DEJAN's Content Optimizer structures its optimization hypotheses across seven distinct Rank Factor classes: Alignment, Substance, Architecture, Style, Framing, Proof, and Other. These categories span low-level syntactic traits to high-level warranting mechanisms, enabling diversity-aware Bayesian sampling to methodically test diverse editorial strategies against re-ranker models.
| Class | Mechanism Scope | Example Factors |
|---|---|---|
| Alignment | Query and intent relevance | Query token matching, search intent fit, locality signals |
| Substance | Factual and entity content | Named entities, numeric quantification, comparisons |
| Architecture | Layout and physical structure | Lead positioning, bulleted lists, header hierarchies |
| Style | Voice, tone, and register | Register tier, assertion strength, accessibility |
| Framing | Content angle and format | Comparative frames, definition guides, how-to formats |
| Proof | Authoritative grounding | Expert citations, statistical backing, certifications |
| Other | Catch-all for emergent patterns | Novel or cross-cutting mechanisms |
Referenced by
