The 120–150 Word Answer Capsule: The Exact Passage Structure AI Engines Extract

The Short Version

An answer capsule is a self-contained 120–150 word answer placed directly under a heading, written so it makes sense with zero surrounding context. In Princeton's GEO study, adding answer capsules was one of the most reliable visibility boosts measured, in the same family as statistics (+41%) and expert quotes (+37%) (Aggarwal et al., 2024). The reason is mechanical: retrieval-augmented systems pull passages, not pages, so every passage must carry its own answer.

Why LLMs Extract Passages, Not Pages

Google's AI Mode, ChatGPT with browsing, and Perplexity do not read a page the way a person scrolls it. They retrieve candidate passages and assemble an answer from the fragments that best match the query. Research across citation studies puts a sharp number on this: around 44% of LLM citations come from the first 30% of a page. If your core answer sits in the conclusion, the retrieval system often never reaches it.

The practical consequence is that every section of your article is a candidate for a citation on its own. Write each section as if a stranger will read only that section, on its own, with no other context from the page.

What an Answer Capsule Actually Is

A capsule has five parts:

• A direct first sentence that answers the question in the heading. No throat-clearing.
• One core claim stated with a concrete fact or number where possible.
• A supporting explanation of why the claim is true, in 2–3 sentences.
• A self-contained boundary — no "as mentioned above," no "we discussed earlier."
• A length of roughly 120–150 words, short enough to be extracted whole, long enough to be substantive.

The heading above the capsule matters as much as the capsule itself. Question headings signal exactly what the passage answers, which is why question-based headings outperform topic labels in both human scanning and AI extraction.

The +40% Data Point and Why It Holds

The Princeton study tested six content optimization strategies across 10,000 queries and measured visible citation lift: statistics at +41%, answer capsules at +40%, expert quotes at +37%, inline citations at +22%. Keyword stuffing, by contrast, reduced visibility by 10%. Two lessons:

• Extractability tactics compound. Statistics, capsules, and quotes are all ways to make a passage self-sufficient and quotable.
• Over-optimization backfires. AI models detect stuffed, unnatural copy and deprioritize it.

Five Reusable Capsule Templates

1. The Definition Capsule

"X is [definition]. It is distinct from [adjacent concept] because [key difference]. In practice, [concrete example]. Since [date or source], [what changed]."

Use this for any concept your audience might ask AI to define. Definitional passages are citation anchors for the query and its related forms.

2. The Comparison Capsule

"To choose between A and B, compare [criterion 1], [criterion 2], and [criterion 3]. A wins on [attribute] and suits [scenario]. B wins on [attribute] and suits [scenario]. If your priority is [use case], pick [choice]."

Comparison content performs strongly in AI answers — decision queries are among the highest-volume types in AI search.

3. The How-To Capsule

"To [goal], follow these steps. Step 1: [action]. Step 2: [action]. Step 3: [action]. You know it worked when [checkpoint]. Budget roughly [time or cost]."

Step sequences map cleanly onto the instruction lists AI engines generate.

4. The Data Capsule

"[Claim] with [number] in [timeframe], according to [source], [date]. A follow-up study by [source] found [second number]. The practical implication is [what the reader should do]."

Each data capsule should carry two dated facts — the same two-statistics-per-section discipline used in research-fed articles.

5. The Decision Capsule

"Choose [option] when [condition]. Choose [option B] when [different condition]. If you are unsure, [tiebreaker test]. The cost of guessing wrong is [downside]."

Decision capsules answer the "which should I pick" queries that AI Mode data shows are growing fastest.

Before and After: One Heading, Two Paragraphs

Before (vague, not extractable): "Schema markup is quite important for modern search. Many experts agree that adding structured data to your pages helps search engines and AI systems understand your content better. There are different types of schema and you should probably use some of them."

After (self-contained answer capsule): "Schema markup tells search engines and AI systems what a page is about in a language they parse directly. Pages with schema are cited more often by AI tools, though adding schema to pages that were already cited made no measurable difference (Ahrefs, 2026). For most sites the highest-value types are Article, Organization, Breadcrumb, and FAQ — they resolve the entity, the author, and the question-answer pairs AI engines extract most."

The after version is a standalone answer. It could be cited on its own, with nothing else from the page.

Capsule Rules Checklist

• Put the direct answer in the first sentence of each section.
• Keep each capsule between 120 and 150 words.
• Use question headings above every capsule.
• Include at least two dated statistics per article section.
• Never reference other sections — "as discussed" breaks extraction.
• Front-load your conclusion; 44% of citations come from the first 30%.

When Capsules Overlap with FAQ Schema

Answer capsules and FAQ markup serve the same goal from different directions: both create clean question-answer pairs for machines to extract. A page built with capsules gives you the content; structured data gives you the explicit Q&A mapping. They are complementary, not competing — see how FAQ schema performs specifically in AI Overviews for the pairing strategy that works.

The Bottom Line

Answer capsules are the cheapest extractability upgrade available to any site. They require no design, no schema, and no tooling — only a stricter way of writing sections. Structure every heading as a question, answer it in 120–150 words in the first block, and you have built the exact passage shape retrieval systems are looking for.