LLMOcheki launches analysis of what your AI-cited pages have in common

Itera, Inc. (head office: Shinjuku-ku, Tokyo; Representative Director: Takayuki Muto) has begun offering, in LLMOcheki, its tool for measuring and improving LLMO, an analysis of the characteristics of your own pages that AI has cited, together with an analysis of citation tendencies engine by engine.

Background: do cited pages have anything in common?

As AI search comes into wider use, being introduced and cited in an AI’s answer has become a new route to customers. Yet what makes a page likely to be cited has remained a matter of rules of thumb and generalities.

The selection algorithms of each AI are not published, so no definitive answer is available. What is possible is to gather the pages of your own that were actually cited and look for what they share — to check whether there is a structural difference, within your own site, between the pages being chosen and the pages that are not. This release makes that check something you can run on your own data.

Feature 1: characteristics of your cited pages

The pages of yours that AI actually cited are extracted, and their characteristics are listed: whether structured data such as FAQ is present, how many headings there are, how much body text.

Because these are shown alongside the averages for the site as a whole, you can see whether the cited pages are skewed in some way. If citations cluster on pages carrying structured data, for instance, rolling that out to other pages becomes the next thing to do.

The check runs on your own data rather than on generalities, so it is not thrown off by differences in industry or site structure.

Feature 2: citation tendencies by AI engine

For each AI engine, the make-up of the sources cited (your own, competitors, third parties), the types of source most often cited (comparison sites, reviews, Q&A, video, social and so on, in sixteen categories), the leading cited domains, and the share of UGC and social media are displayed.

Engines differ in the kinds of source they refer to. For an engine where you are seldom introduced, you can consider a direction for your work in terms of building exposure in the kinds of source that engine refers to most.

Feature 3: automatic recording of article audit scores

Each time an article is generated or saved, an audit of whether its structure is easy for AI to read is run automatically and the score recorded. The criteria are fixed, so the same article always produces the same score.

Matching the accumulated scores against actual citations lets you keep testing what form of article tends to be cited.

On reading these results

What this feature shows is an observation based on our own measurement data. It does not reveal the selection criteria of any AI, nor does it show that a causal relationship exists between the characteristics displayed and citation.

It is offered as a clue for deciding what to work on.

What comes next

We will continue to extend the features that support the whole path from measurement to improvement, including judging the effect of work carried out and reflecting it in improvement proposals, and presenting steps appropriate to the stage a business is at.

About LLMOcheki

LLMOcheki is a tool for measuring and improving LLMO. It measures daily how your brand is introduced and cited across eight AI search engines including ChatGPT, Gemini and Perplexity, and covers analysis, article creation and verification of results in a single tool.

LLMOcheki https://llmocheki.com/