Source-backed content generation

Example: turn a lost comparison into a grounded page

Sample Brand loses a high-intent comparison prompt to Competitor A. The Opportunity explains what was found and recommends a direct comparison page. The team opens the Comparison Page Writer, adds the product truths that must be represented, reviews the research sources, and exports a draft for human approval.

Step by step

1. Start from measured evidence

Choose a prompt loss, source gap, audit finding, or topic you want to address. Starting from an Opportunity keeps the proposed content tied to an answer the monitoring system actually observed.

2. Choose the right writing workflow

Use Article Writer for a new long-form page, Article Refining for an existing draft, or Comparison Page Writer for a head-to-head buying question. Set the topic, key points, language, and image treatment before generation.

3. Research before drafting

AEO Mantis collects relevant facts with their source URLs first. Review the evidence, add approved reference pages, and remove anything that should not support the final claim.

4. Edit, verify, and publish

The generated draft uses the approved research and preserves its citations. Verify product truths, claims, links, tone, and formatting before exporting it to your CMS; publication is never automatic.

Why grounding matters for GEO

Engines cite pages that state checkable claims. An article assembled from verified facts — with sources — is exactly the citable-atom structure that wins citations (how AI engines choose what to cite). Ungrounded AI content is the opposite: fluent, unciteable, and increasingly filtered.

Honest limits

Generation consumes plan quota per article, and research runs are metered per plan. Failed jobs refund their quota automatically. The output is a strong grounded draft — your product truths and voice still need your edit before publishing. That's by design; publish-without-reading is how brands end up cited for errors.