Can our internal audit team review the methodology?

Yes, under a structured diligence process. Public materials stay at architecture altitude. Deeper methodology discussion is available for qualified opportunities under NDA, without transferring source IP.

Will competitors reverse-engineer RankCraft from this page?

No. This page intentionally excludes weights, prompts, thresholds, code paths, and aggregation recipes. It explains the business system, not the secret sauce.

Your tenant configuration, product inputs, and resulting monitoring records remain yours under contract. Platform IP — engines, models, and methodology — remains Mercury Technology Solutions intellectual property.

Can RankCraft run inside our environment?

Enterprise programs support cloud SaaS plus private-cloud and on-premise deployment paths where required by policy.

Is Mercury's GEO Audit a black box?

No. RankCraft GEO publishes measurement categories, data sources, access controls, security posture, and remediation language. What stays unpublished is the proprietary scoring layer — weights, prompts, and aggregation methods — the same way a credit model is governed without printing the formula on a marketing site.

How is a GEO score calculated at a high level?

Configured AI engines are queried on a controlled schedule. Responses are normalized, attributed, and scored under dual methodologies: site readiness (Method A) and citation potential (Method B), plus citation presence, share of voice, framing/safety, and trend signals. Exact weights and calibration curves remain Mercury IP and are discussed under NDA when required.

What is the difference between GEO methodology and proprietary scoring?

Methodology is the process: scope, observe, normalize, score, act, re-measure. Proprietary scoring is the encapsulated math inside that process. Mercury's GEO Audit methodology page explains the dual-methodology architecture; this transparency page explains the disclosure boundary around that architecture.

Can we attach this page to a vendor-risk or procurement review?

Yes. This page is written as a public leave-behind for executives, risk, and procurement. Pair it with the GEO Audit service page and, for shortlisted deals, request the NDA diligence pack covering architecture, security questionnaires, and data handling.

Enterprise Transparency

Updated August 2026

Not a black box. A governed measurement system.

Executives should understand how RankCraft measures AI visibility — without competitors receiving a free blueprint of the Mercury GEO Engine. This page draws that line clearly.

Free AI-visibility audit See the disclosure boundary

Definition

What controlled transparency means

Controlled transparency is Mercury's public policy for RankCraft GEO audits: buyers can inspect what is measured, where the data comes from, who can see results, and how findings are remediated. The proprietary scoring layer — dimension weights, prompts, and aggregation recipes — is not published, because publishing it would transfer competitive advantage rather than increase trust.

Read the GEO Audit methodology · See the GEO Audit service

Key takeaways

What executives should remember

RankCraft is a governed GEO measurement system, not a mystery score for the board.
Open by design: categories, data sources, presentation, access control, security, and remediation language.
Protected IP: exact weights, prompts, thresholds, heuristics, source code, and R&D corpora.
Trust comes from lineage, repeatability, and actionability — the same standard used for enterprise risk models.
Deeper methodology is available under NDA for qualified diligence; public pages stay at architecture altitude.

The executive concern

Trust requires clarity. Competition requires protection.

When a platform scores brand visibility across ChatGPT, Perplexity, Gemini, and other AI engines, buyers rightly ask: what is measured, can the process be audited, or is this a mystery score for the board?

RankCraft answers with controlled transparency: open enough for governance, closed enough to protect intellectual property and your competitive edge. Read the GEO Audit methodology.

For your board

Explainable categories, data lineage, and decision rights — not a mystery number.

For your teams

Clear inputs, outputs, and recommended actions they can operationalize.

For risk & compliance

Tenant isolation, RBAC, encryption posture, and deployment options you can review.

Against competitors

Scoring weights, agent logic, and ranking heuristics remain proprietary IP.

Disclosure boundary

We share structure and governance. We protect the methods that create durable advantage.

Open by design

What enterprise buyers, auditors, and operators can inspect.

What is measured: citations, share of voice, readiness dimensions, trend signals
Where data comes from: configured AI engines, your domains, approved product inputs
How results are presented: dashboards, exports, alerts, executive summaries
Who can see what: tenant isolation, role-based access, product scoping
Security & deployment options: SaaS, private cloud, on-premise paths
Remediation language: findings mapped to practical fix categories

Protected IP

Not published publicly. Available only under NDA / enterprise diligence where required.

