NEWS

News

Press release

LLMOcheki launches analysis of what AI Overview is talking about, and where only competitors are used as the source

Itera, Inc. (head office: Shinjuku-ku, Tokyo; Representative Director: Takayuki Muto) released two features in LLMOcheki, its tool for measuring and improving LLMO, on 13 October 2026: AIO Topics, which extracts the subjects Google AI Overview covers and compares whether you or your competitors are cited on them, and Competitor Sentiment Score, which compares the tone in which AI answers speak about you and about your competitors.

Subjects are extracted automatically from the body of an AI Overview, and for each subject you can see who is being used as the source. Both are included within existing plans at no additional cost.

Background: the question that comes after “are we showing up?”

An AI-generated summary, AI Overview, now appears at the top of Google results more and more often. It summarises the content of several sites and shows links to the sites it used as sources. Users frequently resolve their question from the summary alone, so search rank no longer tells you whether your information is reaching anyone.

LLMOcheki has until now measured, keyword by keyword, whether an AI Overview appeared and whether you were cited in it. But knowing you were not cited does not tell you what to write next, because you cannot see what the AI Overview is saying about that keyword.

This release goes further, into the substance of the summary, its sources, and the tone of the answer.

Feature 1: AIO Topics

The body of each AI Overview shown for a registered keyword is saved at the time of the daily rank measurement, and the subjects corresponding to its headings are extracted and aggregated.

The subject list shows, for each subject, how often it appears (the proportion of measurements with an AI Overview that included it), the change between the last seven days and the seven before, and a priority that factors in monthly search volume. A high priority means a subject that AI Overview returns to repeatedly and that also has substantial search demand.

The coverage table shows, for each keyword, how many days your own site and each registered competitor site appeared among the sources of the AI Overview. A keyword where you are absent and only a competitor appears is one where only that competitor is being used as the basis for the subject.

When writing an article, this tells you whether your own pages cover the subjects AI Overview is addressing, and which subjects rest on competitors alone. For each keyword you can also see when a subject first and last appeared, how the cited domains break down, and an excerpt of the latest body text.

Feature 2: Competitor Sentiment Score

For your brand and your registered competitors as they appear in AI answers, the tone of the answer is judged as positive, neutral or negative, and compared on a scale from minus one to one across four axes: informational, comparative, problem and purchase.

The score is the number of positive answers minus the number of negative answers, divided by the number of answers in which the brand was mentioned. One means every mention was positive; minus one means every mention was negative.

An axis where your score is lower than a competitor’s is one where AI introduces you less favourably for that kind of question. A low score on the problem axis points to what is being described as your weakness; a low score on the purchase axis points to how price or adoption is being explained. Either can be examined in the list of answers and acted on.

Note that this judges the tone of an answer, not whether its content is factually correct. Where the number of answers mentioning a brand is small, the score is swayed by a handful of them.

Also added

  • Competitor mention column: how many engines a registered competitor’s name appeared on when the keyword was put to AI
  • Citation share by domain over time: the daily trend in the share of AI answers citing your domain and the most-cited domains
  • Citation rate split by brand mention: answers that mentioned your brand and answers that did not, each with the rate at which your URL was cited
  • AI Overview appearance and trend: columns for appearance rate and your citation rate per keyword, and a day, week or month toggle on the appearance trend
  • Reputation breakdown: the tone of answers mentioning your brand, by observation axis and question type

Availability

  • Released: 13 October 2026
  • For: customers on a plan that includes LLMOcheki’s AI search features
  • Cost: no additional charge

AIO Topics and competitor sentiment accumulate from measurement data recorded on or after the release date.

Comment from our Representative Director

“Showing someone that they are not cited in AI Overview does not tell them what to write next. What is the AI saying about that keyword, and who is being used as the basis for it? Without breaking it down that far, it does not connect to the work of writing an article. Finding the subjects where only a competitor is used as the source is, I think, the most practical part of this.”

About LLMOcheki

LLMOcheki is a Japanese-built tool for measuring and improving LLMO. It measures daily how much your brand is introduced in AI search on ChatGPT, Google and elsewhere, and carries through to the improvements that get you introduced. It covers measurement of mentions and citations in AI answers, analysis of cited sources, rank measurement including AI Overview, and support for writing articles AI is more likely to cite.

It puts the same question repeatedly and records the result together with its statistical spread.

LLMOcheki https://llmocheki.com/

Originally published at prtimes.jp