Different output is not a different page job
Two AI tools can return different branches, examples, caveats, or sources for the same prompt. Those differences may reveal questions worth checking. They do not show which URL currently owns the user job, whether the site already answers it, or whether the evidence is ready to publish.
The page decision therefore starts after the comparison. Keep useful branches as declared inputs, then inspect the actual site inventory before adding work to the roadmap.
What can travel downstream
Preserve the topic, user job, branches, questions, entities, source roles, provider label, date, visible mode, and any known settings. Treat every imported branch as user-supplied planning material rather than a hidden query or verified demand signal.
A portable record makes the handoff reviewable. It also preserves provenance when a branch later becomes a section, research task, or rejected idea.
- Keep the original topic and primary user outcome.
- Keep the visible provider and run context without inferring hidden internals.
- Deduplicate wording variants before page review.
- Mark source suggestions as unverified until the source is opened and checked.
What the site must still prove
Name the current canonical owner, page promise, indexability, evidence state, and next step for the reader. Then classify coverage as no owner, a section gap, a material page gap, substantial coverage, overlap, or unknown.
This is the decision-critical evidence. Without it, a long fan-out can only support a research task—not a reliable URL recommendation.
Keep the two product roles separate
AI Fanout is the separate upstream workspace for planning branches and public-answer evidence. SEO Fanout decides how those inputs should change a real page set. Both share the same operator, so neither is independent validation of the other.
If you already have an AI Fanout export, import it locally and complete the inventory review here. If you only need to compare provider outputs, stay in the upstream evidence workflow.