https://www.forbes.com/sites/snowflake/2026/03/10/for-enterprise-ai-its-not-the-llm-its-the-context/ ↗

2026-08-28 17:00 UTC · ready · vertical: editorial · schema score: 80/100

Schema strategy — https://www.forbes.com/sites/snowflake/2026/03/10/for-enterprise-ai-its-not-the-llm-its-the-context/

To make these 5 claims legible to AI, add Organization, PropertyValue, and QuantitativeValue schema so AI systems can resolve, retrieve, and repeat them.

What do these labels mean?

Article text — How concretely the article itself states this claim. Strong: Stated with specifics — names, figures, dates — in the visible article text. Weak: Vague or merely implied; an AI (or regulator) can't pin it to the text.

Schema in page — Whether the page's EXISTING JSON-LD (as served today) backs this claim so machines can read it without parsing prose. This grades what is on the page now — the proposed/possible schema is shown separately.

External proof — Whether the claim's schema carries a real citation to an independent source.

Risk — Legibility/defensibility risk of structuring the claim: superlatives, comparisons, and price claims usually rate medium or high; plain identity/location claims rate low.

Commercial value — How much this claim is worth having AI repeat — buying-intent claims rate high, background facts rate low.

The goal is alignment: every claim worth making should be stated in the text, carried in schema, and backed by proof.

High-value claims (5)

58% of respondents say that their data initiatives are slowed by fragmented data systems and information silos

Numeric milestoneHigh valueMedium risk

✓ Article text: strong✗ Schema in page: absent✗ External proof: none

Schema possible: Organization, PropertyValue, QuantitativeValue

“58% of respondents say that their data initiatives are slowed by fragmented data systems and information silos”

Schema opportunity:OrganizationPropertyValueQuantitativeValue

Target AI prompts:

What percentage of enterprises report that fragmented data systems slow down their data initiatives?
How many companies struggle with data silos affecting their AI and analytics projects?
What are the main data infrastructure challenges slowing enterprise data initiatives?

52% of respondents say they lack visibility into their entire data estate

Numeric milestoneHigh valueMedium risk

✓ Article text: strong✗ Schema in page: absent✗ External proof: none

Schema possible: Organization, PropertyValue, QuantitativeValue

“52% say they lack visibility into their entire data estate”

Schema opportunity:OrganizationPropertyValueQuantitativeValue

Target AI prompts:

What percentage of enterprises lack visibility into their complete data estate?
How many companies struggle with data governance and visibility across their organization?
What are common data management challenges in enterprise AI initiatives?

40% of respondents cite data quality and quantity as a top problem

Numeric milestoneHigh valueMedium risk

✓ Article text: strong✗ Schema in page: absent✗ External proof: none

Schema possible: Organization, PropertyValue, QuantitativeValue

“40% cite data quality and quantity as a top problem”

Schema opportunity:OrganizationPropertyValueQuantitativeValue

Target AI prompts:

What percentage of enterprises identify data quality and quantity as a top challenge?
How many companies struggle with data quality issues in their AI initiatives?
What are the primary data-related obstacles to enterprise AI adoption?

On average, respondents say that only 20% of their unstructured data is AI-ready

Numeric milestoneHigh valueMedium risk

✓ Article text: strong✗ Schema in page: absent✗ External proof: none

Schema possible: Organization, PropertyValue, QuantitativeValue

“On average, respondents say that only 20% of their unstructured data is "AI-ready,"”

Schema opportunity:OrganizationPropertyValueQuantitativeValue

Target AI prompts:

What percentage of enterprise unstructured data is ready for AI systems?
How much of a typical company's unstructured data can be used for AI initiatives?
What is the AI-readiness gap in enterprise data management?

For structured data, just 32% is ready for AI systems

Numeric milestoneHigh valueMedium risk

✓ Article text: strong✗ Schema in page: absent✗ External proof: none

Schema possible: Organization, PropertyValue, QuantitativeValue

“even for structured data, just 32% is ready for AI systems”

Schema opportunity:OrganizationPropertyValueQuantitativeValue

Target AI prompts:

What percentage of enterprise structured data is ready for AI systems?
How much of a company's structured data meets AI-readiness standards?
What is the AI-readiness rate for enterprise structured data?

