GEO Studio measures two layers. AI Visibility is what models say about a brand. Content evaluation is whether a URL is fit to be retrieved and cited. Hover a column in the product for the one-line definition; this page is the full catalog.
Every scored metric shown in GEO Studio.
AI Visibility
Scores from scheduled prompt runs. They describe how a brand shows up in AI answers — not how a page ranks in classic search.
Visibility
Also called Visibility Index · avgVisibilityScore
0–100Higher is better
How often and how well the brand appears in AI answers (0–100).
Critical
< 40
Warning
40–59
Good
60–79
Excellent
80–100
How it’s calculated
75% mention quality + 25% citation quality, each scaled by √(appearance rate) so a one-off spike ranks below reliable presence. Missing components count as 0.
When it moves
Headline ranking for topics, prompts, brands, and competitors. A drop is usually absence or weak presence — not a classic search rank change. Split mention vs citation score to see which side moved.
In product · KPI tiles, Topics, Prompts, Brands, Competitors
How positively the brand is spoken about when AI mentions it (0–100).
Negative
< 30
Mixed
30–69
Positive
70–100
How it’s calculated
Average sentiment of the brand’s mentions in finished runs. Empty when the brand was not named.
When it moves
Negative sentiment with high mention volume is worse than being absent — the model is teaching a weak story. Empty means not mentioned, not a zero score.
In product · KPI tiles, Topics, Prompts, Brands, Answers log
The brand’s share of AI answers that name any in-scope brand in this topic (0–100).
How it’s calculated
This brand’s answer-appearances in the topic ÷ all in-scope brands’ answer-appearances, counting each brand once per answer. Context brands (ChatGPT, Google) are excluded from the denominator.
When it moves
Visibility can look healthy while share falls because more brands entered the answers. The denominator also grows with prompt volume and discovery, so read it next to the topic leader, not as a standalone percentage.
How well the brand is named in the answer — role, placement, prominence, and sentiment (0–100).
Critical
< 40
Warning
40–59
Good
60–79
Excellent
80–100
How it’s calculated
Composite of how the brand is referenced when named, averaged across finished runs.
When it moves
High mention count with a low score means you are named as an afterthought. Mentions outweigh citations in Visibility (~3:1), so this often moves the headline even without a link.
Includes
RolePlacementProminenceSentimentScope of coverage
In product · KPI tiles, Topics, Prompts, prompt detail overlays
Also called Citation Score · Citation quality · avgCitationScore
0–100Higher is better
How well a linked URL supports the AI answer (0–100). Different from how often it is cited, or where it ranks.
Critical
< 40
Warning
40–59
Good
60–79
Excellent
80–100
How it’s calculated
Per-citation quality for linked URLs — claim importance and how prominently the citation is shown — averaged for the brand or URL. Cited-pages tables use avgCitationScore; topics and prompts use avgCitationQualityScore.
When it moves
Volume without quality is a vanity count. Low score on an owned URL usually means the wrong page or a weak passage was selected.
Average rank in the AI answer. Lower is better; empty means not cited.
How it’s calculated
Mean citation position across finished runs. Null when there are no citations — not position 0.
When it moves
Rising (worse) position with stable citation count means you are still linked but no longer leading. Read next to Citation Quality: a weak #1 is worse than a strong #3.
Count of brand mentions in finished runs in the date range.
When it moves
Use this to see whether you are in the conversation at all, including answers with no URL. Interpret with Mention Quality and Sentiment — a negative spike is not a win.
How many times a URL was linked as a source in AI answers.
How it’s calculated
Count of citations in finished runs in the date range.
When it moves
If mentions rise and citations stay flat, the model knows the brand but will not attach a URL — usually a citable-passage problem on the page you want linked.
8–10 maintain, 6–7.9 targeted improvements, 4–5.9 moderate rewrite, below 4 major rewrite or rethink the page’s role. Strong health does not guarantee prompt Visibility — the page may never be selected for that query.
In product · Evaluated Pages, page detail, Optimize queue
How precisely each section addresses its intended topic. Earlier sections carry more weight.
Irrelevant
< 4
Tangential
4–5.9
Supportive
6–7.9
Essential
8–10
How it’s calculated
Mean of chunk weighted scores. Each chunk is scored on specificity, completeness, accuracy, clarity, and purpose (0–1), scaled to 0–10, then weighted so earlier sections count more.
When it moves
Low precision with decent recall means generic or rambling sections — a chunk-retrieval and citation-safety problem. Tighten passages so each chunk makes one concrete claim.
