--- title: "CatalogIQ Methodology: Open Catalog Quality Standards | CatalogIQ" description: The living, public methodology behind the CatalogIQ Index. Read it, evaluate it, comment on it. Version-tracked and open for review. image: https://magnetlabs.ai/hubfs/logos/catalogiq-logo-on-light.png --- [![CatalogIQ](https://magnetlabs.ai/hubfs/logos/catalogiq-logo-on-dark.png)](https://magnetlabs.ai/?hsLang=en) [Platform ▼](https://magnetlabs.ai/platform?hsLang=en) [The CatalogIQ IndexOne number, two sub-indices](https://magnetlabs.ai/platform?hsLang=en)[FrameworkTwo views from one crawl](https://magnetlabs.ai/platform/catalog-quality-framework?hsLang=en)[FeaturesWhat the score turns into](https://magnetlabs.ai/platform/features?hsLang=en)[RemediationsMeasurement first, then the fix](https://magnetlabs.ai/platform/catalog-enrichment?hsLang=en) [Catalogs ▼](https://magnetlabs.ai/industries/hvac-distribution?hsLang=en) [HVAC/R DistributionCatalog quality & regulatory readiness](https://magnetlabs.ai/industries/hvac-distribution?hsLang=en)[B2B Complex CatalogsDeep, multi-system catalogs](https://magnetlabs.ai/industries/b2b-distribution?hsLang=en)[B2C Consumer CatalogsReady for AI shopping agents](https://magnetlabs.ai/industries/b2c-retail?hsLang=en) [Methodology](https://magnetlabs.ai/methodology?hsLang=en)[Pricing](https://magnetlabs.ai/pricing?hsLang=en)[Company](https://magnetlabs.ai/company?hsLang=en) [Login](https://ciq.magnetlabs.ai) [Request your assessment](https://magnetlabs.ai/tools/catalog-scoring?hsLang=en) Methodology · v8.0 # How the score is built. The dimensions, the weights, and the standards we score against, published in full. Because independence only means something if it is auditable. [Read it in full →](https://magnetlabs.ai/methodology#model) [See the framework](https://magnetlabs.ai/platform/catalog-quality-framework?hsLang=en) The Index, at a glance 0 to 100 A · Catalog Quality and Discoverability8 dims · 60% B · AI Commerce Readiness4 dims · 40% Return Risk signalweight 0 Why it is public ## A private method makes "independent" a claim. A published one makes it a fact you can check. If we kept the scoring private, independence would be marketing. Publishing the dimensions, the weights, and the standards we score against is what lets an analyst, a buyer, or an AI agent verify the number instead of trusting it. The transparency gives the credibility needed to trust and defend the quality score. The model ## 13 dimensions, 12 weighted, one signal beside them. Twelve weighted dimensions resolve into the two sub-indices, blended into the headline at the default split of **60% Sub-index A and 40% Sub-index B**. Return Risk is reported beside the Index at weight zero. Weights below are the v8.0 published defaults; each column sums to 100% within its sub-index. DimensionSub-indexWt · of sub-index 01Attribute CompletenessSub-index A20% 02Attribute Value ConsistencySub-index A18% 03Classification CorrectnessSub-index A15% 04FindabilitySub-index A15% 05Specification AccuracySub-index A14% 06Technical DocumentationSub-index A8% 07DescriptionsSub-index A6% 08ImagerySub-index A4% 09Structured DataSub-index B30% 10Agent Required AttributesSub-index B28% 11Feed Protocol ComplianceSub-index B24% 12Natural-Language ParseabilitySub-index B18% 13Return-Risk Score derived signalReported besideweight 0 v8.0 published defaults. Weights are informed starting positions, review pending calibration against the scored corpus. Channels and protocols ## What Sub-index B is measured against. Sub-index B does not score in the abstract. Each of its dimensions is graded against the live requirements of the surfaces that now route machine-driven demand: the marketplace feeds a product must qualify for, and the agentic protocols an agent transacts through. Feeds ### Feed Protocol Compliance scores against the feeds The marketplace and shopping-feed requirements that decide eligibility and rank. We score your catalog against what each feed demands. Google Shopping Amazon Walmart Protocols ### Structured Data, Agent Required Attributes, and Natural-Language Parseability score agent readiness Readiness against the agentic commerce protocols that route machine-driven demand: the Agentic Commerce Protocol (ACP) behind ChatGPT checkout, and Google's Universal Commerce Protocol (UCP) behind AI Mode and Gemini. These specifications change constantly; the score reflects current requirements. ACP · ChatGPT checkout UCP · Google AI Mode, Gemini Standards ## What the score is built on. ![ETIM](https://magnetlabs.ai/hubfs/logos/std-etim.png) ETIM Classification standard for technical products, used in distribution. Leads on B2B and HVAC/R. ![ECLASS](https://magnetlabs.ai/hubfs/logos/std-eclass.png) ECLASS