EnvTrace Detection and Evaluation Principles
EnvTrace shows the browser, device, and network signals that a website can observe during the current visit.
The assessment is designed to answer three questions:
- What can the website see right now?
- Do these signals make sense together?
- Which results need further review?
Where possible, EnvTrace separates direct observations, external network intelligence, and EnvTrace-derived conclusions. This makes it possible to see not only the result, but also why that result was produced.
Detection results describe the current environment
IP addresses, routes, browser versions, system settings, and network conditions can change. Timezone, language, WebRTC behavior, browser APIs, and other device signals may also change when the browser profile, device, or network changes.
For that reason, an EnvTrace result describes:
the environment observable to the page at the time the test was performed.
The time of the test is part of the result. If the network, browser profile, or system configuration has changed, running the test again is more useful than relying on an older report.
Direct observations, external intelligence, and derived conclusions are different
EnvTrace results come from three layers.
Direct observations
These are values that the current page can read or compute directly. Examples include:
- User-Agent
- Platform
- Client Hints
- display and device capabilities
- Canvas, WebGL, and Audio signals
- language and timezone
- WebRTC addresses
- the current public exit IP
These values describe what the page can actually observe.
External network intelligence
These are values supplied by external network data sources. Examples include:
- IP country and city
- ASN
- network provider
- proxy, VPN, or Tor indicators
- abuse-risk information
- results from the blocklists checked during the test
These results depend on the coverage and update cycle of the underlying data sources.
EnvTrace-derived conclusions
EnvTrace uses the first two layers to produce additional assessments, such as:
- whether browser identity signals agree with one another
- whether the current browser profile contains obvious conflicts
- which primary network type best describes the current IP
- whether browser timezone and language align with the current network environment
- whether WebRTC exposes a different public IP
- whether the combined signals contain items that need further review
A conclusion should always be traceable back to the signals that produced it.
Missing data is not treated as a passing result
Missing data does not automatically mean that a signal is normal. It also does not automatically mean that something is wrong.
When EnvTrace does not have enough information to make a reliable determination, it prefers states such as Unknown, Pending, or Unavailable instead of inventing a complete answer. For example:
- if network-quality intelligence is unavailable, the IP is not automatically described as clean
- if there is not enough timezone reference data, the result is not forced into match or mismatch
- if the current exit IP is unavailable, another WebRTC address is not automatically labeled as a public-IP leak
- if the browser does not support a particular API, unsupported is not counted as a successful check
Some assessments therefore also show coverage. Coverage indicates how much of the planned assessment had enough information to run.
A high score with limited coverage does not mean the same thing as a high score with nearly all checks completed.
A single signal rarely describes the whole environment
One value is usually not enough to establish the full state of a browser, device, or network. For example:
- A User-Agent describes the identity the browser reports, but it does not independently prove the underlying device.
- WebGL can reveal useful graphics information, but it does not by itself establish the complete operating system.
- An ASN describes the autonomous system announcing an IP prefix, but it does not identify the current user.
- IP geolocation describes the approximate location of a network exit, not the physical GPS location of a person.
EnvTrace therefore focuses on a broader question:
Do multiple independent signals support one another, or do they conflict?
This is the basis of both browser fingerprint assessment and environment consistency checks.
Consistency means that signals can reasonably coexist
EnvTrace compares signals within the browser and across the browser and network environment. Examples include:
- whether User-Agent, Client Hints, and Platform agree
- whether browser identity conflicts with graphics, hardware, or device-form-factor signals
- whether key identity values differ between the main page, iframe, and Worker contexts
- whether browser timezone and language align with the current network environment
- whether WebRTC exposes a different public IP
When the available signals can reasonably coexist under the current rules, they are treated as consistent or matching. When there is a clear conflict, the result is marked for review.
However: a consistent result does not mean that a third-party platform will accept the environment. A platform may also use account history, server-side records, behavioral signals, device history, and other information that EnvTrace cannot observe.
