+ Uses only supplied facts and gives clear choices by reader type.
- Some verdict-change factors are repetitive or broadly framed.
Establishes the criteria and the table before the prose, so the post is not just two lists of pros and cons.
| Category | Marketing › Blog |
|---|---|
| Tags | AnalyzingDraftingCreator |
Structure a comparison post. Produce in this order: 1. The decision the reader is making, in one sentence. Everything else serves this. 2. Criteria — five to seven, derived from that decision and ordered by how much they affect it. *Criteria that do not change the decision do not belong in the post.* 3. Comparison table: criteria as rows, items as columns. Fill only from the facts I gave you. Write "to verify" in every other cell. 4. A verdict per reader type — "choose A if…", "choose B if…". Not a single winner. 5. What would change the verdict, so the post can be updated instead of rewritten. Rules: - *Do not fill the table from memory.* Product details go stale and a wrong cell is the thing readers catch. Unverified means "to verify". - If one item wins on every criterion, say that plainly — but check first whether the criteria were chosen to make that happen. - No conclusion that all options are fine. That is the one thing a comparison post may not say.
Comparison posts fail for lack of criteria. This derives them from the decision the reader is trying to make and compares only on those, leaving unknown cells as things to verify rather than filling them in.
ChatGPT best follows the format and fact limits. Gemini is concise but overstates some points, while Claude is specific yet materially violates the table rules.
+ Uses only supplied facts and gives clear choices by reader type.
- Some verdict-change factors are repetitive or broadly framed.
+ Its concise table and verdicts make the comparison easy to scan.
- It overstates local ownership and UI effects beyond the given facts.
+ Explains the criteria and founder-specific choices concretely.
- It breaks the required placeholder and adds unsupported server claims.
| Criterion | ChatGPT | Claude | Gemini | Leader |
|---|---|---|---|---|
| Instruction following | 10 | 7 | 9 | ChatGPT +11% |
| Accuracy | 9 | 7 | 8 | ChatGPT +13% |
| Specificity | 10 | 9 | 9 | ChatGPT +11% |
| Structure | 9 | 9 | 9 | Tie |
| Right length | 9 | 7 | 9 | Tie |
Scored 1–10 by gpt-5.6-sol with model names hidden (2026-09-24). This is an AI review, not a measurement.
We gave three models the same input and copied their answers unedited. Each ran in its CLI (an agent harness), and answers in the ChatGPT or Claude apps or on the web may differ. Outputs are in Korean.
Items: 노션(Notion) vs 옵시디언(Obsidian) vs 크래프트(Craft) Reader's decision: 개인 지식 관리(PKM) 도구를 새로 정착시키려는 1인 창업자가 셋 중 하나를 골라 6개월 이상 쓸 도구를 정하는 것 Facts I verified: 노션은 웹 기반, 협업 기능 강력, 오프라인 사용 제한적, 무료 플랜 있음. 옵시디언은 로컬 마크다운 파일 저장, 플러그인 생태계 크고 무료지만 상업적 이용 시 유료(연 50달러), 협업 기능 약함. 크래프트는 macOS/iOS 중심, UI 디자인 우수, Windows 지원 미흡, 월 구독료 약 9,900원
1인 창업자가 개인 지식 관리(PKM) 도구를 새로 정착해 6개월 이상 사용할 때, 노션·옵시디언·크래프트 중 자신의 작업 방식에 가장 잘 맞는 도구를 고르는 결정이다.
결정에 미치는 영향이 큰 순서다.
| 기준 | 노션(Notion) | 옵시디언(Obsidian) | 크래프트(Craft) |
|---|---|---|---|
| 작업 환경과 플랫폼 호환성 | 웹 기반 | to verify | macOS/iOS 중심, Windows 지원 미흡 |
| 오프라인 사용성 | 오프라인 사용 제한적 | to verify | to verify |
| 데이터 저장 방식과 장기 보존성 | to verify | 로컬 마크다운 파일 저장 | to verify |
| 협업 | 협업 기능 강력 | 협업 기능 약함 | to verify |
| 확장성과 사용자화 가능성 | to verify | 플러그인 생태계 큼 | to verify |
| 비용 부담 | 무료 플랜 있음 | 무료지만 상업적 이용 시 연 50달러 | 월 구독료 약 9,900원 |
| 인터페이스 만족도 | to verify | to verify | UI 디자인 우수 |
| Announce Milestone | |
| Place calls to action inside a blog post | |
| Turn your own experience into a blog post | |
| Build an FAQ section for a blog post | |
| Write four different blog openings |