Optimization Engine
Turn Content Genome findings into typed, deterministic recommendations proposing exact DOM-level changes to improve AI understanding.
Private beta — available to agency partners
This capability is live and working for agencies in GeoScript's private beta program. Learn about agency access →
What Optimization Engine does
The Optimization Engine is GeoScript's recommendation layer. It reads the structured Content Genome for a client's pages and applies deterministic rules to produce typed recommendations — specific, actionable proposals for changes that will improve how AI engines understand and represent the business. Recommendations are not generic suggestions: each one proposes an exact change at the DOM level, which can be reviewed and then applied via the Automated Deployment pipeline.
How it works
Six deterministic rules (v1)
The current engine runs six rules: missing_local_business_schema (no JSON-LD LocalBusiness block detected), missing_faq_block (no structured FAQ section), thin_content (page body below minimum content threshold), missing_meta_description (no meta description tag), multiple_h1 (more than one H1 tag on the page), and missing_service_area_block (no geographic service area reference in content). Each rule is deterministic — the same Content Genome input produces the same recommendation output every time.
Typed recommendations
Every recommendation has a type that corresponds to the rule that triggered it. The type determines how the Automated Deployment pipeline packages and applies the change. A missing_local_business_schema recommendation includes the exact JSON-LD block that would be injected. A missing_faq_block recommendation specifies the FAQ structure to add. Typed recommendations mean the deployment layer does not need to guess what to do — it executes a specific, pre-validated intervention.
Connected to deployment
Recommendations produced by the Optimization Engine flow into the Automated Deployment pipeline. Agencies review recommendations before deployment — no change is applied without approval. The deployment pipeline then packages the approved change into a signed manifest, applies it via the Standard Script runtime, and runs independent verification to confirm the change landed as intended.
What Optimization Engine does not do
Clear boundaries increase trust. These limitations are documented explicitly, not buried in footnotes.
The Optimization Engine does not generate prose content — it proposes structural and schema changes, not copy rewrites.
Recommendations are not guarantees of AI engine visibility improvement — they address observable structural gaps; actual engine behavior is not controlled by GeoScript.
The engine does not crawl third-party sites or analyze backlinks — it reasons over the client's own Content Genome.
The six rules in v1 are the complete set — the engine does not apply rules beyond what is listed.
Recommendations require agency review and approval before deployment — the engine does not auto-apply changes.
Related
Other features
Content Genome
Private betaDeterministically extract typed content blocks, structured data, and FAQs from every page into a stable typed representation.
Standard Script
Private betaA lightweight, fail-open browser tag that fetches a signed deployment manifest and applies constrained typed-DSL changes to the DOM at runtime.
Automated Deployment
Private betaRisk-aware approval, signed packaging, preflight, activation, independent verification, and rollback for every deployed intervention.
GeoScript is in private beta
Agency partners get access to the full capability set — acquisition, monitoring, website intelligence, optimization, deployment, and reporting — for $299/month per active client.