B2B classification and product-description standard for technical and industrial products. ![UNSPSC](https://magnetlabs.ai/hubfs/logos/std-unspsc.png) UNSPSC Global product and service classification standard. ![GS1](https://magnetlabs.ai/hubfs/logos/std-gs1.png) GS1 GDSN plus GS1 identifiers, including GTINs: the global standards for product data and identification. ![Google](https://magnetlabs.ai/hubfs/logos/std-google.png) Google The Google Product Taxonomy, the category structure for shopping surfaces. ![Amazon](https://magnetlabs.ai/hubfs/logos/std-amazon.png) Amazon Marketplace product-data requirements, leading on consumer catalogs. Confidence & limits ## Honesty about confidence is part of the design, not a caveat bolted on. The methodology grades the strength of evidence behind each measure and flags provisional ones until they are calibrated against a scored corpus. A score is always traceable to the evidence behind it. DIRECTMeasured straight from the catalog data. The strongest evidence class. ADJACENTInferred from closely related signals where a direct measure is not yet available. VENDORTaken from a vendor claim, disclosed as such and weighted accordingly. Versioning & governance ## Version-tracked, and open to challenge. A methodology you cannot track over time is one you cannot trust. Every change to the dimensions, the weights, the standards, or the rules is versioned and recorded in a public change log. If you think a weight is wrong or a standard is misapplied, there is a way to say so, and a process that acts on it. Change log v8.02026 · 03 · current Multi-standard waterfall and per-dimension standards priority. Data-driven channel definitions. The Regulatory Overlay introduced as a sibling system, with per-field provenance at the regulatory layer. Three measurement classes and the three-state missing-data treatment documented as shipped. v72025 · Q3 Sub-index B reduced to four dimensions (Data Freshness deferred under Principle Zero). The HVAC-tuned weight vector introduced. v62025 · Q1 The original five-dimension Sub-index B specification, the first published structure of the Index. Semantic versioning: major = a structural change to the Index, minor = a weight change within the structure, patch = a clarification. The version that produced a score travels with the score. The cadence comment feeds 01 **Submitted.** A challenge to a weight, a standard, or a dimension is logged against the current version. 02 **Reviewed.** Weights are reviewed at least annually against the scored corpus; substantive challenges are weighed in that cycle. 03 **Released.** An accepted change ships as a new version, recorded in the public change log with what changed and why. ### Challenge the methodology. Tell us which weight, standard, or dimension you would change, and the argument for it. Substantive challenges are reviewed in the calibration cycle and credited in the change log. [Submit a methodology challenge →](https://magnetlabs.ai/methodology/challenge?hsLang=en) [Full revision history Every version, with what changed and why. →](https://magnetlabs.ai/methodology/history?hsLang=en) ## Questions about the methodology. ### Why publish the methodology openly? Because independence only means something if it is auditable. If we kept the scoring private, "independent" would be a claim. Publishing the dimensions, the weights, and the standards we score against makes it a fact you can check. ### Can I comment on or challenge the methodology? Yes. The methodology is a living document, and active evaluations and comments are invited to shape the next standards release. If you think a dimension is weighted wrong or a standard is misapplied, that input is wanted, not just tolerated. ### How often does the methodology change? It is version-tracked, with a published version label and a change history. Changes are deliberate and documented rather than silent, so a score is always traceable to the version of the methodology that produced it. ### What standards does the methodology build on? The active standards are ETIM, ECLASS, UNSPSC, GS1, the Google Product Taxonomy, and Amazon's requirements. The methodology page documents how each is used and weighted. ### Is the score provisional or final? The methodology marks measures as provisional until they are calibrated against a scored corpus, and discloses the strength of evidence behind each. That honesty about confidence is part of the design, not a caveat bolted on. ## Read it, evaluate it, then see your own score. [Request your assessment →](https://magnetlabs.ai/tools/catalog-scoring?hsLang=en) [See the Index](https://magnetlabs.ai/platform?hsLang=en) [![CatalogIQ](https://magnetlabs.ai/hubfs/logos/catalogiq-logo-on-dark.png)](https://magnetlabs.ai/?hsLang=en) The