What We Detect - And What We Don't
We believe honest transparency about capabilities and limitations builds more trust than inflated claims. Here is exactly what GDPR Privacy Monitor checks, our confidence levels, and what falls outside our scope.
What We Detect
| Check Category | What We Look For | Confidence |
|---|---|---|
| Pre-consent tracking | Cookies, trackers, and scripts active before user consent interaction | HighHigh (Verified CMP) / Medium (Compatible CMP) |
| Cookie classification | Categorization into necessary, analytics, marketing, functional | HighHigh for 5,000+ known cookies; "unclassified" for unknown cookies |
| CMP detection | Consent banner presence and CMP type identification | HighHigh: CookieYes, Cookiebot, Complianz. Medium: OneTrust, iubenda, Didomi, Quantcast + 38 others |
| Reject flow verification | Whether "Reject" actually stops tracking cookies and third-party requests | HighHigh for Verified CMPs with known reject selectors |
| Banner accessibility | Keyboard navigation, color contrast, touch targets, ARIA labels (WCAG) | HighHigh (browser DOM measurements) |
| Visual prominence | Whether reject is equally prominent as accept (dark pattern detection) | MediumMedium (CSS sampling, may vary by viewport) |
| Cookie wall detection | Whether content is blocked until consent is given | MediumMedium (heuristic-based) |
| Google Consent Mode v2 | Detection and mismatch analysis between declared consent state and observed behavior | HighHigh (runtime interception) |
| Data flow mapping | Third-party data destinations, organizations, and adequacy status | MediumMedium (45% of domains unmapped) |
| External font loading | Google Fonts and other CDN IP leaks before consent (LG München ruling) | HighHigh (URL pattern matching) |
| Mixed content | HTTP resources loaded on HTTPS pages (security of processing) | HighHigh (network request analysis) |
| Insecure cookies | Tracking cookies without Secure flag or appropriate SameSite | HighHigh (cookie attribute inspection) |
| Browser fingerprinting | Pre-consent fingerprinting-sensitive API calls (Canvas, WebGL, fonts) | MediumMedium (compound scoring of intercepted signals) |
| Evidence collection | HAR network log, page screenshots, banner screenshots | HighHigh (100% coverage on successful scans) |
| Preference center analysis | Second-layer category toggles, pre-selected optional categories | HighHigh for Verified CMPs |
| DPA enforcement context | Historical enforcement fines and cases mapped to detected findings | HighHigh (structured enforcement database, 10 EU jurisdictions) |
What We Don't Detect (Honest Limitations)
| Limitation | Why | What We Do Instead |
|---|---|---|
| Server-side tracking | Cannot see server-to-server data transfers | We detect client-side indicators and note the limitation in the report |
| Geotargeted banners | We scan from EU (Finland); may miss geo-specific configurations | We note viewport and location in every report for transparency |
| Mobile app consent | We scan websites, not native mobile apps | Mobile viewport scanning available for responsive sites |
| Legal policy quality | We do not assess privacy policy text quality in depth | AI-powered policy gap detection identifies missing disclosures vs runtime behavior |
| Cookie purpose accuracy | Some cookies lack reliable classification in public databases | We label as "unclassified" with honest scoring rather than guessing |
| Dynamic SPA timing | Single-page apps may load consent banners late or asynchronously | We use adaptive waiting strategies and report "could not detect" when uncertain |
| Post-scan changes | Websites change after we scan them | Continuous monitoring catches regressions; each report is timestamped |
| Cross-origin iframe banners | Some CMPs render banners in cross-origin iframes that cannot be inspected | We detect CMP signals and report banner visibility uncertainty |
Our Philosophy
"Cannot determine" > false positive
When we are not confident about a finding, we mark it as "cannot determine" rather than guessing. We do not invent false positives to pad detection counts. An honest "cannot determine" is more useful than a confident wrong answer.
Risk Score, not Compliance Score
Our 0–100 score indicates technical risk, not legal compliance status. A low score means fewer technical risk indicators were found - it does not certify GDPR compliance. Only legal counsel can determine compliance.
Independent auditor
We do not sell consent banners, CMP software, or cookie solutions. This means we have no conflict of interest - we can tell you honestly when your CMP is misconfigured, even if it's a popular one.
Evidence-first reporting
Every finding is backed by HAR network data and screenshots. You can verify any finding independently using the evidence pack we provide. No black-box scoring - full transparency.
See it in action on your site
Run a free scan and see exactly what we detect on your website, with full evidence and a transparent risk score breakdown.
Run a free scan