Framework How-To Guide aiknowledgesignal.io

How to Determine a Page's Content Type and Epistemic Role

Every page plays one role — primary contribution, synthesis, applied interpretation, or reference. Decide which before you write, declare it, and keep it pure.

Why it matters

Readers, and the systems that summarise for them, need to know what kind of thing a page is. A page of original research is a primary source; a structured synthesis is a useful map of a field; a reference page is a canonical definition. When a single page mixes original research, opinion, and summary, it is hard to tell which parts are load-bearing — and an answer engine is more likely to quote your opinion as if it were your data.

Choosing and declaring an epistemic role up front sharpens the page. It tells the reader (and the model) exactly how to use the content, and it is the foundation for the rest of Phase 3 — you can't make a contribution explicit (3.2) until you know what kind of contribution it is.

What good looks like

The page commits to a single role and says so:

Step by step

  1. Pick one role for the page Before writing, name the page's role: primary contribution, synthesis, applied interpretation, or reference. If you can't decide, the page probably hasn't found its purpose yet.
  2. Declare the role in the opening paragraph State it plainly: "This page is the canonical definition of [term]" or "This is a structured synthesis of [sources]". The declaration orients both readers and AI systems immediately.
  3. Keep the role pure Resist mixing. If you're writing original research, don't pad it with a generic explainer; if it's a synthesis, attribute every borrowed idea and don't pass it off as primary. Purity is what makes the role legible.
  4. Match the schema type to the role Reflect the role in your JSON-LD (Phase 5.1): ScholarlyArticle for research-grade primary work, TechArticle for methods, Article for synthesis, a definition page for reference. The schema should agree with the declared role.
  5. Split pages that mix roles If a draft is doing two jobs — say, presenting a dataset and arguing an opinion — split it. Publish the research as a primary contribution and the argument as a clearly-labelled interpretation that cites it.

Worked examples

1 The four roles, with a declaration

Each role has a recognisable opening declaration. These are the sentences that tell a model what the page is.

RoleOpening declaration
Primary contribution"This page reports our original analysis of [N] AI responses, with methodology and sample size below."
Synthesis"This is a structured synthesis of five 2026 GEO studies, each named and dated."
Applied interpretation"This is an applied interpretation of the GEO study (Aggarwal et al., 2024) for healthcare publishers."
Reference"This is the canonical definition of Corpus Survival Likelihood."
2 Before / after: mixed vs. pure role
Before A generic blog post that mixes some original data, several paragraphs of opinion, and a Wikipedia-style summary — with no stated role. The model can't tell the data from the commentary.
After One epistemic role per piece, labelled: "This is an applied interpretation of the GEO study (Aggarwal et al., 2024) for healthcare publishers." The original data lives in its own primary-contribution page that this one cites.

Common mistakes

MistakeFix
No stated roleDeclare it in the opening paragraph.
Mixing research, opinion, and summary in one pageKeep one role per page; split the rest out.
Synthesis presented as primary evidenceAttribute sources and label it a synthesis (a Hard Rule).
Schema type contradicts the roleAlign the @type with the declared role.
Reference page padded with opinionKeep reference pages clean and definitional.

Limits of this advice

Checklist

  • The page has exactly one of the four epistemic roles
  • The role is declared in the opening paragraph
  • No mixing of primary research, opinion, and summary
  • Schema @type matches the declared role
  • Any page doing two jobs has been split