
What happened
Branding Works added a feature to its LLMO measurement tool that lists, per measured question, the search terms ChatGPT actually used before answering; the demo screen is free and needs no registration.
Why it matters
Since search terms steer which pages an answer is built on, the list gives companies a hint at which wording makes their own pages easier for ChatGPT to pick up.
WHO IT HITSSEO and AI-search-optimization staff at companies that want to know how ChatGPT finds and describes them can use the query lists to decide which wording and pages to prepare.
Summaries like this, every morning.
Branding Works, a Chiba-based company that provides SEO and AI search optimization services, built its LLMO measurement tool to track how AI describes a company. The new fan-out query view extends that from the output — how a company is mentioned — back to the input, namely the search terms ChatGPT assembles from a question and then uses to gather material.
The company's own measurements illustrate why it treats that step as worth exposing. Across 39 measurement days between 2026年9月11日 and 2026年10月26日, covering 17 question types about where to hire SEO and AI search help, ChatGPT searched before all 1,990 answers, using one search term per answer. Repeat the same question and the term changes: queries born from a single question averaged 63 kinds over 39 days.
ChatGPT also adds words the question never contained. Asked for affordable, capable SEO firms, it searched 「日本 格安 SEO 会社 評判」, adding "評判"; searches with words absent from the question totaled 132. Branding Works reads this as the AI looking for reputation and reviews to check capability — an inference on its part. About 10% of queries against Japanese questions (190) were run in English, which is why the company suggests listing English names for the company and its services.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
Edge Sound Research announced a $2 million round led by Stormbreaker Ventures, bringing total funding to $9 mi…
Nvidia committed $3 billion to Meta's equity offering, part of Meta's plan to raise up to $95 billion, and the…

US investment-grade bond sales were on track to reach $1.76 trillion once deals priced on October 26, 2026, to…

Crunchbase data shows global robotics venture funding reached $9.9 billion in Q3, more than double the $4.3 bi…

In its IPO filing cover letter, Anthropic wrote that Claude's "only incentive" should be giving a helpful answ…

OpenAI says its models proved the Exponential Time Hypothesis false, proved the local smoothing conjecture in…