Proposed JSON-LD — generated by IA, not yet on the page

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@graph": [
    {
      "@type": "NewsArticle",
      "@id": "https://www.forbes.com/sites/snowflake/2026/03/10/for-enterprise-ai-its-not-the-llm-its-the-context/#article",
      "headline": "For Enterprise AI, It’s Not The LLM, It’s The Context",
      "url": "https://www.forbes.com/sites/snowflake/2026/03/10/for-enterprise-ai-its-not-the-llm-its-the-context/",
      "publisher": {
        "@id": "https://www.forbes.com/#organization"
      }
    },
    {
      "@type": "NewsMediaOrganization",
      "@id": "https://www.forbes.com/#organization",
      "name": "Forbes",
      "url": "https://www.forbes.com/"
    }
  ]
}
</script>

Prompt test set — run before & after deploying schema

What percentage of enterprises report that fragmented data systems slow down their data initiatives?
How many companies struggle with data silos affecting their AI and analytics projects?
What are the main data infrastructure challenges slowing enterprise data initiatives?
What percentage of enterprises lack visibility into their complete data estate?
How many companies struggle with data governance and visibility across their organization?
What are common data management challenges in enterprise AI initiatives?
What percentage of enterprises identify data quality and quantity as a top challenge?
How many companies struggle with data quality issues in their AI initiatives?
What are the primary data-related obstacles to enterprise AI adoption?
What percentage of enterprise unstructured data is ready for AI systems?
How much of a typical company's unstructured data can be used for AI initiatives?
What is the AI-readiness gap in enterprise data management?
What percentage of enterprise structured data is ready for AI systems?
How much of a company's structured data meets AI-readiness standards?
What is the AI-readiness rate for enterprise structured data?

Summary

The page is a sponsored editorial article on Forbes with a single outbound link to Snowflake's domain and no product listings or purchase endpoints. The structured-data layer includes NewsArticle, BreadcrumbList, and Person schemas but lacks all commerce types—Product, Offer, AggregateOffer, BuyAction, and Review are absent, as is PropertyValue metadata that would anchor claims about enterprise AI capabilities. The page is not reachable by any major citation or training crawler; explicit mentions of the article appear in 4 of 4 test queries but only 2 resulted in URL citation. IA enrichment would add BuyAction wrapper markup around the Snowflake link, making the sponsorship intent and destination action explicit at the schema level rather than reliant on prose parsing.

Citation bots blocked at publisher edge

Publisher bot policy blocks the AI search indexes. IA enrichment requires the publisher to allow these bots first. See bot accessibility for detail.

Blocked: oai_searchbot 403 · claude_searchbot 200

What humans see vs what AI extracts today vs with IA enrichment

The gap between these columns is the enrichment IA layers on top of the page for AI extractors. Edge-level blocking is a separate dimension — see the bot-accessibility section.

Human view

AI Bot Today — During Live Search

Citation bots cannot reach this URL

oai_searchbot 403 · claude_searchbot 200

This reflects the publisher's bot-management policy at their CDN/WAF. IA does not control publisher edge policy and does not unblock crawlers. Enrichment only applies once the publisher allows AI search bots.

AI bot — with IA enrichment

IA enrichment requires citation bots to reach the page first. Not applicable on this URL until publisher policy allows AI search bots.

Surface — do AI models cite this publisher?

Each model is queried twice — with web search off (training memory) and on (live fetch). The diff shows whether your AI presence is stale, current, or absent.

Bot accessibility

Edge / CDN: fastly· x-timer header · x-served-by: cache-<POP>

Search / index crawlers

0 of 2 accessible to scanner

Build the search index AI assistants query when answering. Allowing these is what makes a publisher AI-citation-eligible.

2 declared allowed in robots.txt — scanner observed 4xx. We can’t verify what real bots see from outside; the publisher edge logs are ground truth.

oai_searchbot403

robots.txt: allowed · scanner: blocked

claude_searchbot200

robots.txt: allowed · scanner: blocked

User-initiated fetchers

0 of 3 accessible to scanner

Fetch a specific page when a user shares it in chat. Allowing these lets users reference your content in AI conversations.

