UNDERSTAND · TRAFFIC QUALITY
Traffic quality: know how much of your traffic is real.
Some automated visitors run JavaScript, load your analytics tag and appear in GA4 as sessions, sometimes engaged ones, sometimes conversions. Traffic Quality shows the evidence from your logs, consented browser signals and GA4, and which numbers it may affect.
- Evidence, not a verdict
- Consented browser signals
- Nothing blocked or changed
3 sources
server logs, browser signals, GA4
5 verdicts
from verified bot to insufficient evidence
Versioned
a deterministic, published policy
No IDs
no identifiers or fingerprints collected
INSIDE THE PRODUCT
See automation in the numbers you report.
Three kinds of evidence count different populations, so they are shown side by side and never added together.
Evidence
Three kinds of evidence, kept apart
Server and CDN logs see bots that never run JavaScript. Browser signals see automation that does. GA4 shows the aggregate sessions, engagement and key events that may be affected.
- Server and CDN requests
- Consented browser signals
- Aggregate GA4 data
| Source | Sees | Population |
|---|---|---|
| Server & CDN | Bots without JS | Requests |
| Browser signals | Automation with JS | Page views |
| GA4 | Sessions & events | Aggregates |
Verdicts
Reviewable verdicts with the reasons shown
Each measured page view is a verified bot, high-confidence suspected automation, possible automation, no material indicators or insufficient evidence, with every signal, its reliability and what else can produce it.
- Five clear verdicts
- Signal reliability shown
- Deterministic, versioned policy
- Strongnavigator.webdriverPresent at load
- StrongCrawler name from wrong networkOutside the operator’s addresses
- ModerateHeadless rendering hintsNeeds corroboration
- WeakNo interactionOnly corroborates
GA4 impact
See which GA4 numbers may be inflated
Find the landing pages, countries, devices and key events where suspected automation concentrates, before you report a conversion rate or cut a campaign.
- Affected landing pages
- Countries and devices
- Key events at risk
Conversion integrity
Check key events before you report them
Find key-event bursts, purchase events that outnumber transactions, session starts above sessions, and duplicated Google tags on the homepage, such as one measurement ID configured twice.
- Key-event bursts and impossible ratios
- Duplicated Google tags
- Dates and pages affected
- Highpurchase > transactions3 days with more purchases than orders
- MediumDuplicate tagOne measurement ID configured twice
- MediumKey-event burstgenerate_lead ×6 on 14 Oct
- OKsession_startIn line with sessions
Protection rules
Route suspected automation, only after review
Optional rules move high-confidence automated page views to a separate GA4 property. A rule runs in shadow first, needs explicit approval, and produces a Tag Manager configuration you apply yourself, with rollback steps.
- Observe, shadow, then approve
- Staff visits caught or spared in the preview
- Versioned with one-click roll back
| Measure | Routed | Note |
|---|---|---|
| Page views | 4.1% | High confidence only |
| Tab sessions | 3.6% | |
| Staff visits | 0 | Logged-in editors spared |
| Missed after load | 1.2% | Signals arrive too late |
Alerts
Hear when the pattern changes
Hourly checks raise an alert when browser signals stop arriving, the suspected share jumps, a new burst of high-confidence automation appears, or a rule would catch staff or route too much.
- Signals stopped arriving
- Jump in suspected share
- Rules that would over-route
- NewHigh-confidence burst312 page views on /promo/ on 21 Oct
- RuleShadow ruleWould route 24% of page views; review
- InfoSignalsArriving normally
WHAT IT MEASURES
What Traffic Quality counts, and what it never does.
What each metric tells you, and where it comes from.
| Metric | What it tells you | Source |
|---|---|---|
| Page views by verdict | Measured page views as verified bot, high-confidence suspected, possible, no material indicators or insufficient evidence. | Consented browser signals |
| Suspected tab sessions | Browser tabs with suspected automation, by landing page and device. | Consented browser signals |
| Estimated automated range | A range of automated browser page views shown beside GA4 sessions, never subtracted from them. | Browser signals + GA4 |
| GA4 anomalies | Unusual days by channel and landing page, with corroboration. | Google Analytics 4 |
| Automated requests | Bots that never run JavaScript, seen in server and CDN logs. | Crawl Analytics sources |
| Conversion integrity | Impossible event ratios and duplicated tags. | GA4 + homepage scan |
UNDER THE HOOD
How Traffic Quality reaches a verdict.
How the numbers are produced, so you can judge them for yourself.
Signals with known reliability
Strong signals, such as an automation flag present at load or a crawler name from the wrong network, can decide a verdict. Moderate signals need corroboration, and weak ones, such as no interaction, only corroborate.
A deterministic, versioned policy
The same signals always produce the same verdict. The policy is versioned, so a change in the rules is visible rather than silently rewriting history.
Collected after consent
With the WordPress plugin option on, a small script sends one beacon per page view after consent, with a few categorical indicators and no identifiers. It never runs on cart, checkout, account, login or admin pages.
Shown beside GA4, never subtracted
Where browser signals cover a period, an estimated range of automated page views appears next to GA4’s sessions. GA4 history is never corrected or rewritten.
HOW IT WORKS
From a suspicious spike to a reviewed decision.
Four steps from connecting your data to a decision you can stand behind.
