

The delivery ledger is appended to by the publishing paths themselves — metadata and schema pushes, FAQ publishes, and page publishes — so a row exists only when something genuinely went live. Each row names the company, the item type, the page URL, and the moment of publication.

It exists so an outcome promise can be verified rather than asserted. The running count feeds the guarantee card on the customer dashboard, and the same rows are the evidence TGP reviews when a first-month delivery guarantee period closes short.

Because publishing requires customer approval and a connected publish target, a quiet ledger usually means work is queued for approval rather than that nothing was produced — which is why guarantee periods pause instead of paying out in that case.

Auto-discovered competitors are proposed by Finn (Gemini) using the same brand context that seeds tracked prompts: your industry, value proposition, product names, and site content. They are stored in the competitors table with source='auto' so they can be visually distinguished from manual entries.

Auto entries participate in every downstream signal: “X vs Y” prompt suggestions, competitor citation detection on prompt runs, and Share-of-Voice calculations on the Benchmarks dashboard. Users can hide any auto entry from the Competitors card if it isn’t a real competitor, and hidden entries are not resurrected by later Auto-discover runs.

MCP and REST integrators (like Quid) can trigger auto-discovery headlessly via the discover_competitors tool, so a downstream agent never has to route the user back into our UI to bootstrap the competitor set.