Data
Every observation, recommendation, and outcome lives in a structured, tenant-isolated, immutable-where-it-matters data architecture — the evidence base the entire platform reasons over.
Four data planes
GeoScript's evidence base is organized around four data planes: the observation plane (raw AI engine responses, preserved before parsing), the content plane (the Content Genome — structured extraction from a client's website), the deployment plane (signed intervention packages, preflight records, verification results), and the measurement plane (before-and-after visibility attribution, report snapshots). These planes are kept separate — each is immutable where it matters and feeds the next stage in the product loop.
Tenant isolation: the foundational rule
Every data record in GeoScript is owned by a specific agency workspace. The platform enforces tenant isolation at the service layer, not just the query layer — a workspace cannot access another workspace's observations, clients, opportunities, or reports. This is enforced at the database policy level and is treated as a Tier 1 invariant: code that violates tenant isolation does not ship.
Immutability where it matters
AI visibility observations are immutable — once recorded, a raw response cannot be altered. Report snapshots are immutable — once published, the numbers are frozen. Opportunity score components are versioned — formula changes produce new version records, not overwrites. This design means the data underlying a report or pitch room accurately reflects what was true at the time it was recorded, not a retroactively adjusted version.
Determinism: the observable behavior standard
GeoScript's data processes are deterministic wherever possible — the same inputs produce the same outputs. The prompt universe generator produces the same 1,118 prompts for the same client inputs every time. The Content Genome extractor produces the same typed blocks from the same page content every time. The Optimization Engine produces the same recommendations from the same genome every time. This makes the system auditable: unexpected outputs point to changed inputs, not random behavior.
Methodology transparency
GeoScript publishes the methodology behind its scored metrics and data processes. The Opportunity Score methodology documents every component, its weight, the missing-evidence penalty, and the formula version. The data freshness methodology documents how timestamps are assigned, how stale evidence is handled, and why published report snapshots are isolated from live data updates. The methodology section is kept current as the platform evolves.
What Data is not
Explicit limitations (§127)
- ✕GeoScript does not sell, license, or share client observation data with third parties.
- ✕GeoScript does not use one client's data to benefit another client's recommendations or scores.
- ✕GeoScript does not claim perfect causal attribution — the measurement plane records before-and-after differences; causal inference requires acknowledging confounders.
- ✕GeoScript does not store or process personally identifiable information about a client's end-customers — observations are about AI engine responses, not individual consumer data.
- ✕GeoScript is not a general-purpose data warehouse or analytics platform — the data architecture is purpose-built for the AI visibility measurement and optimization loop.
Related
Other pillars
Monitor how AI engines respond to the queries that matter for your clients — capturing mentions, recommendations, and citations across a prompt universe built around each business.
Find businesses with measurable AI visibility gaps, score each opportunity, track your pipeline, and deliver evidence-backed pitches — before a competitor does.
GeoScript crawls a client's public website, extracts a structured content evidence base, and builds the Content Genome that powers every recommendation.
Understand which AI crawlers and agents are reaching a client's site, what they're doing, and why — with verified identity and declared purpose.
Turn Content Genome findings into specific, typed recommendations — then deploy them through an approval-gated, verifiable system that leaves an immutable record.
Immutable, period-scoped reports that capture AI visibility change, deployments made, and measured outcomes — shareable with clients and presentable under agency branding.