AISEOgic — Complete Audit Methodology
The AISEOgic audit methodology is the exact set of formulas, weights, and scoring rules the tool uses to turn raw page signals into the 0-100 scores you see in a report, covering both deduction-based and weighted-ratio scoring across every category.
How the Overall Score Is Calculated
The Overall Score (0–100) is a weighted average of category scores. The category set and weights change depending on which modules are included (Free vs. Pro).
Free Plan — 5 Categories
| Category | Weight |
|---|---|
| Technical SEO | 22% |
| On-Page SEO | 27% |
| Performance | 22% |
| Security | 11% |
| Accessibility | 8% |
Fixed: The Free-plan weights sum to 90%, not 100% (22% + 27% + 22% + 11% + 8% = 90%) — and enabling different Pro modules can push the total anywhere from 90% to 107%, depending on the combination. This used to be a real bug: the Overall Score formula multiplied each category by its raw weight without dividing by the total, so the result could mathematically overshoot 100 (confirmed live at 102%+ with all four Pro modules enabled) or undershoot what a true average would show. This has since been corrected — see the formula below.
Pro Plan — Up to 9 Categories
| Category | Weight |
|---|---|
| Technical SEO | 18% |
| On-Page SEO | 22% |
| Performance | 18% |
| Security | 10% |
| Accessibility | 7% |
| AI SEO | 10% |
| GEO | 10% |
| AEO | 6% |
| LLMO | 6% |
Overall Score Formula
overall_score = round( Σ(category_score × category_weight) / Σ(category_weight) )Each category score ranges from 0 to 100. If a category did not run, it is excluded from both the numerator and the weight set.
The Two Scoring Methods Used Across Categories
AISEOgic uses two different scoring formulas depending on the size and number of checks in each category.
1. Deduction-Based Scoring — scoreCategory
This method is used for categories with fewer checks:
- Technical SEO
- On-Page SEO
- Security
- Accessibility
The score starts at 100, and points are deducted for every failed check.
score = 100 − Σ(deduction for every failed check)Deductions by Severity
| Severity | Deduction |
|---|---|
| Critical | −20 |
| High | −11 |
| Medium | −5 |
| Low | −2 |
The score is floored at 0 and capped at 100.
2. Weighted-Ratio Scoring — scoreCategoryByRatio
This method is used for categories with 20+ checks:
- AI SEO
- GEO
- AEO
- LLMO
Because these categories contain many checks, deduction stacking could cause the score to fall to 0 too easily. Instead, each check contributes proportionally to the overall result.
score = round( earned_points / possible_points × 100 )Points by Severity
| Severity | Points |
|---|---|
| Critical | 4 |
| High | 3 |
| Medium | 2 |
| Low | 1 |
- Passed checks earn their assigned points.
- Failed checks earn zero points.
- The score is floored at 5, so it never displays a misleading 0.
- The score is capped at 100.
- A category with zero checks defaults to a neutral score of 50.
Performance and Accessibility Scoring
The Performance category uses live Lighthouse/PageSpeed data directly when available, per Google’s own PageSpeed Insights documentation. If that data is missing, it falls back to the deduction-based scoring method.
The Accessibility category follows the same approach, using Lighthouse’s own accessibility (a11y) score when available and falling back to the deduction-based method when it is not.
Category-by-category detail
Revenue Impact Estimate
Revenue impact formula: revenue_impact = round( 300 + (critical_issues × 800) + (high_issues × 250) )
A flat heuristic estimate in USD/month, not a personalized financial projection. It scales only with issue counts and severities, not the site’s actual traffic, conversion rate, or industry.
This must be labeled as an illustrative estimate on any user-facing page, not a guarantee or data-backed forecast. Presenting it as precise would undercut the honesty standard held elsewhere in this product.
AI Enrichment (Optional, Pro)
If configured with Gemini and Wikidata API keys, two scores receive a live blend:
AI SEO score: 60% rule-based / 40% Gemini-assessed content quality.
GEO score: +8 if Wikidata confirms the entity and it has a Wikipedia page, +4 if confirmed without a Wikipedia page. According to research from Google’s own Knowledge Graph documentation, this kind of cross-referencing is one of the signals AI knowledge systems use to confirm a real-world entity actually exists.
This step is wrapped in try/catch and silently skipped if the enrichment file or API keys are missing. It never breaks or blocks a score.
For the public page: only claim “Gemini-verified” or “Wikidata-verified” scoring with a disclaimer that it applies when those integrations are configured and succeed, not unconditionally.
System-Wide Limitations
All checks are rule-based/deterministic against extracted page signals. Except for the optional Gemini/Wikidata step, nothing here queries ChatGPT, Claude, Perplexity, or Gemini live to see if they would actually cite the page.
GEO, AEO, and AI SEO are readiness proxies, not live citation tests.
Broken-link/redirect checks: sample up to 15 external links with lightweight HEAD requests. They are not a full-site crawl.
Performance scoring: depends on Google PageSpeed/Lighthouse being reachable. On failure, it silently falls back to the rule-based estimate.
Category weights: are engineering-chosen allocations reflecting known platform behavior and reported market share. They are not derived from a controlled outcome study.
Category weights: don’t always sum to exactly 100% (Free plan totals 90%; enabling AEO + LLMO alongside AI SEO + GEO totals 107%) — the Overall Score formula normalizes by the actual total weight of whichever categories ran, so this doesn’t affect the final 0–100 result.
“A methodology that hides its own imperfections isn’t transparent — it’s marketing. This page documents what the code actually does, bugs included.”