SEO helpful-content proxy

Business value

• Reduces risk of publishing low-value pages at scale.
• Creates an early warning signal before Search Console data arrives weeks later.
• Prioritizes editor attention toward pages most likely to damage search quality.

Agent job to be done

Act as a helpful-content QA analyst. Decide whether the page demonstrates concrete value, originality, and source/provenance signals for the target query.

format: articleintended_use: publishdomain: SEO helpful content

When to call VeracityAPI

Run before publish, and again before updating already-indexed pages after major rewrites.

What text to submit

Primary content, title, meta description, H1/H2s, intro, key comparison sections, and conclusion. Exclude global nav/sidebar/footer.

Decision policy

• allow: low risk and content_trust_score >= 0.65.
• revise: medium risk or evidence flags generic guidance, thin comparisons, or unsupported claims.
• human_review: high risk on money, safety, health, or product recommendation pages.
• aggregate policy: for section scoring, use the worst section as page risk unless only boilerplate is flagged.

Automation recipe

• SEO agent generates or updates page.
• Extractor sends unique body sections to VeracityAPI.
• Agent maps evidence spans back to headings and line numbers.
• Revision agent adds firsthand details, named examples, data, screenshots, or sources.
• Page can publish only after score falls below configured threshold.

Evidence spans agents should inspect

• thin comparisons with no concrete differentiators
• generic advice matching thousands of pages
• unsupported best/cheap/safe claims
• absence of examples, data, source links, or firsthand details

Policy pseudocode

if (result.recommended_action === "allow") continueWorkflow();
if (result.recommended_action === "revise") rewriteWith(result.evidence, result.recommended_fixes);
if (result.recommended_action === "human_review") queueForHumanReview(result);
if (result.recommended_action === "reject") discardOrRebuild();

KPIs to track

• pre-publish block rate
• percentage of pages improved before indexation
• ranking/indexing delta for passed vs failed pages
• organic traffic retained
• helpful-content QA queue size

What can go wrong

• This is not a direct Google classifier.
• Helpful content includes UX, links, reputation, and user satisfaction outside the submitted text.
• Use evidence spans to fix pages; do not chase scores blindly.

Cost and latency notes

Analyze only is $0.005 per 1,000 characters; Analyze + revise with auto_revise=true is $0.010 per 1,000 characters. Both round up to the nearest 1,000 characters. Short captions/emails usually cost $0.005; longer pages or chapters scale linearly by length. Current v0.1 latency is LLM-bound, so batch/concurrent orchestration is recommended for high-volume pipelines.