Prompt Research from ThriveStack citedby finds the prompts buyers ask in AI search

Prompt Research in ThriveStack citedby, shown with sample data for a made-up brand, Acme Corp.

SEATTLE, Wash. – November 02, 2026 – Prompt Research, a new ThriveStack citedby feature launched today, builds AI search prompt panels from 12 brand signals so marketers can measure how often AI engines name their brand.

AI engines like ChatGPT, Google Gemini, Google AI Mode, Claude and others publish no query data, so brands have to guess which questions to track, and the same question can return a different brand list on every run. Prompt Research gives brands a prompt panel: a fixed set of real buyer questions run on a schedule, used to track share of voice and other marketing metrics in AI search.

Prompt Research in ThriveStack citedby, shown with sample data for a made-up brand, Acme Corp.

“An AI visibility score is only as good as the prompts that track it,” said Gururaj Pandurangi, Founder and CEO of ThriveStack. “If you track vanity prompts, your dashboard says you are winning while buyers never see you. ThriveStack citedby’s Prompt Research picks the questions from your brand, and 11 other buyer signals.”

Prompt Research takes a brand kit, topics, personas, competitors, markets, languages and buyer intent in one pass. It returns a panel that users can edit before anything runs. Key details:

• Tagged by signal. Every prompt carries its topic, persona, intent, market, language and branded or unbranded tag, so results can be cut by any of them.
• Weighted by intent. The default mix is 20% informational, 40% commercial and 40% transactional prompts. Between 80% and 90% of prompts stay unbranded to measure discovery.
• Sized by formula. A startup panel runs about 45 prompts. A growing brand runs about 360. An enterprise panel can pass 10,000.
• Run on your schedule. Each prompt runs on every AI engine at least once a week, up to daily. University of St. Gallen researchers found a single AI run has a standard error of 0.370 on brand detection. Seven runs bring it under 0.10, so results are reported as a range over several weeks.

Choosing prompts to track in ThriveStack citedby. Each question arrives tagged by persona, branding, intent and country. Sample data.

Mention rate, share of voice and citation rate are reported by engine, persona, intent, market and others. A team can track, for example, that it wins buyer prompts in the U.S. and does not appear in Germany. Brands and agencies can use the same panel for AI search tracking, also called AEO (answer engine optimization) or GEO (generative engine optimization).

Prompt Research is available today in all ThriveStack citedby plans. The full method, including all 12 signals and sizing by company stage, is at thrivestack.ai/citedby/features/prompt-research.

About ThriveStack

ThriveStack citedby is an AI visibility tool for brands and agencies that measures citation gaps across 10+ AI engines, fixes them and attributes revenue to AI search. It is a product of ThriveStack Corporation, based in Seattle, Washington, and led by Founder and CEO Gururaj Pandurangi (ex-Microsoft, Oracle, Zscaler), a three-time founder with two exits. Learn more at thrivestack.ai/citedby.

Media Contact
Company Name: Thrivestack corporation
Contact Person: Gururaj Pandurangi
Email: Send Email
State: WA
Country: United States
Website: https://www.thrivestack.ai/citedby