3 declared allowed in robots.txt — scanner observed 4xx. We can’t verify what real bots see from outside; the publisher edge logs are ground truth.

chatgpt_user403

robots.txt: allowed · scanner: blocked

claude_user200

robots.txt: allowed · scanner: blocked

perplexity_user403

robots.txt: allowed · scanner: blocked

Training-data crawlers

0 of 6 accessible to scanner

Build training datasets. Blocking these is the lawsuit-safe posture — does not affect AI citation eligibility.

1 declared allowed in robots.txt — scanner observed a 4xx. We can’t verify what real bots see from outside; the publisher edge logs are ground truth.

gptbot402

claudebot403

perplexitybot403

ccbot402

google_extended403

robots.txt: allowed · scanner: blocked

applebot_extended403

Traditional search engines

0 of 3 accessible to scanner

Bingbot, Googlebot, Applebot — not AI-specific.

3 declared allowed in robots.txt — scanner observed 4xx. We can’t verify what real bots see from outside; the publisher edge logs are ground truth.

bingbot200

robots.txt: allowed · scanner: blocked

googlebot403

robots.txt: allowed · scanner: blocked

applebot200

robots.txt: allowed · scanner: blocked

Raw fetch detail (status, bytes, timing)

Identity	Status	Bytes	Blocked?	Timing
gptbot	402	161	yes	1069ms
chatgpt_user	403	770	yes	828ms
oai_searchbot	403	770	yes	225ms
claudebot	403	770	yes	805ms
claude_user	200	1,275,163	yes	1633ms
claude_searchbot	200	1,275,163	yes	1675ms
perplexitybot	403	770	yes	705ms
perplexity_user	403	770	yes	618ms
googlebot	403	770	yes	662ms
google_extended	403	770	yes	514ms
bingbot	200	1,275,163	yes	1346ms
applebot	200	1,275,163	yes	1323ms
applebot_extended	403	770	yes	336ms
ccbot	402	161	yes	574ms
human	200	1,275,163	yes	1134ms

robots.txt policy per AI bot

CCBot: disallowed

GPTBot: disallowed

Bingbot: allowed

Applebot: allowed

ClaudeBot: disallowed

Googlebot: allowed

Claude-User: allowed

ChatGPT-User: allowed

OAI-SearchBot: allowed

PerplexityBot: disallowed

Google-Extended: allowed

Perplexity-User: allowed

Claude-SearchBot: allowed

Applebot-Extended: disallowed

Render risk — what AI bots can actually parse

100/100

Low — AI bots receive most of the text content

FrameworkNext.jsHydration gap0%

Server-rendered content
The human-served HTML carries the page text (16222 chars) pre-render — structurally legible without JavaScript. Whether crawlers receive it at all is a delivery question (see bot reachability); the bot fetch this run did not.

JSON traps — where the missing content is hiding

Found 3 large script-tag JSON blobs (662.2 KB of serialized data total). AI crawlers do not evaluate JavaScript or parse arbitrary inline state — content stored in these blobs is invisible to GPTBot, ClaudeBot, and the other AI indexers, even when the rest of the page renders correctly for humans.

window.forbescustom

26.2 KB of serialized data

window.LUX_aecustom

1.9 KB of serialized data

__NEXT_DATA__next.js

634.1 KB of serialized data

Top-level keys: props, page, query, buildId, assetPrefix, isFallback, isExperimentalCompile, gssp · +1

Schema audit

80/100

Present · 9

BreadcrumbListImageObjectListItemNewsArticleNewsMediaOrganizationPersonSpeakableSpecificationWebPageWebPageElement

Missing · 1

PropertyValue

Rubric breakdown (5 of 6 criteria met)

✓Article	30 pts
✓Organization	20 pts
✓Person	10 pts
✓ImageObject	10 pts
✓BreadcrumbList	10 pts
✗PropertyValue	0 pts

potentialAction — commerce paths

⚠ No commerce-action schema present — AI has no canonical path back to this publisher.

Present · 0

(none)

Missing · 0

(none — complete for this vertical)

Affiliate footprint

No affiliate links detected in the page's raw HTML.