"The Will Scale (W-Score): A Psychometrically Validated Index of Brand Visibility in Generative AI Engines (GEO) — Protocol, Open Dataset, and Preregistered Confirmatory Extensions"

Engineering notes

Engineering notes will be added by the aipentium editorial team.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为大语言模型、生成式AI、ChatGPT相关技术、计算机视觉、深度学习等高价值论文补充中文说明。

Original abstract

This registration provides a public, timestamped record of the protocol, instrument, and dataset of the Will Scale (W-Score), a composite formative index (0–100) of brand visibility in generative AI engines, and preregisters the confirmatory analyses of its next validation phase. The W-Score aggregates five axes — spontaneous mention (40%), position (20%), recommendation (15%), source diversity (15%), and cross-LLM consistency (10%) — and classifies brands into five interpretive bands (Invisible, Shadow, Supporting, Reference, Authority). The primary study audited 552 Brazilian brands (550 brand–segment pairs) across 22 market segments through 6,611 valid queries to four large language models (ChatGPT, Gemini, Perplexity, and Claude), using three category-neutral question templates per brand that never name the brand, with fixed model snapshots, registered seeds, and SHA-256 hashing of every raw response. Classic SEO indicators (Moz Domain Authority, Page Authority, backlinks, referring domains, estimated organic traffic, Lighthouse/CrUX) were measured in parallel as external criteria. IMPORTANT DISCLOSURE: data collection for the primary study was completed in June 2026, prior to this registration, and the primary validation battery has been analyzed and reported. This registration therefore serves two distinct purposes, kept explicitly separate throughout: (a) a transparent, timestamped archive of the protocol, instrument, and dataset of the completed study; and (b) a genuine preregistration — prior to any data analysis — of the confirmatory extensions specified in the Analysis Plan (PLS-SEM validation of the formative index, confirmatory factor analysis on the presence subset, reconciliation reanalysis of the diversity axis, and a second longitudinal collection wave).

Links and sources

Comments