Entity‑first SEO: how I build a content brief with entities, info gain, and sources in 12 minutes

Posted 4 months ago by Dino Cajic

Keywords are table stakes. Entities are how modern search, and now AI systems, actually understand a topic. When your brief is built around entities and information gain, you give models and humans the same thing they want: clear coverage of “things, not strings,” plus original insights that move the conversation forward.

Below is the exact, time‑boxed workflow I use to go from a blank page to a publish‑ready brief in 12 minutes. It’s AI‑assisted, repeatable, and it scales across teams.

What “entity‑first” really means (in plain English)

An entity is a real‑world thing (person, place, product, concept) with attributes and relationships. Google formalized this with the Knowledge Graph so it can disambiguate and connect meaning across the web. If your brief starts with entities (and how they relate), you’ll outline the actual questions, attributes, and comparisons people need, not just a pile of synonyms.

Two guardrails make entity‑first work:

  • Structured data: Mark up the page so machines can align your content to recognized entities (use JSON‑LD where possible).
  • Quality signals: Draft with the same factors raters use (experience, expertise, authoritativeness, trust) knowing rater feedback evaluates systems and doesn’t directly rank pages. It’s still a great bar for usefulness.

The 12‑minute brief (stopwatch version)

Minute 0–1: Define the job‑to‑be‑done.

One line: “Who’s the reader, what decision do they need to make, and what’s the win if they make it?”

Minute 1–3: Map the entity set.

List the primary entity, a short list of adjacent entities, and the key attributes/relations people use to evaluate them. Think “what it is,” “how it works,” “metrics,” “alternatives,” “risks,” “standards,” “people/brands,” “timeframes.”

Minute 3–5: Baseline the market conversation with AI.

Have your LLM scan top results and extract H2/H3s, questions, and claims. You’re not copying; you’re finding the overlap everybody covers.

Minute 5–7: Decide your information gain.

Information gain is the net new value your page adds beyond what’s already on page one. Google has even explored ranking methods based on “information gain scores,” emphasizing content that contributes new, additional information. In other words: be the page that actually adds something.

Minute 7–9: Lock sources.

Pick 3–5 primary sources (official docs, standards, datasets, patents) and 2–3 secondary sources (industry analyses) that underpin your info‑gain sections. Favor originals over summaries.

Minute 9–10: Draft the outline.

Write the H2s/H3s so each heading maps to an entity or relationship. Add one sentence under each heading stating the claim and the source you’ll use.

Minute 10–11: Add structure.

Define the schema types you’ll ship (Article/FAQ/HowTo/Product/Organization) and the internal links that connect this piece to your cluster. Use JSON‑LD and fill sameAs where appropriate to align with known entities.

Minute 11–12: Package for production.

Three title options, a one‑sentence hook, bullets for info‑gain sections, source list, schema plan, internal links, CTA, and KPIs.

The 1‑page brief template (copy/paste)

Working Title Options
1) …
2) …
3) …

Reader & JTBD
• Who it’s for:
• Decision / outcome:

Primary Entity
• Name:
• Definition (one line):
• Related entities (5–10): 
• Attributes & relationships to cover:

Query & Intent
• Core intent:
• Secondary intents:

Information Gain (what we add beyond page one)
• New angles:
• Data/diagrams we’ll produce:
• Contrasts/comparisons we’ll own:

Outline (H2/H3)
• H2: …
  – Claim (one line):
  – Source to cite:
• H2: …
  – Claim:
  – Source:

Sources (link + why it matters)
• Primary (docs, standards, datasets, patents):
• Secondary (analyses, reviews):

Structured Data & Entity Alignment
• Schema types:
• Key properties (author, date, headline, mainEntity, sameAs):
• Internal links (anchor text -> URL):

Distribution & KPIs
• Channels:
• Lead metric (e.g., read% or assisted conversions):
• Supporting metrics (e.g., avg. scroll, citations earned):

My AI prompt trio (fast and reliable)

1) Entity mapper (2 minutes) “Act as a topical graph builder. For the topic ‘{TOPIC}’, list the primary entity, 10–15 adjacent entities, and the attributes & relationships that matter to a buyer or practitioner. Return as three flat lists: Primary, Adjacent, Attributes/Relationships. Keep it specific, no fluff.”

2) Market overlap (2 minutes) “From the top 10 results for ‘{TOPIC}’, pull H2/H3s and the explicit claims under each. Return a deduped list of covered subtopics and common claims. Do not generate any new claims.”

3) Info‑gain finder (2 minutes) “Given the entity set and common coverage above, propose 5–7 unique angles that add new information. Prioritize original data, practical frameworks, local context, quantified comparisons, or counterintuitive findings. For each, suggest one primary source we can cite.”

How I pick and use sources (without slowing down)

  • Start with primaries: vendor or standards docs, original datasets, patents, and official guidelines. That’s your defensibility. For search topics, Google’s own docs are gold for how machines interpret content and markup.
  • Contextualize with secondaries: respected trade pubs or analysis that frame the “why now.”
  • Cite lightly, verify deeply: one inline citation where you make a factual claim; keep most of the narrative in your voice.
  • Align to how quality is evaluated: even though rater evaluations don’t directly rank your page, they mirror what “helpful for people” looks like. Use them as your quality checklist.

Why this plays well with AI‑led search

LLMs prefer content that’s organized around entities and relationships because it reduces ambiguity and makes retrieval cleaner. Schema gives machines explicit clues, and info gain gives humans a reason to share and link. That combination future‑proofs content whether the answer appears in a search result, a chat response, or a vertical AI assistant.

Quick example of an info‑gain move

If “passkeys vs passwords” is saturated with definitions and pros/cons, your info‑gain might be: a field‑tested rollout playbook with measurable risk reductions by org size, plus a calculator that estimates support ticket impact. That’s new, useful, and cite-able, exactly what “information gain” encourages.

Ship checklist

  • Entity lists mapped to headings
  • At least one original element (chart, table, calculator, or mini‑study)
  • Primary sources attached to the claims they support
  • JSON‑LD added, validated
  • Internal links placed, anchors written for entities
  • CTA and one lead KPI defined

Leave a Reply