The Digital Craftsmanship Index
CraftScore™
Methodology
and Governance
The complete technical reference for the CraftScore™ framework — how the score is built, how each index is calculated, how evidence is classified, and how the benchmark is governed to remain fair, defensible, and evergreen.
-
01Executive SummaryWhat CraftScore™ is, why it exists, and what problem it solves
-
02Framework ArchitectureThe four-index structure, design principles, and weight rationale
-
03Core FormulaThe top-level CraftScore™ equation, fully annotated
-
04Index Deep-DivesO, S, V, and B indexes: sub-formulas, components, and signal definitions
-
05Companion ScoresPaid Amplification, Confidence Score, and Industry Percentile
-
06Tier SystemFive tiers, their boundaries, and what each signals for remediation priority
-
07Benchmark GovernanceEvidence hierarchy, recency windows, missing data protocol, anti-gaming rules
-
08Versioning ProtocolWhat is frozen, what updates, and how the framework stays evergreen
-
09Worked ExampleFull CraftScore™ calculation for a real brand, step by step
-
10Limitations and Calibration RoadmapKnown constraints of v1.0 and the path to global benchmark calibration
01 — Executive Summary
A benchmark the world
can trust and verify.
CraftScore™ is a 0–100 digital health index designed to measure a brand’s complete digital ecosystem in a single, comparable number. Unlike single-metric tools such as Domain Authority or social follower counts, CraftScore™ evaluates four interconnected dimensions: Owned Media performance, Shared Media authority, Search and AI Visibility, and Brand Authority.
The framework was developed from the Digital Craftsmanship Series — three volumes of practitioner methodology covering Shared, Owned, and Paid media. CraftScore™ is the measurable output of that methodology: a score that any brand, in any industry, can receive and act on.
Why CraftScore™ was needed
The digital marketing industry suffers from metric fragmentation. Brands track Domain Authority in isolation, social followers as a proxy for influence, and paid ROAS as a proxy for marketing health. None of these metrics reflects the compound, systemic health of a brand’s digital presence. None is cross-comparable across industries. None accounts for the emerging AI discovery layer where ChatGPT, Gemini, and Perplexity now influence purchasing decisions before a user ever clicks a search result.
CraftScore™ fills this gap with a single, weighted, auditable composite score. One number. Four dimensions. Five tiers. A governance layer that prevents gaming and ensures the score means the same thing for a restaurant in Kuala Lumpur as it does for a SaaS company in Singapore.
What this document covers
This white paper is the complete technical reference for CraftScore™ v1.0. It covers every formula, every weight, every signal definition, and every governance rule in full. The methodology is published openly because a benchmark that cannot be audited cannot be trusted.
Openness Principle
CraftScore™ operates as an open methodology benchmark. The formulas, weights, and governance rules published in this document are the actual system used in all audits. No proprietary black-box scoring. If a brand receives a CraftScore™, it can read this document and understand exactly why.
02 — Framework Architecture
Built on four dimensions.
Frozen at the foundation.
CraftScore™ is structured as a three-layer system. The core formula is frozen. Companion scores provide context. A governance layer ensures integrity. Each layer serves a distinct purpose and none can substitute for another.
Design Principles
Four principles governed every architectural decision in the CraftScore™ framework:
- Consistent: The same formula applies to every brand, regardless of industry, size, or geography. Context is added via companion scores and percentile layers, never by changing the core formula.
- Computable: Every input is observable or inferable from external, publicly accessible signals. Audits do not require internal access. A brand’s CraftScore™ can be computed independently by any qualified auditor using this document.
- Causal: The four indexes are ordered by causality, not by importance alone. O (Owned) is the foundation. S (Shared) and V (Visibility) are growth drivers that feed off the foundation. B (Brand Authority) is the lagging compounding result. Fix in order O, V, S, B and the whole score rises.
- Evergreen: The architecture absorbs platform-level change without structural revision. AEO and GEO are defined abstractly. New AI platforms slot into existing categories. Sub-signals update annually. The core formula does not change.
