QUICK ANSWER
Read all five metrics together
The five metrics measure appearance, evidence, competitive share, absolute volume, and platform range. Even identically named metrics produce different figures when valid-response rules or units of analysis differ. The goal is to overlap metrics and find the bottleneck—not react to one metric in isolation.
SUMMITFEED MEASURED
SUMMITFEED website audit results
Final local production-HTML re-audit: 2026-08-06
82/82
Public URLs returned HTTP 200
82/82
Canonical matched the current URL
166/166
JSON-LD blocks parsed successfully
0
BLOCKER · HIGH findings
These figures are technical validation results for public URLs, canonical tags, and JSON-LD—not AI Visibility. They are kept separate from AI response metrics and indicate whether the site's measurable foundation is healthy.
5 METRICS
The five metrics at a glance
01 · Appearance
AI Visibility
Visibility rate
Valid responses mentioning the brand ÷ all valid responses × 100
Does our brand appear in the answer?
Check: Separate branded and non-branded questions.
02 · Evidence
Citation Rate
Official-source citation rate
Valid responses linking an official URL ÷ all valid responses × 100
Is our document used as evidence?
Check: Verify both the citation URL and whether the link resolves.
03 · Share
Share of Voice
Mention share
Brand mentions ÷ mentions of all comparison brands × 100
How much share do we hold against competitors?
Check: Fix the comparison-brand list and duplicate-mention rule.
04 · Volume
Mention Volume
Absolute mention volume
Total brand mentions observed during the measurement period
How often did the brand appear within the measurement scope?
Check: Record question count, platform count, and repeat count.
05 · Range
Platform Coverage
Platform coverage
Platforms where the brand appeared ÷ all measured platforms × 100
Is visibility distributed across platforms?
Check: Read it together with AI Visibility for each platform.
Overlapping metrics explain the cause
One value only shows movement. Two metrics together reveal which layer is blocked and which task to check first.
| Observed combination | What it suggests | Check first |
|---|---|---|
| High Visibility · low Citation | The name is known, but official documents are not used as evidence | Indexed pages with definitions, figures, procedures, and comparison tables |
| High Citation · low Visibility | Documents are used, but the brand connection is weak | Publisher, author, and company-name presentation |
| High Visibility · low SOV | The brand enters the candidate set but holds a small share | Conditions under which competitors are introduced first |
| High Coverage · wide platform variance | One platform is lifting the combined average | Platform-level measurement rules and results |
| More mentions · flat Visibility | The measurement scope may simply have expanded | Question, platform, and repeat-count change history |
| Stable metrics · no inquiry change | Visibility is disconnected from the next action | Cited landing pages and their conversion paths |
What belongs in a monthly report?
A report should support the same measurement next month. Record calculation conditions and supporting data beside each value.
| Metric | Condition to record | Supporting data |
|---|---|---|
| AI Visibility | Valid-response count and unit of analysis | Results by platform and question type |
| Citation Rate | Accepted URL scope | Cited page and link-reachability result |
| Share of Voice | Comparison-brand list | Questions showing only the brand or only competitors |
| Mention Volume | Questions × platforms × repeats | Mentions per response |
| Platform Coverage | Platform and model list | AI Visibility by platform |
REPORT RULE
Do not collapse the metrics into one score
The five metrics use different measurement rules. Put changed and unchanged metrics side by side to reveal the next task.
- Question-set version and change history
- Platform list and model versions
- Country, language, and login state
- Repeat count and measurement cycle
- Valid-response rules and unit of analysis
- Brand-name variants and comparison-brand list
- Storage location for full responses, citation URLs, and collection time
Frequently asked questions
What are the five GEO performance metrics?
AI Visibility, Citation Rate, Share of Voice, Mention Volume, and Platform Coverage. They measure appearance, evidence, competitive share, absolute volume, and platform range.
How do AI Visibility and Citation Rate differ?
AI Visibility counts whether the brand name appears. Citation Rate counts whether an official URL is connected as a source. A named brand and an official document used as evidence are different outcomes.
How many questions should a GEO measurement use?
Start by checking how many percentage points one response can move the total. Repeat the same questions under the same conditions to measure run-to-run variance, then expand the question and platform set.
Can Share of Voice be used directly as a KPI?
First inspect the calculation. Period comparisons require the comparison-brand list, duplicate-mention rule, and whether the number is observed or modeled.
What should I check if metrics rise but inquiries do not?
Start with the cited URLs and the next-action path on those pages. An old article or a page with no inquiry path can be cited repeatedly without creating business action.
Are SUMMITFEED's published figures AI Visibility results?
No. The 82/82 and 166/166 figures on this page are technical website-audit results. They are separated from AI response metrics and show whether the site's measurable technical foundation is operating correctly.
CONCLUSION
Read the calculation rules before the number
Define valid responses and the unit of analysis first, then overlap appearance, evidence, and competitive share. SUMMITFEED separates technical audit results from AI response metrics and stores questions, platforms, timestamps, and full responses to decide the next reinforcement task.