aiVIP™ • FAQ+
Detailed Scanner & Scoring Explanations
Additional information about the aiVIP Scanner V3.1.0 methodology, score calculation, weighting, confidence and interpretation.
Scoring methodology
How to interpret your result
What does my aiVIP score mean?
The aiVIP score is a readiness benchmark, not a prediction of whether an AI engine will cite or recommend your business. It summarises a defined set of website and business-context signals that can be checked consistently.
The value of the score is mainly comparative: it helps you identify weaknesses, make improvements and rescan using the same rules. The individual findings are often more important than the headline number.
How is the V3.1.0 score calculated?
V3.1.0 is designed around two components:
90% Observable Website Evidence
Technical, discovery, indexing, machine-readable, entity and supporting signals detected by the scanner.
10% Business Context
Positive questionnaire answers supplied by the user. These are useful supporting context but are not independently verified.
The final raw score is calculated from these components and then converted into the displayed whole-number score using the V3.1.0 rounding rule.
What does “deterministic” mean?
A deterministic scanner uses explicit rules. It does not ask a generative AI model to decide whether your website “looks good”. Each check has a defined condition, status, importance and scoring behaviour.
This makes the result more repeatable and auditable: the same inputs, under the same rules version, should produce the same score.
Why didn’t my website score 100?
Points can be lost because an important signal is missing, because the scanner detected only partial evidence, or because a check could not be reliably completed. A lower score does not automatically mean the website is poor; it means the scanner detected gaps against the current methodology.
Focus first on where the points were lost, especially Critical and High-importance findings.
What do the V3.1.0 score bands mean?
0–19
Very Low Readiness
Major foundations are absent.
20–34
Low Readiness
Important foundations remain weak.
35–49
Early Readiness
Some foundations exist, with material gaps.
50–64
Developing Readiness
A meaningful base exists; several improvements remain.
65–74
Good Readiness
Good foundations with identifiable gaps.
75–84
Strong Readiness
Most important foundations are in place.
85–100
Advanced Readiness
A high level of detected readiness.
How do my questionnaire answers affect the score?
The questionnaire contributes a deliberately limited 10% of the final score. This allows useful business practices to be recognised without allowing self-reported information to overpower observable website evidence.
The V3.1.0 methodology gives different importance to the positive ticks. Publishing useful content carries the highest share of the questionnaire component, followed by clearly identified experts, then analytics/measurement and maintained policy/business-information pages.
A positive tick can add supporting points, but it does not change a contradictory observed finding from Missing or Review to Present.
What does the scanner actually check?
The automated assessment includes checks covering homepage availability, HTTPS, selected headers, robots.txt, sitemap files, AI crawler access, sitemap discovery, page title, meta description, primary heading, canonical URL, indexing directives, document language, JSON-LD structured data, business/contact signals, Open Graph metadata, favicon, mobile viewport and selected accessibility indicators.
Some checks are more important than others and therefore carry different scoring weight in V3.1.0.
Why are checks weighted differently?
Equal weighting can make a minor presentation signal mathematically as important as whether the homepage can be indexed. V3.1.0 corrects this by assigning greater influence to signals that have a stronger practical relationship with discoverability, accessibility to crawlers and machine understanding.
For example, homepage availability and indexability are treated as much more important than a favicon or an experimental guidance file.
What do Present, Review, Missing and the other statuses mean?
Present
The required signal was detected. It normally receives full credit for that rule.
Review
Some evidence exists or the condition is imperfect or ambiguous. V3.1.0 normally gives partial credit.
Missing
The expected evidence was not detected. It receives no points for that rule.
Blocked / Failed / Not checked
The scanner could not reliably assess the rule. These states also affect confidence.
What is the confidence score?
Confidence answers a different question from readiness: how much of the important observable evidence could the scanner reliably assess?
V3.1.0 is designed to calculate confidence using the importance of successfully checked rules rather than treating every check as equally significant. User questionnaire answers do not inflate technical scan confidence.
