Prizvox
PricingCompareMethodologyBlogAboutContact
Log inStart free trial

Methodology

How the Prizvox Score
is calculated

Traditional SEO metrics (rankings, domain authority, page authority) were built to measure visibility in a world where every answer lived on a blue-link results page. That world is ending. A brand can have a 70 DA, rank in position 1 for every target keyword, and be completely invisible when someone asks ChatGPT the question that should send them there. Most SEO agencies are optimising for the last platform, not the next one. Prizvox measures both.

The Prizvox Score is a weighted composite across five components. Each component has a documented data source, weighting rationale, and boundary condition. If you or a client asked “how exactly is this calculated?” the answer is specific enough to reproduce independently. A score nobody can explain is worse than no score.

Score composition

Performance20%
SEO20%
Authority25%
AEO15%
GEO20%
ComponentWeightWhat a score above 75 means
Performance
20%Site loads in the top quartile. LCP under 2.5 s for real users. INP under 200 ms.
SEO
20%20+ on-page and crawlability checks passing. No critical indexability issues. Canonical, robots.txt, sitemap all clean.
Authority
25%Domain Authority above 40. Sufficient third-party trust to compete for target SERPs. Page Authority indicates individual page strength.
AEO
15%Structured data present and valid. Author signals detectable. Content formatted for featured snippet and AI Overview eligibility.
GEO
20%Brand appears consistently across AI platform responses to discovery queries. Score above 80 indicates strong AI search presence.

What happens when data is unavailable

Clients will ask this. The answer is documented before they ask.

Real-user data unavailable

Site has insufficient traffic for real-user field data. Performance score uses lab data only. A notice badge appears on the score card. The score is still valid — lab data is a reasonable proxy for low-traffic sites.

Authority data unavailable

Authority component falls back to a public fallback source. Score is footnoted. A caveat badge on the score card indicates reduced authority data.

GEO data unavailable

The GEO component scores 0 and its 20% weight is redistributed proportionally to the other four components. The GEO tab shows a notice.

Partial audit failure

The Prizvox Score reflects available data. A caveat badge on the score card lists the failed modules. The score is not inflated. Missing modules do not count as passing.

Private or blocked domains

If the HTTP scrape returns a 403 or requires authentication, the audit completes with the data available from the DNS, sitemap, and external API signals. Limitations are noted.

GEO scoring: how it works

This component is unique to Prizvox. The methodology is auditable.

1

Prompt generation

A set of discovery prompts is generated for each domain. Prompts are location-aware and industry-contextual, drawn from signals identified during the site crawl. They reflect the types of questions a real buyer asks an AI assistant before shortlisting vendors.

2

Platform queries

Each prompt is sent to all four major AI assistants in parallel so results reflect the current state of each platform at the time of the audit.

3

Response analysis

Each response is analysed for brand presence: whether the brand appears prominently, appears peripherally, or is absent entirely. The analysis checks the brand name, domain, and known brand variants across the full response text.

4

GEO score

The GEO score reflects the proportion of platform responses in which the brand has a meaningful presence. A score above 80 indicates the brand is consistently surfaced across AI platforms for relevant discovery queries. A score of 0 means the brand does not appear in any response.