The Three-Layer Architecture
| Layer | Components | Purpose |
|---|---|---|
| Layer 1: Core Score | O, S, V, B indexes | Measures permanent digital health. Cross-comparable. 0–100. |
| Layer 2: Companion Scores | Paid Amplification, Confidence Score | Never blended into CraftScore. Adds commercial context and data quality transparency. |
| Layer 3: Benchmark Context | Industry, Stage, Region percentiles | Positions the absolute score relative to comparable peers. Grows in accuracy as the audit database scales. |
Weight Rationale
The top-level weights are not arbitrary. Each reflects the causality and controllability of that dimension in the digital ecosystem:
| Index | Weight | Rationale |
|---|---|---|
| O — Owned Media | 35% | Highest weight because the brand controls it entirely. Technical or content failures here suppress every other index. Fix O first; V and B improve automatically. |
| S — Shared Media | 25% | Equal to V because social and PR signals feed search visibility. A brand with strong S but weak O is building authority on sand. |
| V — Visibility | 25% | Equal to S because search visibility in turn drives brand demand and authority. V and S are interdependent growth drivers. |
| B — Brand Authority | 15% | Lowest weight because it is a lagging indicator. Brand authority rises as a consequence of fixing O, S, and V — it cannot be engineered directly in isolation. |
Important Note on Paid Media
Paid media is a companion score (Paid Amplification), not a component of the core CraftScore™. This is a deliberate architectural decision: CraftScore™ measures permanent, compounding digital health. Paid activity is budget-dependent and temporary. A brand spending heavily on ads but with a broken owned media foundation is not healthy — their score should reflect that. The Paid Amplification Score exists to reward brands who amplify a strong organic foundation, not to mask structural weakness.
03 — Core Formula
The equation.
Fully annotated.
The CraftScore™ top-level formula is a weighted composite of four sub-index scores. All sub-indexes are normalised to a 0–100 scale before weighting. The composite is reported as an integer; the decimal is retained internally for calculation precision.
CraftScore™ Core Equation · v1.0 · Frozen
Variable Definitions
| Variable | Full Name | Weight | Scale |
|---|---|---|---|
| O | Owned Media Performance Index | 0.35 (35%) | 0–100 |
| S | Shared Media Authority Index | 0.25 (25%) | 0–100 |
| V | Search and AI Visibility Index | 0.25 (25%) | 0–100 |
| B | Brand Authority Index | 0.15 (15%) | 0–100 |
Weight Verification
All weights sum to 1.00: 0.35 + 0.25 + 0.25 + 0.15 = 1.00. This is a requirement of the framework. If any future revision alters the weights, all four must still sum to exactly 1.00 and a published rationale must accompany the change.
04 — Index Deep-Dives
Every component.
Every signal. Explained.
Each index is a weighted composite of sub-components. All sub-components are scored 0–100 and blended with defined weights. The sub-formulas are frozen at v1.0; the signals within each component are versioned annually.
Purpose: Measures how well a brand controls and optimises its own digital real estate — website architecture, content infrastructure, technical health, CRM, and user experience. A failing O-Index suppresses every other score because owned media is the destination all other channels point to.
O-Index Sub-Formula · v1.0
Sub-Component Definitions
| Component | Weight | What is scored |
|---|---|---|
| Tech SEO | 25% | HTTPS status, robots.txt health, XML sitemap presence and validity, JavaScript rendering behaviour, Core Web Vitals, canonical tag implementation, internal linking architecture, structured data / schema presence, mobile-first indexing compliance, crawl accessibility. |
| Content / AEO Foundation | 25% | Blog or content hub presence, FAQ content depth, on-page optimisation quality, FAQ/Article schema implementation, ability to answer key high-intent questions, content freshness signals, topical authority coverage. |
| UX | 20% | Mobile responsiveness and usability, navigation clarity, page speed, form and conversion pathway friction, accessibility basics, JavaScript-first architecture issues, dead-end pages. |
| GEO Readiness | 20% | Presence of AI-friendly “answer assets” — structured guides, location pages, how-to content, brand story pages, loyalty or service guides. Content that enables AI platforms to cite the brand accurately. |
| Email / CRM | 10% | Email capture presence, list segmentation signals, observable lifecycle flow indicators, estimated engagement benchmarks. Weight capped at 10% because this component is externally unobservable without client access; higher weight would allow estimated data to inflate scores. |
Purpose: Measures how actively and effectively a brand earns attention in external channels it does not fully own. Social, PR, reviews, and community are the four pillars of earned authority. Strong S-Index signals without strong O-Index creates a house with no foundation.
