Kuroma
Answer Box Potential
What this check reads on your page
The same heading outline that Heading Hierarchy reads is walked again, this time including the h1, and for each heading the check counts the words beneath it that are not link text. A section with twenty-five to two hundred such words is a passage an assistant could quote whole; an overlong one earns a quarter credit and a very short one less than half. It asks whether the page has enough of these for its length, expecting about one per hundred and fifty words up to eight, and what share of all headings open one. Three smaller tiers count definition sentences ('X is a', 'refers to', 是指, とは) in the visible text, look for a question-and-answer block as FAQPage JSON-LD, an FAQ section or two question headings, and check for a table with a header row, a definition list, an ARIA grid or HowTo markup.
Our verdict on the evidence
Evidence-backed The real mechanism is passage self-containment: heading-scoped sections that stand alone match retrieval chunking, and fixing context-severed passages measurably cuts retrieval failures.
Anthropic contextual retrieval (self-contained passages cut retrieval failures) · GEO: Generative Engine Optimization (arXiv 2311.09735)
Retrieval works on passages, and a passage that stops making sense without its neighbours is the failure that contextual-retrieval research measures and repairs. A section that answers its own heading is the unit an assistant lifts.
What it is worth
24% of AI Readiness, which is 7.2 points of the overall 100.
How audited pages score on it
Based on 30,738 audited pages.
The median page scores 42, and the middle half of pages score between 19 and 51.
- Good (75 to 100)699 (2%)
- Moderate (50 to 74)9,237 (30%)
- Needs work (25 to 49)11,757 (38%)
- Critical (0 to 24)9,045 (29%)
68% of audited pages score below 50 on it.
Where it fails most and least
Industries that fail it most
- Passenger Transport / Transit: 90% of 138 audited pages fail it
- Banking / Neo-banking: 83% of 363 audited pages fail it
- Government / GovTech: 81% of 141 audited pages fail it
- Publishing / News Media: 80% of 905 audited pages fail it
- Streaming / Video: 79% of 756 audited pages fail it
Industries that fail it least
- AI / Machine Learning: 25% of 477 audited pages fail it
- SaaS / Cloud Software: 37% of 188 audited pages fail it
- MarTech / Marketing Platform: 40% of 201 audited pages fail it
- Developer Tools / Platforms: 41% of 175 audited pages fail it
- Data & Analytics: 44% of 144 audited pages fail it
How to fix it
- Under each heading, write twenty-five to two hundred words that answer the heading on their own, without depending on the paragraph before.
- Open a section with a plain definition sentence: 'X is a ...', 'X refers to ...'.
- Turn the questions readers ask into two or more headings ending in a question mark, each followed by its answer.
- Put comparisons in a table with a header row and procedures in HowTo markup.
- Move headings that only label a link out of the content outline, or give each of them a sentence of text.