Share of Voice: an own-brand study of name ambiguity Published 13 October 2026 by Superstellar LLC, operator of Share of Voice https://shareofvoice.app/blog/ai-brand-name-matching-study SCOPE This report measured our product brand, Share of Voice. Superstellar LLC is the company that operates the product; its legal name was not the measured brand. This is our own product’s report, not another customer’s report. Collected on 10 October 2026, 11:51:46.925–12:41:39.917 UTC. Market: Global, not Switzerland. Category: ai answer visibility monitoring. Audience: marketing and PR teams. 26 question groups, two phrasings, three repeats, four model APIs = 624 answers. All answers have collection status ok. Only 623 have a saved classification. Original prompts and model output are English. Article editions are available in English, German, French and Italian; the data is the same in each edition. FILES answers.json: original prompts and all answer texts, timestamps, requested and resolved models, requested provider settings, saved brand attribution flags. answer-counts.csv: one row per answer; deterministic counts and saved flags. recount.py: Python 3 script using only the standard library, with no network or AI calls. Save beside answers.json, then run: python3 recount.py answers.json EXPECTED TOTALS 624 collected; 623 analyzed; 260 answers containing the phrase (41.7% of 624); 385 occurrences; zero brand mentions attributed by saved automated analysis; zero answer texts containing the literal domain shareofvoice.app. One unclassified answer: q4-1-ChatGPT-1. Each provider has 156 collected answers: ChatGPT: 155 analyzed, 10 phrase matches; openai/gpt-5-mini Gemini: 156 analyzed, 98 phrase matches; google/gemini-3-flash Claude: 156 analyzed, 126 phrase matches; anthropic/claude-sonnet-4.6 Perplexity: 156 analyzed, 26 phrase matches; perplexity/sonar COUNTING RULES Case-insensitive regex: \bshare\s+of\s+voice\b A phrase match means at least one occurrence in the text, not evidence of brand identity. Repeat mentions in one answer count once for answer-level rates. brandNamedBySavedAnalysis is the saved automated attribution, NOT a manual reference label. null in JSON / unknown in CSV means unclassified, not negative. The CSV's phrasing and repeat indices are zero-based. The analysis denominator is 623, while the literal-text denominator is all 624 collected answers. No new classification was generated for this publication. LIMITATIONS - We sell AI visibility reports; this is not independent product validation. - Legacy buyer-panel-v4 used overlapping needs and generic comparison dimensions. This question panel is not validated as representative of buyer demand. - Single-turn API output with provider-specific search is not personalized consumer-app behavior. Requested word/search limits are not proof of compliance. - We did not manually adjudicate all 260 matches. They are not 260 proven errors. - This snapshot does not measure Swiss demand, universal brand absence, audience market share, provider quality, or the causal effect of any SEO changes. - Zero domain matches refers only to answer text. Attached citation metadata includes opaque redirects and was not analyzed for a domain-absence claim. - Vendor claims, pricing, sources and recommendations in archived model output are historical and unverified. They are evidence of output, not our advice. PRIVACY AND PROVENANCE This extract excludes billing, private access links, usage/cost records, search logs, attached citation metadata and the action plan. It contains no other customer's report. Text is preserved verbatim, including provider attribution. The CSV is derived from this extract, not independent validation of its labels. ARCHIVE CHECKSUM (SHA-256 of answers.json) 4773125f6a07f36cd30f0108ed4284fd3f87c2e645714604ca2540c42f1f6d11 EDITORIAL CORRECTION — 13 October 2026 Earlier wording called the measured product brand “our company”. The report measured Share of Voice, not Superstellar LLC. Answers, prompts, saved analysis flags, counts and the archive checksum are unchanged.