S-Index Sub-Formula · v1.0
Sub-Component Definitions
| Component | Weight | What is scored |
|---|---|---|
| Social Health | 30% | Audience size relative to category, posting frequency, content mix (educational vs promotional), engagement rate, platform diversification, social-to-site linking quality. |
| PR / Citations | 25% | Volume and quality (Domain Authority) of earned media placements, recency of press features, presence in authoritative local and industry publications, citation in relevant directories. |
| Reviews | 25% | Average rating across platforms (Google, TripAdvisor, delivery apps, Facebook, App Store), review volume, response rate and quality, trend direction (improving vs declining). |
| Community | 20% | User-generated content (UGC) depth and quality, influencer footprint and structure, ambassador programme presence, local community integration (student, corporate, area partnerships). |
Purpose: Measures how discoverable a brand is across the complete visibility landscape — traditional search engines, AI-embedded search features (AEO), and standalone generative AI platforms (GEO). As AI reshapes discovery, V-Index becomes increasingly predictive of future demand.
V-Index Sub-Formula · v1.0
Critical Signal Separation Rule
AEO and GEO are distinct surfaces. A signal cannot be counted in both. A brand appearing in Google AI Overviews earns AEO credit only. The same brand appearing in ChatGPT earns GEO credit only. This prevents double-counting visibility and ensures each component measures a genuinely separate discovery channel.
Sub-Component Definitions
| Component | Weight | Signal Definition |
|---|---|---|
| SEO Rankings | 35% | Branded vs non-branded traffic mix, ranking positions for key category/location/FAQ terms, organic traffic share vs competitors, click-through rate signals, position improvement trend. |
| AEO Presence | 35% | Definition: Answer visibility across AI-assisted search surfaces. Current signals include: Google AI Overviews (SGE), Featured Snippets, People Also Ask appearances, Bing Copilot citations. Future signals will be added to this bucket as search engines embed new AI answer formats. |
| GEO Presence | 30% | Definition: Answer visibility across standalone generative AI platforms. Current signals include: ChatGPT responses, Google Gemini answers, Perplexity citations, Claude responses, voice assistant mentions. Future AI platforms are added to this bucket as they achieve mainstream usage. |
Evergreen Note on AEO and GEO
AEO and GEO are framework categories, not platform lists. The definitions above describe the category abstractly (AI-assisted search surfaces vs standalone AI platforms). The specific tools cited (Google AI Overviews, ChatGPT, etc.) are current examples within each category. As the platform landscape evolves, examples update inside the same category structure. The framework does not need to change; only the Signal Pack annotation updates annually.
Purpose: Measures the trust, credibility, and entity strength a brand carries across the web. B-Index is intentionally weighted lowest because it is a lagging indicator — it rises as a consequence of sustained excellence in O, S, and V. It also functions as a feedback loop: strong brand authority improves CTR, drives organic search demand, and increases conversion rates independently of direct marketing activity.