Connect the evidence
Connect GA4 and a Crawl Analytics source, and optionally turn on consented browser signals in the WordPress plugin.
Read the verdicts
See page views by verdict, with each signal’s reliability and what else could cause it.
Find the impact
Open the landing pages, channels and key events where suspected automation concentrates.
Decide, then verify
Exclude a segment from analysis, or shadow-test a protection rule before any change.
WITH YOUR AI AGENT
Ask whether a spike was real people.
Your agent reads the Traffic Quality views and can shadow-test a rule, but it can never create, approve or publish one.
- Uses the same data as the web app
- Paid tools estimate their cost first
- Saves the report to the project
YouWas last Tuesday’s traffic spike real?
get_traffic_qualityget_google_analytics_traffic_acquisitionpreview_traffic_rule
AgentProbably not all of it. 38% of page views on /promo/ that day were high-confidence suspected automation with the webdriver signal present. GA4 shows the sessions as direct with 4-second engagement.
YouWhat would a rule for that route?
AgentIn shadow, a high-confidence rule would have routed 4.1% of last week’s page views and no staff visits. Approval stays with you.
Illustration · sample data
USE CASES
Built for the way you work.
How teams put it to work.
Trust your conversion rate
See whether automated sessions are inflating key events before you report them.
Judge campaigns on real visits
Check a traffic spike for automation before crediting a campaign.
Explain an odd month
Show a client which part of a traffic jump showed signs of automation, with the evidence.
Find duplicate tags
Spot one measurement ID configured twice, or gtag.js beside Tag Manager, before it doubles events.
ALSO INCLUDED
More from Traffic Quality.
More capabilities that come with it, in the same project.
Fair to real people
Privacy browsers, assistive technology and VMs are not flagged on their own.
Consent first
Browser signals run only with consent, via the WP Consent API or your banner.
Never on checkout
Signals never run on cart, checkout, account or login pages.
DATA & LIMITS
Know what the numbers can and cannot tell you.
Where the data comes from, how fresh it is, and the questions it cannot answer on its own.
| Source | What it provides | Freshness & limits |
|---|---|---|
| Consented browser signals | Categorical automation indicators per page view. | WordPress plugin 2.2+, with consent; never on cart, checkout, account or login pages. |
| Server and CDN logs | Bots that do not run JavaScript. | From your Crawl Analytics source. |
| Google Analytics 4 | Sessions, engagement and key events that may be affected. | GA4 already removes some known bots without saying which. |
Suspected is not proven
A verdict is evidence under a published policy, not proof of a bot or of a person.
No corrected history
Ndovu never rewrites GA4 history; estimates sit beside GA4 numbers.
Rules are client-side
A rule cannot stop Measurement Protocol spam or bots that never run JavaScript.
Country is never a signal
Country and time zone are context only; they never mark a visitor as automated.
WORKS WITH
Connected to the tools you already use.
The sources and clients this feature reads from or works alongside.
CHOOSE THE RIGHT VIEW
Which view answers which question.
Start where the question points. Each view keeps its own source.
| Your question | Start with | Why |
|---|---|---|
| Is this traffic real? | Traffic Quality | Automation evidence in your analytics. |
| Which crawlers request my pages? | Crawl Analytics | Verified crawlers in your logs. |
| Which pages convert? | Analytics & AI referrals | Sessions, events and revenue. |
LEARN & TRY
Guides and free tools for the same job.
Read how the evidence works, or try a free check before you connect a project.
GA4 landing page analyzer
Read a landing-page export.
UTM campaign builder
Build correctly tagged campaign links.
Crawler directory
Every crawler, its operator, its robots.txt token and how to verify it.
FAQ
Traffic Quality: common questions.
Straight answers about what it does, what it costs and where its limits are.
Does Traffic Quality block visitors?
No. Nothing blocks visitors or changes your tags. It reports evidence so you can decide what to do.
Is “suspected automation” proof of a bot?
No, and “no material indicators” is not proof of a person. Verdicts are evidence under a published policy.
What do browser signals collect?
A few categorical automation indicators per page view, sent after consent. No identifiers, fingerprints, form values or keystrokes.
Which data does it need?
GA4 and a Crawl Analytics data source. Browser signals need the WordPress plugin 2.2 or later with the option turned on.
Does it work without the WordPress plugin?
GA4 views and log evidence work without it. Browser signals and protection rules need the plugin.
Can a rule hide a real sale?
It could, which is why rules start in shadow, need approval and never touch ad conversion tags.
How long is data kept?
Daily aggregates for 400 days, landing pages for 90 days. Removing the WordPress source deletes them.
Can an agent publish a rule?
No. Agents can read the views and preview a rule; creating, approving and publishing stay with people.
Is it fair to privacy browsers?
Yes. Weak signals only corroborate, so privacy browsers, assistive technology and virtual machines are not flagged on their own.
What are the limits of the data?
Suspected automation is not proof of a bot, and “no material automation indicators” is not proof of a person. Nothing blocks visitors or changes your tags. Browser signals hold no identifiers or fingerprints and never run on cart, checkout, account or login pages.
How do I get access?
Platform access is reviewed. Request access with your website and the data sources you use, and the team will help you connect them.
See it with your own data.
Tell us about your website. We will review access and help you connect the right sources.