Run the instrument
Paste a page URL into the lab. INGEST fetches like a machine, extracts atomic facts, and keeps contradictions visible with evidence.
What INGEST is
INGEST is a machine comprehension test harness for search engines and LLM crawlers. The product question is not “Is this optimized for AI?” It is: can a machine attribute important claims cleanly?
It is not a traditional SEO auditor and not an “AI visibility score” product. UNKNOWN stays UNKNOWN. Contradictions stay visible.
How it works
- Fetch like a machine — raw HTML, crawler identities, robots policy, and rendered DOM, kept separate.
- Extract atomic facts — only propositions the source supports. Questions are not claims. Qualifiers survive.
- Compare representations — visible HTML vs JSON-LD vs other machine layers. Contradictions stay visible.
- Preserve evidence — every finding links back to source text, selector, and measurement confidence.
Example: live fact tension
Production pages often publish the same question with incompatible answers across layers:
- Sephora · Structured Offer: “InStock · ships free with Beauty Insider on orders $50+.”
- Sephora · Visible PDP: “Currently sold out online. Check your local store for availability.”
The same class of tension shows up across other surfaces INGEST is built to catch:
- Genentech — labeled usage excerpt vs a stale FAQ still citing a prior revision
- Car sales & rentals — JSON-LD daily rate excluding fees vs “all-in” hero copy
- Apartment rentals — structured rent that omits pet rent quoted on the leasing page
- Vacation stays — free cancellation in schema vs non-refundable visible terms
Scope may reconcile these. INGEST keeps both assertions and does not auto-pick a winner.
What the instrument measures
- Public crawler identities (access matrix)
- Raw HTML vs rendered DOM / JS dependency
- JSON-LD parse validity ≠ schema truth
- Atomic facts with assertion mode
- Human/machine parity classifications (MATCHED, HUMAN_ONLY, MACHINE_ONLY, CONTRADICTORY, STALE, AMBIGUOUS)
- robots.txt / sitemap / machine files
- Permanent regression fixtures and measurement-validity CI
Full contract: INGEST methodology.
Who it is for
Operators — see whether AI-facing answers can misstate offer, pricing, or policy from conflicting page layers.
Practitioners — audit crawler access, structured-data integrity, fact parity, and entity identity with permanent fixtures.
FAQ
What is INGEST?
INGEST is NRLC’s machine comprehension test harness. It asks whether a machine can retrieve, parse, extract, verify, relate, and trust important facts on a page without unnecessary inference. It is not an SEO score product.
How is INGEST different from the Search System Diagnostic?
The public diagnostic on /diagnostics/ runs bounded discovery checks on a host. INGEST goes deeper on a URL: crawler access matrices, raw HTML vs rendered DOM, JSON-LD parse vs schema truth, atomic fact extraction, and human/machine parity with evidence.
Does INGEST decide which conflicting claim is true?
No. When visible copy and structured data disagree, INGEST keeps both assertions and classifies the tension. It does not auto-pick a winner.
Is INGEST a ranking or AI-citation guarantee?
No. It measures machine-readable substrate and fact trustworthiness. It does not claim rankings, traffic, or citation rates.