B-Index Sub-Formula · v1.0
Sub-Component Definitions
| Component | Weight | What is scored |
|---|---|---|
| Backlinks | 30% | Domain Authority of referring domains, number of unique referring domains, presence of high-DA editorial links, link velocity trend, directory and industry vertical representation. |
| Brand Entity | 30% | Wikipedia depth and quality, knowledge panel signals, structured schema clarity (who, what, where in machine-readable form), Google Knowledge Graph entity recognition, third-party authoritative brand pages. |
| Sentiment | 25% | Ratio of positive, neutral, and negative public mentions across monitored channels, nature of sentiment drivers (quality vs service vs price vs values), trend direction, crisis or reputational incident signals. |
| Competitive Position | 15% | Share of voice vs primary competitors in search and social, ability to own or defend a category keyword or concept, distinctiveness of brand position in competitive landscape. |
05 — Companion Scores
Context that never
distorts the benchmark.
Three companion scores appear on every CraftScore™ report. None of them blend into the core score. Each adds a different layer of honesty and context without compromising the integrity of the benchmark number itself.
Paid Amplification Score (P)
Measures how effectively a brand uses paid media to amplify its existing organic and owned presence. Paid activity cannot compensate for a weak O or V score — it can only amplify what is already working.
Paid Amplification Formula
| Component | Weight | What is scored |
|---|---|---|
| Ad Presence + Spend Signals | 40% | Observable paid presence across search, social, and display. Estimated spend signals from tools like SimilarWeb or SemRush. Platform diversification. |
| Landing Page / CRO Quality | 35% | Dedicated landing page presence, page speed, conversion element quality, form friction, trust signals, mobile experience. |
| Audience Targeting Sophistication | 25% | Retargeting pixel presence, custom audience indicators, lookalike signal quality, funnel stage differentiation observable in ad creative. |
Confidence Score
Every CraftScore™ report discloses what percentage of its inputs come from directly observable signals (Class A) versus inferred or self-reported signals (Class B and C). This is the Confidence Score.
| Evidence Class | Definition | Trust Weight Applied |
|---|---|---|
| Class A: Observable | Publicly accessible, auditable in real time | Full weight (100%) |
| Class B: Inferred | Derived from proxies without direct access | 80% weight applied |
| Class C: Self-Reported | Provided directly by the brand being audited | 60% weight applied, flagged in report |
| Confidence Band | Meaning | Report Treatment |
|---|---|---|
| 80–100% | High confidence | Score presented as reliable and actionable |
| 50–79% | Moderate confidence | Score presented as directional; key assumptions noted |
| Below 50% | Low confidence | Score marked indicative only; full audit recommended |
Industry Percentile
The raw CraftScore™ is an absolute measure (0–100) that is cross-comparable regardless of industry. The percentile is additive context that tells a brand where they rank within their peer group.
Three percentile dimensions appear on every full report:
- Industry Percentile: brand score vs audited brands in the same category (F&B, SaaS, B2B, retail, etc.)
- Stage Percentile: brand score vs audited brands at the same business maturity stage (Early / Growth / Mature / Enterprise)
- Region Percentile: brand score vs audited brands in the same primary market
Calibration Note
Percentile data is populated as the CraftScan audit database accumulates brand assessments. Version 1.0 launches with Southeast Asia as the reference market. Global percentiles will be added once the 500-brand milestone is reached, providing statistically reliable category benchmarks.
06 — Tier System
Five tiers.
Each with a distinct prescription.
Once calculated, a CraftScore™ maps to one of five tiers. Tier boundaries are frozen at v1.0 and will be empirically reviewed at the 500-brand calibration milestone. Each tier signals a specific health state and carries specific remediation priorities.
