Skip to content
SEOWofSEOWof
Browse the standard

Standard SW-1:2026.2 · Category 9 of 9

AI & Answer Engine Readiness

Added to the standard in SW-1:2026 and weighted conservatively while the evidence base matures. We inspect only what is objectively verifiable — crawler access, server-rendered content availability, attribution clarity and factual consistency — and we do not score anything that depends on an unverifiable claim about how a model ranks sources.

Weight
5 / 100
Checks
3
Pass-blocking
1
WOF analogue
New-standard compliance
AI-01Minor failweight 2measured

Answer-engine crawlers are not blocked unintentionally

Have you decided, deliberately, which AI crawlers may access your site?

Pass, advisory and fail criteria for AI-01
PassA deliberate, documented position exists, and retrieval crawlers are permitted if citation visibility is a goal.
AdvisoryNo explicit rules, so all crawlers are permitted by default without a decision having been made.
FailRetrieval crawlers are blocked while the business simultaneously expects to appear in AI answers.

Why it matters

Blocking training crawlers and blocking retrieval crawlers are two entirely different decisions with different consequences, and they are routinely conflated. Blocking a retrieval crawler removes you from live answers that would have cited and linked you. Blocking a training crawler does not. Many sites have blocked both by copying a snippet without understanding the distinction — and many CDNs now block them by default.

How we inspect it

We enumerate the agent groups in robots.txt and identify which answer-engine and AI crawlers are permitted or blocked, distinguishing training crawlers from retrieval crawlers that fetch pages to cite in live answers. We report the position; we do not judge it.

Source: robots.txt agent-group analysis plus CDN bot-rule inspection where declared

Remedy

  1. 1List every AI-related agent currently blocked and classify each as training or retrieval.
  2. 2Decide explicitly whether citation visibility is worth content ingestion for your business model.
  3. 3Implement the decision in robots.txt and check your CDN or bot-mitigation layer is not overriding it.
  4. 4Document the decision and its date so it is revisited deliberately rather than inherited.

Verification

Recheck re-reads agent groups and confirms the implemented position matches the declared intent.

Usually fixed by
Owner
Effort
Small · 0.5–2 hrs
Est. impact
medium
Typical cost
NZ$120–400
AI-02Major failweight 2measured

Primary content is available without JavaScript execution

Is your content in the HTML, or only after scripts run?

Pass, advisory and fail criteria for AI-02
PassOver 90% of main content, all headings, internal links and structured data present in the raw HTML.
Advisory70–90% present, with secondary modules hydrated client-side.
FailUnder 70% of main content present, or navigation and structured data only appear after hydration.

Why it matters

Googlebot renders JavaScript, but on a deferred queue — and most answer-engine retrieval crawlers do not render at all. Content that only exists after hydration is therefore invisible to a growing share of the systems that decide whether to cite you, and slower to be indexed by the ones that do. Server rendering has moved from a performance optimisation to a visibility prerequisite.

How we inspect it

We fetch the raw HTML without executing JavaScript and diff main content, headings, internal links and structured data against the fully rendered DOM, reporting the proportion available pre-render.

Source: Raw HTML fetch compared against the fully rendered DOM

Remedy

  1. 1Move primary content, headings, navigation and structured data into the server-rendered response.
  2. 2Adopt server-side rendering or static generation for content routes; keep client rendering for genuinely interactive components.
  3. 3Never gate main content behind an authenticated or client-only data fetch.
  4. 4Test by fetching with JavaScript disabled and reading what remains.

Verification

Recheck re-diffs raw HTML against the rendered DOM and confirms the pre-render content proportion.

Usually fixed by
Developer
Effort
Large · 1–3 days
Est. impact
high
Typical cost
NZ$800–3,200
AI-03Advisoryweight 1derived

Content is structured for accurate extraction

Can a machine lift a correct, self-contained answer from your page?

Pass, advisory and fail criteria for AI-03
PassHeadings pose questions answered directly beneath in self-contained passages; data is tabulated; claims are dated and attributed.
AdvisoryStructure is reasonable but answers are diffuse or require surrounding context to make sense.
FailContent is undifferentiated prose with no extractable structure on pages targeting question queries.

Why it matters

Extraction systems lift passages, not pages. A passage that only makes sense after reading three paragraphs above it will either be skipped or quoted incorrectly, and a misquote is worse than an omission. Structuring for extraction is not a new discipline — it is the same clear writing that helps a human skimming on a phone — but doing it deliberately materially changes citation rates.

How we inspect it

We assess whether questions posed by headings are answered directly beneath them in self-contained passages, whether comparative data is presented in parseable tables rather than prose, and whether claims carry attributable sources and dates.

Source: Content structure analysis: heading-to-answer proximity, list and table usage, claim attribution

Remedy

  1. 1Answer the question in the first two sentences under the heading that poses it, before adding nuance.
  2. 2Write passages that stand alone without requiring the preceding section.
  3. 3Put comparative and numeric information in real tables, not in prose or images.
  4. 4Attribute claims with a source and a date so they can be quoted with confidence.
  5. 5Keep specifics — figures, constraints, named conditions — since these are what get quoted.

Verification

Recheck re-analyses content structure on the nominated pages against extraction criteria.

Usually fixed by
Content
Effort
Medium · 2–8 hrs
Est. impact
medium
Typical cost
NZ$320–1,200