Includes
SpecificityCompletenessAccuracyClarityPurpose
In product · Evaluated Pages, page detail, Copywriting headline bar
How comprehensively the page covers its intended concept, calibrated to page intent.
Poor
< 0.3
Weak
0.3–0.59
Moderate
0.6–0.89
Excellent
0.9–1.0
How it’s calculated
Average of breadth, depth, coherence, completeness, and authority. A product page is not judged like a long guide. Shown as a percentage in some tables.
When it moves
Low recall with decent precision means missing subtopics or thin coverage. Expand structured sections. Below ~0.7 the evaluator typically flags gaps to fill.
In product · Evaluated Pages, page detail, Copywriting headline bar
Also called GSC Query Alignment · GSC alignment · Search demand · GSC
0–10Higher is better
How well the page answers real queries from connected Google Search Console data.
Not answered
< 3
Partial
3–6.9
Well answered
7–8.9
Thorough
9–10
How it’s calculated
Mean of per-question scores (0–10) derived from top GSC queries. Unavailable until Search Console is connected.
When it moves
High Semantic Recall but low Questions Answered means you cover the topic abstractly and miss how people actually ask. Align headings and FAQs to query language.
In product · Evaluated Pages, Copywriting headline bar
How distinctive the page is versus top-ranking competitors.
Commoditized
< 5
Moderate
5–6.9
Differentiated
7–10
How it’s calculated
Mean of the per-competitor uniqueness scores against top-ranking Google results for the topic. The evaluator is asked for a low score where overlap is high and differentiated value is weak; the consistency rules in code run the other way, trimming a competitor's overlap score to fit a high uniqueness score (≥ 9 caps overlap at 3, ≥ 7 caps it at 4).
When it moves
Acceptable precision and recall with low uniqueness means you are structurally fine but interchangeable. Add proprietary proof, data, or workflow — not more generic coverage.
In product · Evaluated Pages, Copywriting headline bar
Also called Update need · Topical opportunity · Update
0–10Higher = more urgent
Refresh urgency — higher means the page needs attention sooner (0–10).
Minimal
0–2
Watch
3–6
Elevated
7–8
Act now
9–10
How it’s calculated
Combines topic-trend momentum, how stale the page is, and QDF (how fast users expect updates). 9–10 is a hot topic with stale content and high freshness demand.
When it moves
Inverted vs most scores: high is a call to act, not a quality win. Pair with Precision/Recall so you refresh the right sections, not just touch the date.
In product · Evaluated Pages, Optimize queue, Copywriting headline bar
How time-sensitive the page’s queries are — not how fresh the page is today.
Timeless
0–1
Stable
2–3
Fast-moving
4–5
How it’s calculated
Mean of per-query QDF scores from top GSC queries. 0 is timeless; 5 means users expect updates about every 30 days.
When it moves
High Content Freshness with a stale page drives Topical Opportunity. Low Content Freshness means an old page can still be healthy. Do not treat a high Content Freshness score as a content-quality failure.
In product · Page evaluation detail (feeds Topical Opportunity and Evergreen Health)
Extra fields the evaluation pipeline produces. They explain a score; they are not the headline numbers in tables and KPI tiles.
Field
Group
Range
Meaning
Unprompted brand mention
Page-eval brand monitoring
0–100%
Share of models that spontaneously cite your brand when given the topic only — not your name. Different from AI Visibility Mention Score (scheduled prompt runs).
Brand-aware sentiment
Page-eval brand monitoring
−1 to +1
Sentiment when models are given your brand name on this page’s topic. Separate from the 0–100 Sentiment column on prompt tables.
Brand trust
Page-eval brand monitoring
1–10
Prompted trust rating across models for this brand on the topic.
Brand quality
Page-eval brand monitoring
1–10
Prompted quality rating across models for this brand on the topic.
Topic trend strength
Freshness & opportunity
none → massive
Whether the intended concept has meaningful momentum right now. Feeds Topical Opportunity.
Page last modified
Freshness & opportunity
Date
Best available last-meaningful-update date (headers, schema, visible text, or search metadata).
Update driver
Freshness & opportunity
Enum
Why Topical Opportunity is high: topic trend, freshness gap, Content Freshness urgency, or a combination.
Google AI Overview citation
Google AI surfaces
Yes / no
Whether this domain is cited in Google’s AI Overview for the page topic. Observational, not a GEO Studio prompt run.
Lacks logical chunks
Structure
Flag
Page has no usable chunk structure. Precision cannot be scored meaningfully until headings and sections exist.
Page changed
Structure
Flag
Main body fingerprint changed since the last crawl — triggers re-evaluation.