Metrics

Dashboard Metrics Guide

What each Epitom dashboard number means, in short.

How measurement works

Epitom asks buyer-style questions to AI engines (Google AI Mode, ChatGPT, Claude, Perplexity, Gemini, and others), then scores each answer for brand presence, framing, placement, and sources.

Term	Meaning
Prompt	A tracked question
Answer	One engine’s reply to one prompt
Owned domain	Your site(s) in Epitom
Competitor	Brands you track for comparison
Period	Selected date range

Where metrics appear

Dashboard page	What you’ll see
Overview	Brand metrics, trends, top sources
Prompts	Per-prompt metrics + Mentions + Volume
Domains / URLs	Cited sources + Gap Score
Crawlability	AI bot access
Insights	Findings from the metrics above

Core brand metrics

Visibility, Share of Voice, Sentiment, and Position — higher is better for all four.

1Visibility

0–100%

How often does our brand show up in AI answers?

Formula

Visibility=(answers mentioning brand ÷ total successful answers) × 100

Range	Reading
0–20%	Rarely present
20–50%	Hit or miss by prompt
50–80%	Usually present
80%+	Nearly always present

Example

28 of 40 answers mention you → 70%

2Share of Voice (SOV)

0–100%

Of brand mentions in an answer, how many are ours?

Formula

SOV=brand mentions ÷ (brand mentions + competitor mentions)

Visibility	SOV	Reading
High	High	Present and dominant
High	Low	Present but crowded
Low	High	Rare but dominant when present
Low	Low	Weak presence

Example

You 3, A 2, B 1 → SOV = 50% for that answer

3Sentiment

0–100

When mentioned, is the framing positive, neutral, or negative?

Formula

Sentiment=avg({positive: 100, neutral: 50, negative: 0})

Score	Reading
~100	Usually recommended
~65–85	Mostly positive
~50	Mostly factual
<40	Often critical

Example

6×100 + 3×50 + 1×0 over 10 answers → 75

4Position

0.0–1.0

How early does our brand appear in the answer?

Formula

Position=avg({top: 1.0, middle: 0.6, bottom: 0.3, absent: 0.0})

Placement	Weight
Top (first quarter)	1.0
Middle	0.6
Bottom	0.3
Absent	0.0

Example

Three top, one middle, one bottom → 0.78

5Prompt-level metrics

Per prompt

What does each Prompts row show?

Column	Meaning
Visibility / SOV / Sentiment / Position	Same defs, one prompt
Mentions	Other brands that also appeared
Volume	7-day mention-rate sparkline

6Retrieval Rate

0–100%

How often is this domain or URL used as a source?

Formula

Retrieval rate=(answers citing source ÷ total successful answers) × 100

Example

Cited in 30 of 100 answers → 30%

7Citation Rate (domains)

0.0+

When this domain is used, how many of its pages show up in one answer?

8Competitor Citations & Brand Mentions

Counts

Is the source contested? Do we already co-appear?

9Gap Score

0–100

Where is the opportunity?

Formula

Owned100 − retrieval%
Other domains≈ retrieval%
Other URLsretrieval% × (1 − brand co-mention rate)

Owned: 100 − retrieval %
Other domains: ≈ retrieval %
Other URLs: discounted if you already co-appear

10Crawlability

Allow Rate %

Can AI crawlers reach our site?

Metric	Meaning
Allowed / Partial / Blocked	Bot access state
Allow Rate	% of tracked bots fully allowed

Example

14 of 20 bots allowed → 70%

11Quick cheat sheet

Guide

What should we do when a pattern shows up?

You see	Next focus
High Visibility, low own-site citations	Cite-worthy content; crawlability
High Visibility, low SOV	Comparison content; PR on AI sources
Low Visibility, high Gap on rival sources	Earn coverage or build alternatives
Late Position / thin mentions	Richer category content
Sentiment drifting down	Review messaging on contested prompts
Many bots Blocked	Unblock crawlers first

12Scales at a glance

Reference

What does higher mean for each metric?

Metric	Scale	Higher =
Visibility	0–100%	More presence
SOV	0–100%	More of brand talk
Sentiment	0–100	More positive
Position	0–1	Earlier in answer
Retrieval Rate	0–100%	Source used more
Gap Score	0–100	More opportunity
Allow Rate	0–100%	More bots allowed