If the homepage cannot be assessed or confidence is too low, the scanner can return an Inconclusive result rather than displaying a misleading numerical score.
Why does the scanner check llms.txt?
llms.txt and related guidance files may be useful as supplementary machine-readable guidance, but they should not be treated as established ranking or citation factors. V3.1.0 therefore gives them a low, supporting or experimental weighting.
This avoids allowing an optional emerging convention to distort the main readiness score.
Why does AI crawler access matter?
robots.txt can explicitly restrict automated agents. If a recognised AI or search crawler is blocked from the whole site, direct access to the content may be limited.
The scanner checks for access restrictions as an observable readiness signal. This does not mean that allowing every crawler guarantees inclusion in an AI result.
Improving your result
From diagnosis to action
Will fixing the findings improve my score?
If a weighted finding improves from Missing or Review to Present, it should normally improve the next score under the same rules version. The size of the change depends on the importance assigned to that rule.
Why rescan?
Rescanning lets you verify whether changes are actually detectable from outside the website. Because the methodology is deterministic, it can provide a useful before-and-after benchmark.
For meaningful comparisons, retain the rules version associated with each historical scan.
Which findings should I fix first?
Prioritise issues that cause the greatest weighted loss and that affect basic access or discoverability. In general, a blocked or unindexable site should be addressed before lower-impact metadata or presentation refinements.
V3.1.0 is intended to surface the most important actionable losses rather than simply list findings in arbitrary order.
Does a high score guarantee AI visibility?
No. External AI/search systems make their own decisions. They may use information from your website, third-party sources, structured data, indexes, knowledge graphs, user-generated content and other evidence.
The scanner measures whether important foundations are detectable. It cannot guarantee that a particular prompt, engine or user will receive your company as an answer.
How is aiVIP different from traditional SEO?
There is considerable overlap: crawlability, indexability, clear titles, structured information and useful content matter to both. aiVIP places additional emphasis on whether machines can clearly understand the entity, attribute claims, identify expertise, find supporting evidence and reference the business in answer-led environments.
Why might I still need a manual AI visibility review?
An automated scanner can detect objective website signals efficiently, but it cannot fully determine whether your content demonstrates expertise, answers commercially important questions, is corroborated externally, is cited by competitors or is actually appearing in AI-generated answers.
Those areas are better suited to deeper aiVIP analysis and human review.
Technical interpretation
Rounding, versioning and limitations
How does V3.1.0 round the final score?
The published whole-number score follows a specific owner-defined rule:
Fractional part ≤ .5: round down.
Fractional part > .5: round up.
Examples: 64.49 → 64 • 64.50 → 64 • 64.51 → 65 • 79.50 → 79 • 79.51 → 80.
Why is methodology versioning important?
The scanner software version and scoring-rules version identify the methodology that produced a result. This prevents a historical score from becoming ambiguous if weights, bands or scoring rules are improved later.
Existing filesystem folder names such as aivipv2 or aivip_v2 are deployment identifiers and are separate from the software/methodology version.
Are questionnaire answers independently verified?
No. Questionnaire answers are explicitly self-reported. They are stored and scored separately from observed website evidence and should be presented to the user as supporting business context.
What does “Scan inconclusive” mean?
An inconclusive result is a protection against false precision. If the scanner cannot reach the homepage or cannot assess enough important evidence with sufficient confidence, it is better to withhold a numerical score than display a number that appears more certain than the evidence supports.
Can an automated finding be wrong?
Yes, occasionally. Websites can implement features in unusual ways, content can be generated dynamically, servers can respond differently to automated requests, and some signals require interpretation. “Review” exists partly to distinguish ambiguous evidence from a clear pass or failure.
If an important finding appears incorrect, it should be manually checked rather than changed solely to satisfy the scanner.
What is the guiding principle behind the V3.1.0 methodology?
The methodology is designed to be evidence-first, importance-weighted, deterministic, transparent, bounded and versioned.
Observable website evidence remains dominant. Self-reported context can support the score but cannot override contradictory technical evidence. More important visibility foundations contribute more than minor or experimental signals.
