When to use this prompt
When building or expanding a category glossary on your site. Glossaries are one of the most underused content investments in B2B because they read as boring, but they are heavily cited by AI engines because the format is exactly what retrieval systems want: a clean term, a self-contained definition, and consistent structure across entries.
A glossary of 30-50 terms in your category, properly built, can be the single most-cited section of your site within a year.
The prompt
<role>Editor producing glossary entries for AI search citation. Each entry is a self-contained, citable passage.</role>
<task>Write a glossary entry for each term below. Every entry follows the same structure so retrieval systems can extract any one and cite it cleanly.</task>
<inputs>
<category>[CATEGORY THE GLOSSARY COVERS, e.g., "product analytics for SaaS"]</category>
<audience>[WHO READS THIS, e.g., "early-career product managers"]</audience>
<terms>
1. [TERM 1]
2. [TERM 2]
3. [TERM 3]
... (add as many as needed)
</terms>
<tone_rules>[OPTIONAL: any tone or voice guidance, e.g., "no marketing copy, neutral encyclopedic voice"]</tone_rules>
</inputs>
<entry_structure>
Each glossary entry must follow exactly this structure:
**Term**
**Definition (1 sentence, ≤30 words):** A self-contained definition. The first sentence must stand alone as a citation.
**Expanded explanation (2 to 4 sentences):** Why the term matters, how it is used in practice, who uses it. Include one specific example or use case. Do not include marketing language.
**Related terms:** 2 to 4 other glossary terms a reader might also want, separated by commas. Use only terms that are likely to exist in this glossary or a sibling glossary.
**Common misconceptions or confusions (optional, 1 sentence):** Only include if there is a genuine confusion the reader should be warned about.
</entry_structure>
<instructions>
1. Generate one entry per term in the input list. Maintain the exact structure for every entry.
2. The first sentence must answer "what is X" directly. No throat-clearing. No "X is a concept that..." Start with "X is..." or "An X is..." or the equivalent.
3. The expanded explanation must include one specific named entity, tool, metric, or example. Generic explanations are not citable.
4. Do not invent definitions. If a term is ambiguous, define the most common usage in the named category and flag the ambiguity in the misconceptions field.
5. Do not link entries to specific brands or products. Glossary entries are reference material, not marketing pages.
6. Do not exceed the structure. No additional sections, no examples block, no FAQs. Glossary entries are tight on purpose.
</instructions>
<output_format>
For each term, exactly:
---
**[Term]**
**Definition:** [≤30 words]
**Expanded explanation:** [2-4 sentences with one named example]
**Related terms:** [2-4 terms]
**Common misconceptions:** [optional 1 sentence, omit if none]
---
(repeat for each term)
</output_format>How it works
The single-sentence definition that opens each entry is the citation surface. AI engines retrieving on a “what is X” query reach first for that opening sentence. If it is missing, vague, or buried, the citation goes to a competitor’s glossary that did the work.
The 30-word cap on the definition is intentional. Citation snippets in AI Overviews and answer engines are typically 20-50 words. A 60-word “definition” gets truncated mid-sentence, which is unciteable.
The “related terms” line is a 2026 best practice that mimics how Wikipedia surfaces context. It also creates internal-linking opportunities when you turn each glossary entry into a page: each related-term anchor becomes a link.
The “do not link to brands or products” guardrail is critical. Glossaries that say “X is a concept best implemented with [Brand Y]” stop being reference material and start being marketing copy. Retrieval systems detect this and discount the page accordingly.
Example output
Cohort analysis
Definition: Cohort analysis is a method of grouping users by a shared starting characteristic, then measuring how their behavior changes over time.
Expanded explanation: Product teams use cohort analysis to evaluate whether changes to onboarding, pricing, or product features improve user retention. A typical cohort is “users who signed up in March 2026,” tracked weekly to see what percentage are still active 30 and 60 days later. Without cohort analysis, lift signals from new releases get washed out by churn from older users.
Related terms: retention curve, churn rate, activation event, day-N retention
Common misconceptions: Cohort analysis is not the same as segmentation; cohorts share a starting time, while segments share an attribute.
Variations
- Term-extraction first: Paste the contents of a category overview page and ask the model to surface 20-40 candidate glossary terms before generating entries. Useful for building a glossary from scratch.
- Comparison entries: For two closely related terms, add a “vs” entry that explicitly compares them. Useful for terms readers confuse (e.g., “AEO vs GEO”).
- SEO mode: Add an instruction that each entry includes one external citation to an authoritative source (Schema.org, Wikipedia, an industry standards body). Boosts the page’s authority signal without making the glossary look promotional.