| Score | Tier Name | Health State | Primary Action |
|---|---|---|---|
| 85–100 | Digital Masterwork | Category leader. Compounding digital advantage active across all four indexes. | Maintain. Invest in competitive moat and V-Index dominance. |
| 70–84 | Digital Craftsman | Strong foundation. Structural integrity across O and V. Ready for category dominance. | Optimise the weakest index. Scale what is working. Begin AEO/GEO investment. |
| 55–69 | Momentum Builder | Active across channels but structurally incomplete. Some indexes performing; others dragging. | Identify the lowest-scoring index. Fix structural gaps before scaling paid spend. |
| 35–54 | Presence Alert | Foundational gaps across multiple channels. Visible but structurally weak. | Fix O-Index first. Then V-Index. Do not invest in paid until foundation is stable. |
| 0–34 | Digital Dark Mode | Near-invisible to search, AI platforms, and potential customers. | Emergency rebuild of owned media foundation. Pause all paid activity until O reaches 50+. |
Structural Flag Rules
In addition to the tier system, CraftScore™ applies conditional structural flags when a single top-level index drops below 15/100. These flags are additive — they appear alongside the composite score and tier, never replacing them.
| Condition | Flag Applied | Meaning |
|---|---|---|
| Index below 15 AND Confidence above 79% | Structurally Compromised | The composite score is masking a critical failure. The flagged index requires emergency remediation regardless of the overall score tier. |
| Index below 15 AND Confidence below 80% | Data Incomplete | Insufficient data to score this index reliably. A full audit with client data access is recommended before acting on this index’s sub-score. |
07 — Benchmark Governance
The rules that make it
a standard, not just a score.
A formula without governance is a scoring tool. A formula with governance is a standard. The CraftScore™ governance layer defines how evidence is classified, how the score expires, how missing data is treated, and how gaming is prevented. These rules are published in full and applied uniformly to every audit.
Governance Rule 1: Evidence Hierarchy
Every input used to calculate a CraftScore™ is classified before it is weighted. The evidence class determines the trust weight applied to that input’s contribution to the score.
| Class | Definition | Examples | Trust Weight |
|---|---|---|---|
| Class A Observable |
Publicly accessible, auditable in real time by any qualified auditor | Website status, schema presence, Google rankings, social posting frequency, review ratings, PR placements | 100% |
| Class B Inferred |
Derived from patterns and proxies without direct access to the system | Estimated organic traffic, ad spend signals, email capture flow inference, audience size estimation | 80% |
| Class C Self-Reported |
Provided directly by the brand being audited; cannot be independently verified | Internal analytics data, CRM open rates, conversion rates, internal revenue attribution | 60% + flagged |
Governance Rule 2: Score Validity Windows
CraftScore™ is a point-in-time measurement. Digital health changes continuously. Scores expire after the following windows, measured from the audit date. An expired score must display “Score Expired” in any public or commercial use.
| Index / Score | Maximum Validity | Rationale |
|---|---|---|
| V-Index (Visibility) | 90 days | Algorithm updates and AI platform changes make visibility signals the fastest-changing dimension |
| S-Index (Shared Media) | 120 days | Social and PR signals evolve moderately fast; seasonal and campaign effects mean quarterly review is standard |
| O-Index (Owned Media) | 180 days | Site architecture and content changes are typically slower; 6 months is a reasonable validity window |
| B-Index (Brand Authority) | 180 days | Backlinks and brand entity signals change slowly under normal conditions |
| Full CraftScore™ | 90 days | The shortest index window governs the full score. A score is only as current as its most volatile component |
Governance Rule 3: Missing Data Protocol
| Scenario | Treatment |
|---|---|
| Component fully unobservable | Score at category median; flag as estimated; reduce Confidence Score proportionally |
| Fewer than 3 signals available for a sub-index | Sub-index flagged as Low Signal; capped at 60/100 maximum regardless of available inputs |
| Entire index unscoreable | CraftScore™ marked Incomplete — no composite number issued until the gap is resolved |
Governance Rule 4: Anti-Gaming Rules
CraftScore™ measures durable health, not activity volume. The following behaviours are explicitly identified as gaming and are scored accordingly:
- Publishing content with no indexation or engagement signal: scored at 0 for content quality components
- Acquiring backlinks from low-DA, non-thematic, or link-farm domains: counted at zero for Backlinks component
- Running paid campaigns against a broken O or V foundation: Paid Amplification Score is separate and cannot lift CraftScore
- Buying followers or reviews: anomalous velocity spikes (follower or review count growing faster than engagement or brand search) are flagged and the inflated sub-score is discounted to organic baseline
- Artificially inflated signal detected during audit: scored at 0 for that component and documented in the report
Governance Rule 5: Cross-Brand Comparability
The raw CraftScore™ is an absolute measure on a 0–100 scale. This means a score of 60 means the same thing whether the brand is a restaurant, a SaaS company, or a B2B manufacturer. Context is not added by changing the formula — it is added by the percentile layer.