Likewise: an inconsistency does not prove that a user is spoofing, automating, or behaving abnormally. Privacy features, proxies, enterprise networks, virtualized environments, browser extensions, and compatibility issues can all create unusual combinations.
EnvTrace reports the technical signals that are observable in the current session.
Browser fingerprint assessment measures internal consistency
EnvTrace does not currently attempt to estimate how many browsers worldwide share the same fingerprint.
Instead, the browser fingerprint assessment asks: do the identity, API, graphics and hardware, execution-context, and stability signals exposed by the current browser profile form a coherent browser environment?
The current assessment covers:
- Identity Consistency
- API Integrity
- Graphics & Hardware
- Cross-context Consistency
- Stability
A higher score means that fewer obvious conflicts were found among the checks that were successfully completed.
It is not a population uniqueness percentile, and it is not a third-party platform risk score.
Network labels come from intelligence and classification rules
Labels such as residential IP, datacenter IP, VPN, proxy, and Tor are not fixed properties embedded inside an IP address.
An IP address does not identify itself as: “I am a residential IP.”
These labels are derived from network intelligence and classification rules.
Different services may classify the same address differently because they use different data sources, update schedules, and classification methods.
EnvTrace reduces related network indicators into one primary network type.
The exact classification order and conditions are documented in the IP & Network Methodology.
IP type, proxy indicators, blocklists, and abuse risk are separate signals
These results answer different questions. For example:
- A residential IP can still have elevated abuse-risk data.
- A datacenter IP can have no known proxy indicator.
- A proxy indicator is not the same thing as a blocklist record.
EnvTrace therefore presents these separately:
- network type
- proxy indicators
- checked blocklist results
- abuse risk
If only some of these results are available, EnvTrace describes only the data that actually returned. Missing values are not automatically treated as normal.
“No hit in checked lists” means that the blocklists successfully queried during the current test did not return a match. It does not mean that the IP is absent from every public blocklist.
Scores summarize findings; they are not real-world probabilities
EnvTrace may combine multiple checks into scores or levels so that users can quickly identify areas that need attention.
A score is a summary of the signals covered by that assessment. It is not:
- the probability that a platform will identify the environment
- the probability that an account will be restricted
- the probability that a browser is “real”
- a probability that an IP is safe
- a risk score used by a third-party platform
A score should always be accompanied by the underlying findings.
The score tells you where to look. The individual signals explain why.
Detection results should be explainable
Each major conclusion should be traceable to a dedicated methodology page.
If a result cannot explain which signals produced it and which rules were applied, it should not exist only as an unexplained score.
The meaning of a signal depends on context
EnvTrace aims to describe the signal itself rather than declaring that a particular environment is always good or bad.
A datacenter network may be completely normal for a server or cloud workload. The same network characteristic may require additional review in a different use case.
Network type, risk signals, and consistency results should therefore be interpreted in the context of the user’s actual task.
EnvTrace’s role is to expose the signals and explain the assessment, not to reduce every use case to one universal verdict.
How to read an EnvTrace report
A report can be read in three layers.
What was observed
Browser, device, IP, country, network provider, timezone, language, WebRTC, and other raw signals.
How EnvTrace interpreted it
Browser fingerprint consistency, network classification, network-quality signals, and browser-to-network consistency.
Why something needs review
Open the individual finding to see which signals triggered it and which methodology applies.
When a result affects your next decision, use the corresponding methodology page to review the full rule set.
Versioning
Version 1.0 defines the common detection and evaluation principles currently used by EnvTrace.
This document or the relevant methodology page will be updated when there is a material change to:
- the signals being collected
- classification logic
- scoring logic or important thresholds
- data sources that materially affect the meaning of a result
- the meaning of a conclusion presented to users
Visual changes, wording corrections, and implementation changes that do not affect the meaning of the assessment do not necessarily require a version change.