Content Brief Builder

Customers do not buy products, they hire them to do a job. A functional job is a practical task; an emotional job is a feeling; a social job is how the outcome is perceived by others. Retrieval works on semantic meaning, so content that addresses the feeling and the situation gets retrieved for queries that never state them outright.

The opening is the strongest citation zone: 44.2 percent of AI citations came from the first 30 percent of content in a 1.2 million answer analysis. But if the content type mismatches the reader's stage, the piece is skipped regardless of how good the opening is. GEO's leverage sits upstream, in research and comparison; at the point of purchase the reader is already convinced and the job is a frictionless path, not more persuasion.

• Awareness — educational content, foundations, problem explanation. Inferred from Category and Problem queries.
• Consideration — comparison guides, expert reviews, feature breakdowns. Inferred from Solution and Competitive queries.
• Decision — authority signals, proof, evidence. Inferred from Product queries.

Category Entry Points are the situations that trigger category consideration. The 7W framework organises them into seven dimensions: Why, When, Where, With whom, With what, While, and How you feel. A CEP-anchored piece describes the situation before introducing the solution, which matches how real queries are phrased — situation first, product second — so the piece becomes findable through situational queries and not only product ones. Skip this for foundational explainers.

AI systems often cite a single section rather than a whole page. A page with six self-contained sections, each answering a real query, gives six chances to be cited rather than one. These are the sentinel queries you track to measure citation lift. A page can target one query or several — clustering saves pages, it does not raise citation odds.

AI systems use named-entity recognition to decide what a page is about, then map how entities relate. Vague references — "our product", "this tool", "it" — are invisible to that layer, and an unnamed entity cannot be cited or connected. The pattern: first mention uses the full name plus its category; later mentions use a shorter but still specific form; never pronoun-only.

Every major section must re-establish its primary entity in full, because sections are extracted independently of each other.

Name all the kinds that appear, not only your own brand: the product or service; any technology or method it uses; a component, material or ingredient; an attribute or segment the piece addresses; each competitor named in a comparison; and every named expert with their credentials. Add a row per entity below and pick its role; the Schema.org type is suggested from the role and stays editable.

Entity and naming decisions are made in Content; the JSON-LD implementation belongs to Technical. Keep your legal entities distinct if you have more than one — a parent organisation, an operating company and a consumer brand are three entities, not one.

Why evidence is decided at brief stage. Statistics, expert quotations and credible source citations each lifted AI visibility by up to 40 percent in the Princeton study, with source citations reaching up to 115 percent for a page ranked fifth — but only if the evidence is real. And a published claim never decays: an AI restates it with full confidence in month eighteen, stripped of your hedges. The rule is that every number traces to a primary record — the test report, dashboard export, spec sheet or published study it came from, not an article that repeated it. If you cannot name the record, do not publish the number.

The universal requirements, which apply in every industry and market: claim substantiation, endorsement and testimonial disclosure, and AI-generated content disclosure. Under substantiation doctrine the evidence must exist before the claim is published, the test conditions must match the claim's conditions, best-case figures must disclose what is typical, and "up to X" still needs data showing X is achievable — "up to" is not a loophole.

Classification decides the route. Claims are classified by category and by substantiation status. The two together produce a risk level, and the risk level names who has to review it before it ships.

An Answer Capsule is a 40 to 60 word, self-contained answer placed directly under the title or a heading, with no links inside the block. It names the entity, states the core claim, and carries one verifiable number. In a roughly two million session audit, 72.4 percent of cited blog posts contained one — the single most common trait of cited content — and over 90 percent of cited capsules were link-free.

It is the executable form of the answer-first principle: 44.2 percent of AI citations came from the first 30 percent of content, and the start of a long context is the model's highest-attention zone. Put citations in the supporting text below the capsule, not inside it. Evidence tier: correlational.

Every switch is governed by two forces pushing forward — the push of current frustration and the pull of the new solution — and two holding back: the anxiety of switching and the habit of the status quo. The unaddressed anxieties are exactly what comparison-stage readers ask AI about.