Consumers' information-discovery journey is shifting from traditional search engines to AI answers. Generative AI synthesizes multiple sources into a single recommended answer. Today, customers ask AI directly which product is best, and accept the brands that AI mentions and recommends as their candidates.
Taking the CEP (Category Entry Point) as its starting point, AI Optimizer queries AI directly with the questions real customers are likely to ask. It then analyzes those answers and delivers the following:
Brand & competitor visibility diagnosis — Analyzes which brands appear in AI answers and what AI used as its basis (citations).
Owned-media improvement direction — Recommends the information structure with which you should reinforce the pages you directly control (homepage, detail pages) so that AI calls on them.
Earned-media improvement direction — Guides you on the reasons to believe (RTB) you should establish across external media and communities that you cannot directly control.
To begin an analysis, enter three things.
Select a CEP project — Choose an existing CEP project, or create a new one with
Start a new CEP project. You can preview an example viaView sample projecton the right.Enter your brand & product information — Enter the brand name or product name that will serve as the basis for analysis. The country (language) automatically follows the CEP you selected.
Enter your content URLs — Enter URLs for owned media (your pages) and earned media (YouTube, external media, etc.). You can enter up to 10, and you may also skip this.
Once you finish entering and click Run, the analysis begins.
Note — Branching by whether URLs are entered If you enter your content URLs, the AI Response Analysis Expert (with URLs) runs, diagnosing the gap against your pages as well. Conversely, if you do not enter URLs, there is no comparison target (your pages), so the AI Response Analysis Expert (without URLs) runs, dissecting the structure of the AI answer itself and the brand-occupancy landscape without comparing your pages. In both cases the agent name is identical; only the display condition differs.
Based on the CEP, Nano Intent (detailed context), and KBF (Key Buying Factors) within the CEP project, AI Optimizer automatically generates the prompts (questions) to be posed to AI. On this screen, you review the questions generated for each CEP.
When the analysis finishes, the full Overview screen appears first. Before drilling into the detailed metrics of an individual CEP, this is the stage where you grasp the project's overall opportunity landscape and attack priorities on a single screen. The screen is composed of two areas, left and right.
Left — GEO map and input information — Along with the GEO map (4-quadrant chart), it displays the brand & product information and the list of content URLs that served as the basis for analysis, in the order they were entered.
Right — CEP card list — Presents each CEP as a card. Cards are grouped by the four sections of the GEO map and sorted in priority order (1st–4th).
The map, the core of GEO analysis, plots each CEP (consumer context) on two axes: CEP Interest (X-axis) and AI Recall (Y-axis). Using 50 points, the midpoint of both axes, as the boundary, it is divided into four sections (quadrants), and the brand response strategy differs by section.
Section | Interest · Recall | Status | Strategy |
|---|---|---|---|
Top-Priority Opportunity | Interest high · Recall low | User search demand (CEP Interest) is high, but within AI answers your brand's AI Recall and citation rate are low. | Strengthening GEO is urgent. Produce comparison content centered on Key Buying Factors (KBF), and apply structured data such as FAQ and product schema markup so that AI engines better learn and cite your content. |
Core Competition | Interest high · Recall high | Both CEP Interest and AI recommendation/citation of your brand are high — a section where dual occupancy is complete. | Defense and expansion are key. Continuously monitor competitors' citation patterns, expand your occupancy into adjacent CEPs, and cultivate brand advocates. |
Niche Strength | Interest low · Recall high | AI strongly recommends your brand, but consumer search demand for this context itself is low. | Conversion strengthening and awareness expansion are needed. Imprint brand assets (DBA) through broad-reach campaigns and optimize the funnel from search to purchase. |
Untapped Territory | Interest low · Recall low | Both CEP Interest and AI Recall are low. | This is a proactive-investment section. Publish foundational content that defines the category, secure long-tail SEO, and build a long-term base by entering communities and the like. |
Each CEP's position is determined by its CEP Interest (X-axis) and AI Recall (Y-axis) scores. The specific scoring system (how coordinates are calculated) is explained in the CEP Detailed Analysis dashboard.
Each card on the right consists of a CEP chip, the CEP phrase, and two bar metrics.
AI Recall — A bar accumulating your brand's mention score and your content's citation score. (e.g., mention 40 + citation 50 = 90 points) The sum of the two scores becomes the Y-axis coordinate on the GEO map.
CEP Interest — A bar representing the search demand for that CEP. The CEP with the largest total search volume within the project is set to 100%, and the rest are shown as relative ratios; this value becomes the X-axis coordinate on the GEO map.
Clicking a card or the button on the right takes you to that CEP's detailed analysis screen. A detailed explanation of how metrics are calculated can be found in the detailed analysis dashboard.
Clicking a specific CEP card in the full Overview opens that CEP's detailed analysis screen. The detailed analysis is composed of two parts.
AI Response Analysis — Review detailed information such as the score, mentions, citations, and full response text for the CEP you clicked.
AI Agent — Generates a concrete strategy guide based on the analysis results.
