Anatomy of a closed-loop citation: detect → write → publish → prove

Everyone talks about AI visibility. Almost nobody shows the mechanics. This is a walk-through of one full cycle inside SiteSonar's GEO Autopilot — from detection to proven lift.

Day 0 — Detection

SiteSonar runs your prompt list against ChatGPT, Perplexity, Gemini, and Claude. For every prompt where a competitor is cited and you aren't, it captures:

  • The full generated answer per engine.
  • The cited sources and anchor text.
  • Your competitor's cited passage.
  • The nearest matching page on your own site (or the fact one is missing).

Day 0–1 — Drafting

The Content Agent takes each gap and drafts three things:

  1. An answer-first intro — the first two sentences directly answer the buyer's question, using your brand as the named entity.
  2. An FAQ block — 3–6 sub-questions with concise answers, ready for FAQPage schema.
  3. JSON-LD schema — FAQPage, HowTo, or Article as appropriate.

Every draft is written against your extracted brand guidelines and knowledge base, so the voice matches the rest of the site.

Day 1–3 — Approval

Drafts land in the approval queue. A reviewer can:

  • Edit copy in place.
  • Attach it to an existing page or create a new one.
  • Approve, reject, or request a rewrite.

Nothing publishes without a human click. This is not optional — it's the guardrail that makes closed-loop safe.

Day 3 — Publishing

Approved drafts ship through your connected publishing surface:

  • WordPress plugin — direct update to page or post.
  • GitHub — commit to a branch you choose, PR-ready.
  • MCP / AI-agent connector — for Webflow, Shopify, Wix, or custom CMS.

llms.txt updates automatically to reflect the new/updated pages.

Day 3–10 — Re-crawl

AI engines re-crawl on their own cadence. Perplexity is fastest (often within days); Google AI Overviews follows the main index (typically 2–3 weeks). SiteSonar tracks the crawl status of each affected URL.

Day 10 — Proof

Seven days after publish, SiteSonar re-runs the exact prompts against every engine and computes:

  • Citation rate before → after per engine.
  • Position of your citation inside the answer.
  • Whether the competitor is still cited alongside you or replaced entirely.
  • Any secondary prompts where the new content is now surfacing.

What one cycle looks like in the dashboard

MetricBeforeAfter (Day 10)
Citation rate (ChatGPT)0%62%
Citation rate (Perplexity)0%78%
Position in answer—1st cited source
Secondary prompts lifted04

Numbers are illustrative — actual lift varies per prompt, engine, and category — but the shape of the report is the same for every cycle.

What happens next

The next detection sweep runs immediately, using the fresh state of your site. Gaps that closed disappear from the queue. New gaps enter. With the AI Agents add-on, the whole loop runs autonomously on a schedule you set.

Where to go next

  • See GEO Autopilot in more detail →
  • Every capability in the platform →
  • Why the closed loop wins →