This is the same model used by credit rating agencies and academic testing: the underlying score is universal; the interpretation is contextualised by peer comparison. It ensures that CraftScore™ remains a true benchmark rather than a relative ranking that shifts based on who is being compared.
Source Hierarchy
When conflicting signals are found across sources, the following hierarchy determines which is used: (1) Direct platform data where accessible, (2) Third-party audit tools (SemRush, Ahrefs, SimilarWeb for relevant metrics), (3) Manual observation from the auditor, (4) Brand-provided data (Class C, weighted accordingly). The source used for each scored component is documented in the full audit report.
08 — Versioning Protocol
Frozen where it matters.
Adaptive where it must be.
The CraftScore™ versioning protocol is what makes the framework evergreen. The core architecture is permanent. The sub-signals that feed into it update annually to reflect platform changes, new AI surfaces, and evolving digital best practices.
| Layer | Version Status | Update Rule |
|---|---|---|
| Top-level weights (O/S/V/B) | Frozen — v1.0 | Only changed with published rationale and 6-month advance notice. Requires empirical calibration data from 500+ brand audits before any revision is considered. |
| Sub-index formulas | Frozen — v1.0 | Same rule as top-level weights. Sub-formulas cannot change without published rationale and notice. |
| Sub-component signals | Versioned annually | Updated each January. Published as “CraftScore™ Signal Pack vX.0 — [Year]”. New platforms, deprecated signals, and signal weight adjustments within each component are covered here. |
| Tier boundaries | Frozen — v1.0 | Reviewed at the 500-brand calibration milestone. Will only change if empirical score distribution data shows current boundaries are systematically mis-calibrated. |
| Governance rules | Living document | Updated as edge cases are encountered and resolved. All changes logged with date and rationale in a public changelog at iamjaychong.com/craftscore-governance. |
Version Labelling
Every CraftScore™ report must display the version under which it was calculated:
- Core framework: “CraftScore™ v1.0”
- Signal pack: “Signal Pack v1.0 — 2026”
- Combined display: “CraftScore™ v1.0 · 2026 Signals”
09 — Worked Example
Nando’s Malaysia:
A full CraftScore™ calculation.
This section walks through a complete CraftScore™ calculation for a real-world brand. Every step is shown to demonstrate how the formula translates observable signals into an actionable benchmark number.
Brand Overview
Nando’s Malaysia operates as a well-known international F&B chain with significant Facebook audience, observable paid activity, and a generally positive sentiment profile. The audit was conducted using publicly accessible signals only (Class A and B), resulting in a Confidence Score of 82%.