The prompt and AI response analysis for the selected CEP are expanded. It is organized into sub-tabs.
Summarizes the core metrics of the analysis, such as your content & product information, AI response scores, and CEP Interest.
① AI Recall Diagnosis (Y-axis score)
A gauge that aggregates "how well AI calls on your brand" on a scale of 0–100. Depending on the score, it is displayed in five grades — Insufficient, Lacking, Fair, Excellent, and Outstanding — and this score becomes the Y-axis coordinate on the GEO map. The score ranges for each grade are shown on the scale at the top of the gauge, and the detailed evaluation criteria can be checked in the Scoring Criteria beside it.
② Mention & Citation Detail Scores
The AI Recall in ① is the sum of the two scores below, each weighted half (50% each).
Your brand mention score — Looks at how prominently your brand appears in the AI answer. It is graded in stages, from Preferred Recommendation (recommended first) through mentions as a comparison target or candidate, down to Not Mentioned.
Your content citation score — Looks at how well the sources AI brought in as the basis for its answer match your actual URLs. It is graded in three levels: Content Citation (an exact match with a registered URL) / Domain Citation (a loose match at the same-domain level) / Not Cited.
The higher both scores are together, the more it means AI trusts your brand and actively recommends it. The criteria for assigning each grade are organized under Scoring Criteria on the screen.
③ CEP Interest (X-axis score)
Represents "how often people search for this question." It is the sum of the search volumes of all search terms related to that CEP. Along with a search-volume trend chart, it presents the number of related keywords and total monthly average search volume, plus a table of search volume per individual keyword on the right, so you can directly verify the demand that underlies the score. However, because search volumes differ enormously by keyword (from hundreds to hundreds of thousands), the values are gently compressed so that smaller values are not buried by larger ones, then the CEP with the greatest demand is set to 100 points and the smallest to 0 points — a score lined up so they can be compared with one another. This score becomes the X-axis coordinate on the GEO map.
Compares how many times and in what context your brand/product and competitors' were each mentioned in the AI answer. It shows the mention count per response (Response 1, 2, 3), your share, and your weight relative to the next-ranked brand, so you can grasp at a glance how prominently you appear within the answer.
Organizes the external domains AI used as the basis for composing its answer, along with the cited URLs of your content. In a table, you can check which domains were cited how many times per response, and at which level your content (URL) was captured — Content Citation or Domain Citation.
A simple mention indicates whether a brand appeared in the answer, while a citation indicates whether AI adopted that source as its basis. You need to look at both metrics together to accurately judge the quality of your visibility.
Review the full text of AI's response to each prompt as is. You can directly examine the actual answer content that served as the basis for the analysis metrics.
Shows the related keywords explored together with that CEP, along with their monthly average search volume. Use this to set priorities when planning content.
This is a strategy analyst that converts analysis results directly into actionable content and marketing guides. On the CEP detail screen, clicking a specific CEP activates the AI Response Analysis Expert, and follow-up buttons at the bottom of its response carry on into the detailed owned-media and earned-media (reasons-to-believe media) strategies.
When you click a CEP, it operates. It splits into two modes depending on whether your content URLs are entered (same name, only the display condition differs); in either mode it delivers response diagnosis, owned media, and earned media (reasons-to-believe media) as one integrated briefing, after which follow-up buttons at the bottom of the response continue into the detailed strategies.
AI Response Analysis Expert — With URLs (includes gap diagnosis)
Diagnoses why your pages were not sufficiently surfaced in the AI answer and carries it into reinforcement directions. It bundles a response & entity gap diagnosis (topic groups A/B/C), owned-media information-structure design, and reasons-to-believe (earned) media design — showing what you are missing and what competitors occupy, with owned/earned reinforcement directions aligned in the same context.
AI Response Analysis Expert — Without URLs (no comparison of your pages)
Without your URLs, diagnoses the response structure, brand occupancy, and citation sources from the AI response alone and forms an entry hypothesis. It points out structural entry positions, the reasons-to-believe you need, and the external media to secure, then designs a new owned-media information structure and earned-media signals. Even with no pages of your own, it presents the direction of an entry strategy. (Replaces the gap-diagnosis mode above.)
Below the AI Response Analysis Expert's integrated briefing, two follow-up analysis buttons appear. Clicking each button designs a detailed, channel-by-channel execution strategy for the single CEP you selected.
Owned Strategy — Reshapes or newly designs the channels you directly control (product detail pages (PDP), category pages (PLP), brand pages, blogs/guides, FAQs, product structured data) into an official, AI-readable reference information structure. It presents CEP-KBF-RTB reference-information design, content blocks and per-channel revision plans, a technical strategy, and an execution priority (Now · Next · Later). If you entered your URLs it guides a "revision direction" that refines existing pages; if not, a "creation direction" that builds from scratch.
Earned Strategy — Designs what needs to be confirmed in the external channels you do not directly control (reviews, communities, expert evaluations, creators, distribution platforms, press/PR, social). It organizes the channels to check and the conditions to confirm so that external trust signals point in the same direction as your owned media's official reference information.