independent measurement layer for B2B catalog quality. Built by Magnet Labs. [Request your assessment →](https://magnetlabs.ai/tools/catalog-scoring?hsLang=en) Platform [CatalogIQ Index](https://magnetlabs.ai/platform?hsLang=en)[Framework](https://magnetlabs.ai/platform/catalog-quality-framework?hsLang=en)[Features](https://magnetlabs.ai/platform/features?hsLang=en)[Remediations](https://magnetlabs.ai/platform/catalog-enrichment?hsLang=en) Catalogs [HVAC/R Distribution](https://magnetlabs.ai/industries/hvac-distribution?hsLang=en)[B2B Complex Catalogs](https://magnetlabs.ai/industries/b2b-distribution?hsLang=en)[B2C Consumer Catalogs](https://magnetlabs.ai/industries/b2c-retail?hsLang=en) Resources [Methodology](https://magnetlabs.ai/methodology?hsLang=en)[Glossary](https://magnetlabs.ai/glossary?hsLang=en)[Compare](https://magnetlabs.ai/compare?hsLang=en)[The Problem No One Owns](https://magnetlabs.ai/resources/catalog-quality-problem?hsLang=en)[Assessment kit](https://magnetlabs.ai/tools/catalog-assessment?hsLang=en)[Blog](https://magnetlabs.ai/beyond-the-catalog?hsLang=en) Company [About Magnet Labs](https://magnetlabs.ai/company?hsLang=en) [Contact us](https://magnetlabs.ai/contact?hsLang=en)[Trust & data handling](https://magnetlabs.ai/security/data-handling?hsLang=en)[Pricing](https://magnetlabs.ai/pricing?hsLang=en) © 2026 Magnet Labs. CatalogIQ is a product of Magnet Labs. [Privacy Policy](https://magnetlabs.ai/privacy-policy?hsLang=en)[Data handling](https://magnetlabs.ai/security/data-handling?hsLang=en) ```json { "@context" : "https://schema.org", "@type" : "Organization", "brand" : { "@type" : "Brand", "name" : "CatalogIQ" }, "founder" : { "@type" : "Person", "name" : "Rob Desmarais", "sameAs" : [ "https://www.linkedin.com/in/robdesmarais/" ] }, "foundingDate" : "2024", "logo" : "https://magnetlabs.ai/hubfs/logos/catalogiq-logo-on-light.png", "name" : "Magnet Labs", "sameAs" : [ "https://www.linkedin.com/company/magnet-labs-catalogiq/" ], "url" : "https://magnetlabs.ai/" } ``` ```json { "@context" : "https://schema.org", "@type" : "BreadcrumbList", "itemListElement" : [ { "@type" : "ListItem", "item" : "https://magnetlabs.ai/", "name" : "Home", "position" : 1 }, { "@type" : "ListItem", "name" : "Methodology", "position" : 2 } ] } ``` ```json { "@context" : "https://schema.org", "@type" : "TechArticle", "creator" : { "@type" : "Organization", "name" : "Magnet Labs" }, "dateModified" : "2026-06-28", "headline" : "The CatalogIQ Methodology", "inLanguage" : "en", "mentions" : [ { "@type" : "Thing", "name" : "ETIM" }, { "@type" : "Thing", "name" : "ECLASS" }, { "@type" : "Thing", "name" : "UNSPSC" }, { "@type" : "Thing", "name" : "GS1" }, { "@type" : "Thing", "name" : "Google Product Taxonomy" }, { "@type" : "Thing", "name" : "Amazon" } ], "proficiencyLevel" : "Expert", "version" : "8.0" } ``` ```json { "@context" : "https://schema.org", "@type" : "FAQPage", "mainEntity" : [ { "@type" : "Question", "acceptedAnswer" : { "@type" : "Answer", "text" : "Because independence only means something if it is auditable. If we kept the scoring private, \"independent\" would be a claim. Publishing the dimensions, the weights, and the standards we score against makes it a fact you can check." }, "name" : "Why publish the methodology openly?" }, { "@type" : "Question", "acceptedAnswer" : { "@type" : "Answer", "text" : "Yes. The methodology is a living document, and active evaluations and comments are invited to shape the next standards release. If you think a dimension is weighted wrong or a standard is misapplied, that input is wanted, not just tolerated." }, "name" : "Can I comment on or challenge the methodology?" }, { "@type" : "Question", "acceptedAnswer" : { "@type" : "Answer", "text" : "It is version-tracked, with a published version label and a change history. Changes are deliberate and documented rather than silent, so a score is always traceable to the version of the methodology that produced it." }, "name" : "How often does the methodology change?" }, { "@type" : "Question", "acceptedAnswer" : { "@type" : "Answer", "text" : "The active standards are ETIM, ECLASS, UNSPSC, GS1, the Google Product Taxonomy, and Amazon's requirements. The methodology page documents how each is used and weighted." }, "name" : "What standards does the methodology build on?" }, { "@type" : "Question", "acceptedAnswer" : { "@type" : "Answer", "text" : "The methodology marks measures as provisional until they are calibrated against a scored corpus, and discloses the strength of evidence behind each. That honesty about confidence is part of the design, not a caveat bolted on." }, "name" : "Is the score provisional or final?" } ] } ```