Step 1: Score Each Sub-Component
O-Index Components
| Component | Weight | Score | Weighted Contribution | Key Signal |
|---|---|---|---|---|
| Tech SEO | 25% | 22 | 5.5 | robots.txt timeout, JS-only ordering pages, zero schema, no Core Web Vitals pass |
| Content / AEO Foundation | 25% | 35 | 8.75 | No blog hub, no FAQ schema, limited on-page content depth |
| UX | 20% | 55 | 11.0 | Mobile responsive but JS-first architecture creates ordering friction |
| GEO Readiness | 20% | 40 | 8.0 | No structured answer assets, limited city/location pages |
| Email / CRM | 10% | 50 | 5.0 | PERi-Perks programme observable; lifecycle flows unclear |
| O-Index Total | 38 | 38.25 | ||
S-Index Components
| Component | Weight | Score | Weighted Contribution | Key Signal |
|---|---|---|---|---|
| Social Health | 30% | 72 | 21.6 | Strong Facebook audience, acceptable engagement rate, consistent posting |
| PR / Citations | 25% | 45 | 11.25 | Brand mentions present but limited high-DA editorial features recently |
| Reviews | 25% | 58 | 14.5 | Ratings generally positive; low response rate to negative reviews |
| Community | 20% | 48 | 9.6 | UGC present but no structured influencer programme observable |
| S-Index Total | 56 | 56.95 | ||
V-Index Components
| Component | Weight | Score | Weighted Contribution | Key Signal |
|---|---|---|---|---|
| SEO Rankings | 35% | 42 | 14.7 | High reliance on branded terms, minimal non-branded category ranking |
| AEO Presence | 35% | 18 | 6.3 | Zero Featured Snippets, not appearing in AI Overviews for category queries |
| GEO Presence | 30% | 30 | 9.0 | Partial chatbot mentions; cited only in branded queries, not category queries |
| V-Index Total | 31 | 30.0 | ||
B-Index Components
| Component | Weight | Score | Weighted Contribution | Key Signal |
|---|---|---|---|---|
| Backlinks | 30% | 45 | 13.5 | Moderate DA referring domains; limited recent high-DA editorial links |
| Brand Entity | 30% | 65 | 19.5 | Strong global Wikipedia anchor; knowledge panel present |
| Sentiment | 25% | 60 | 15.0 | Predominantly positive; flavour and service are primary positive drivers |
| Competitive Position | 15% | 40 | 6.0 | Peri-peri category partially owned; competitors gaining in non-branded search |
| B-Index Total | 52 | 54.0 | ||
Step 2: Apply Core Formula
Calculation — Nando’s Malaysia
Step 3: Apply Structural Flags
No index falls below 15/100. V-Index at 31 is low but above threshold. No structural flag applied. The low V score is noted in Top 3 Actions.
Step 4: Canonical Report Output
10 — Limitations and Calibration Roadmap
What v1.0 is. What it
is not yet.
Intellectual honesty is a component of the CraftScore™ standard. This section documents the known limitations of v1.0 and the path to resolving them.
Known Limitations of v1.0
- Founder-designed weights: The current weights are intellectually justified and logically consistent but have not yet been validated against empirical outcome data. They represent the best available practitioner judgment for v1.0. They will be tested against real business outcomes (revenue correlation, brand search growth, organic traffic growth) as the audit database scales.
- Southeast Asia reference market: Percentile benchmarks at launch are calibrated to Southeast Asia. Applying them globally introduces regional bias until a sufficient cross-regional dataset is built.
- Class B and C input reliance: Several sub-components — particularly Email/CRM and some GEO signals — are partially inferred rather than directly observable. The Confidence Score addresses this transparently, but it remains a constraint of external-only audits.
- Tier boundary calibration: The five tier boundaries (85, 70, 55, 35) are set by judgment at v1.0. They may not perfectly reflect the actual score distribution across a global brand dataset.
Calibration Roadmap
| Milestone | Action | Expected Outcome |
|---|---|---|
| 100 audits | First percentile baseline established for F&B and SaaS in SEA | More accurate Industry Percentile for key categories |
| 250 audits | Outcome correlation study: CraftScore vs organic traffic growth, brand search volume | First empirical validation of weight rationale |
| 500 audits | Full tier boundary calibration review; expand to global percentile benchmarks | CraftScore™ v1.1 tier adjustments if data warrants |
| 1,000 audits | Industry-specific signal weighting analysis; consider industry modifier layer | Potential introduction of optional industry lenses for sub-component weighting |
The Benchmark Promise
CraftScore™ v1.0 is strong enough to launch, use in commercial reports, and sell as a concept. It is not yet globally calibrated. The difference between a scoring tool and a true global benchmark is the calibration database. That database is built one audit at a time, and every report generated by CraftScan contributes to it. The methodology is sound. The standard is being built transparently, in public, from the first audit onward.