Exact scoring formulas, dimension weights, and calibration curves
Agent prompts, routing rules, and internal decision thresholds
Query-generation heuristics and competitive-aggregation methods
Model fine-tuning corpora and proprietary feature engineering
Source code, infrastructure internals, and client-specific tuning packs
Raw third-party engine response corpora used for platform R&D

Why the line is drawn here

Anyone can sample AI answers. The durable value is how those answers are normalized, scored across dual methodologies, correlated across engines, and turned into prioritized action — refined across enterprise deployments. Publishing that layer would not make you safer; it would make your competitors faster.

What RankCraft measures

Categories you can put on a slide. Not the proprietary math behind each score.

AI citation presence

Whether your brand, products, or owned sources appear when target queries are asked across monitored engines.

Share of voice

Relative mention frequency versus named competitors within defined query sets and markets.

Site readiness (Method A)

Six high-level dimensions covering technical accessibility, content clarity, structure, and related readiness factors.

Citation potential (Method B)

Whether content is structured to be retrievable and selectable by generative engines — reported as outcome bands, not secret weights.

Brand framing & safety signals

Misstatements, outdated product framing, and adverse narrative patterns surfaced for review.

Trend & anomaly signals

Direction of travel, volatility, and unusual movement that warrant operator attention.

How the system works

A process view executives can follow. Implementation detail stays inside Mercury Technology Solutions.

01
Define scope
Products, markets, languages, competitors, and priority query themes are configured for your tenant.
02
Observe engines
Configured AI engines are queried on a controlled schedule. Responses are captured with run metadata for auditability.
03
Normalize & score
Responses are cleaned, attributed, and scored under dual methodologies. Proprietary steps are encapsulated as services.
04
Act & re-measure
Teams receive prioritized findings and recommendations, then track movement over time.

Architecture at executive altitude

Enough structure for diligence. Not a threat model for reverse-engineering.

Public / safe to share

Experience layer

Executive & operator dashboards
Alerts & exports
Role-aware views

Control plane

Auth & tenant resolution
RBAC & product scope
API contracts

Intelligence core

Mercury GEO Engine U2
Monitoring jobs
Forecasting services

Internals of the intelligence core are proprietary and not disclosed on this page.

Data plane

Tenant-scoped business data
Job coordination store
Per-tenant analysis graphs

Evidence without oversharing

You should not need source code to trust an enterprise measurement system. You need lineage, repeatability, and accountability.

Traceable runs

Each monitoring cycle can be tied to query set, engine, time window, and tenant scope — so results are discussable, not mystical.

Repeatable process

The same configured scope produces comparable outputs over time, enabling trend reviews and before/after remediation tracking.

Human-readable findings

Scores are paired with category-level explanations and recommended next steps your teams can own.

Security & control

Your data boundary is part of the product

Transparency is incomplete without tenancy and access clarity. RankCraft is multi-tenant with product-level scoping and role-based controls.

Tenant isolation for business records and operator access
Role model spanning admin, analyst, and viewer patterns
Encryption in transit; encryption at rest in supported deployments
Enterprise deployment paths: cloud SaaS, private cloud, on-premise options
No requirement to publish proprietary engine internals to operate securely

What a diligence pack typically includes

Architecture briefing
Layered system overview at the same altitude as this page.
Security questionnaire support
Enterprise vendor-risk responses.
Data handling summary
What is stored, retained, tenant-scoped, or external.
Optional NDA deep-dive
More methodology for shortlisted deals only; still not full IP transfer.

FAQ

Executive FAQ

Trust comes from defined measurement categories, repeatable runs, tenant-scoped evidence, and actionability — the same standard used for many enterprise risk and credit models. You validate outcomes against your domain reality; you do not need the formula printed on a public webpage.

Keep reading GEO vs SEO · Trust Architecture · Get cited by AI · Glossary

Clear enough for the boardroom. Closed enough to stay defensible.

RankCraft is not a mystery score. It is a governed GEO measurement system — transparent in process, protected in method, accountable in outcomes.

Book executive briefing Request a GEO Audit

Mercury Technology Solutions · RankCraft GEO · Confidential methodology protected as proprietary IP

GEO programmes from HK$6,000/mo · 6–12 month engagements · Senior-led, no juniors · See GEO tiers (T1–T5)

Not for businesses looking for a brochure website. We build AI-era revenue infrastructure for enterprises where pipeline is measured, not impressions.

Written by James Huang, Founder & CEO, Mercury Technology Solutions · Reviewed by the Mercury GAIO Practice · Updated August 2026