You buy clicks; what arrives are visits. ValidVisit scores each one against 100+ data points and pins it to the campaign, source and publisher that sent it — so you can prove which traffic was never real.
TaboolaProspecting · site deals-hub.example · placement 447134Invalid
PropellerAdsPush RON · zone 4471 · placement 8884Valid
AdsterraDirect · zone 3391 · placement 33133Invalid
OutbrainContent · publisher daily-briefing.example · placement 209388ValidNetworks we score — wherever you buy traffic · search · social · native · banner · push · pop
All 74 networks
AdNow
AdskeeperCoContent.ad
Kadam
MediaGo
MGID
NewsBreak
Outbrain
Revcontent
SourceKnowledge
Taboola
Yahoo Native
AdOperator
AdsCompassAdsterra
DAO.AD
DatsPush
EvaDav
EZmobMeMegaPush
Mondiad
Noviclick
OnClickA
PropellerAds
Push.House
Pushground
ReachEffect
RichAds
ROIads
RollerAds
TacoLoco
Traffic Nomads
UngAds
ActiveRevenue
Adcash
AdMaven
ClickAdillaMost fraud tools identify every visitor to flag a few — cookies, device fingerprints, raw IPs on file. ValidVisit reads the click’s origin, the device class and the behaviour instead, so nothing personal is ever stored.
Powerful — but it turns your tag into a privacy liability and a long legal review.
Passes security review. No cookie banner to add.
Averages hide the leak. Move the sliders and see how many of your monthly clicks were likely never real — then trace them to the source in the console, where verdicts feed the cut-lists and postbacks you act on.
a month spent on clicks that were probably never real.
25,000 of 250,000 clicks · at your $0.40 CPC
benchmarks put it above 10%
stays in your browser
An estimate from your numbers, not a measurement. Your real figure comes from your own traffic, scored 0–100 per source. Why your own rate is the only benchmark that matters →
— industry benchmarks put invalid traffic above 10% of paid clicks · see the benchmarks
A caught bot, a borderline visitor and a real human — scored side by side on one 0–100 scale, with the source and the signals behind each verdict. The model doing the work stays ours.
Every verdict in your dashboard names the exact checks a click failed. Publicly, the checklist stays sealed — so bots can’t study for the test, and neither can our competitors.
Traffic bought from servers and anonymizers, not from people.
Machine-driven browsers and sessions posing as visitors.
Clicks and scrolls no human hand produced.
Pages fetched by software that never renders them.
Connections that contradict their own story — clock, location, device.
Zones and publishers whose clicks keep failing the same checks.
Each category resolves to the exact checks a click failed — named in your dashboard.
A phone-shaped visitor on a datacenter line, no human input — a real browser with no real person behind it.
Illustrative case — every failed check is named in your dashboard, weighed from 100+ data points. Here, the checklist stays sealed.
“14% invalid” sounds fine — until you see which sources the average is hiding. Scored down to the publisher and zone you actually buy at, worst first.
2 of 5 sources flagged·8,254 invalid clicks surfaced·example data — see yours
PropellerAds · zone 33913,400 clicks · 66% invalid
Taboola · site deals-hub.example4,500 clicks · 58% invalid
Zeropark · source 882017,000 clicks · 24% invalid
ExoClick16,000 clicks · 7% invalid·
sportstream.example · zone 4419
2,000 clicks · 45% invalid
sportstream.example · zone 7781
4,000 clicks · 4% invalid
newsloop.example · zone 2204
10,000 clicks · 1% invalid
Outbrain · publisher morning-brief.example30,000 clicks · 2% invalid
Drop the placements that only send bots. Keep the ones sending customers.
100+ data points per visit → one 0–100 quality score, attributed per campaign, source and publisher — with a postback you can act on.
A single ~10 KB gzipped script catches every arriving visit. Campaigns keep running exactly as they are — nothing in your funnel changes.
Each visit is weighed against 100+ data points and reduced to one 0–100 quality score — the more that look wrong, the lower it scores.
Valid, suspect or invalid by source, campaign and publisher — with a postback. Cut what wastes budget, keep what converts.
<script defer src="https://cdn.validvisit.io/p.js" data-site="YOUR_SITE_ID"></script>No tag manager, no funnel changes — see the full install & pipeline
Native, pop & push, referral and ecommerce — each scored at the unit of inventory your channel leaks at.
Descend from publisher to site to the single widget that leaks — so one bad placement can’t hide inside a good campaign.
Break it down by publisher → site → widgetDetection built on scored signals — not on profiling your visitors. The paper trail is for the fraud, never for the humans.
raw IP203.0.113.47a1f9…7c2non-reversible token, hashed on arrival — never written to disk
One mode, minimized by design — no cookies, no raw IPs, no fingerprint identities — and one ~8 KB script inside a 15 KB budget. Read the security overview →
Start with a free trial at launch — every tier gets the full scoring engine; tiers differ by volume, sites and depth. Join the waitlist to be first in.
For a single site getting started
For teams running paid traffic
For agencies & high volume
7-day trial at launch (card required) · cancel anytime · billed on visits inspected · no silent plan upgrades
Every visit scored 0–100 across 100+ data points, attributed to the source that sent it — across 74 ad networks.
One script · raw IP never stored · GDPR legitimate-interest basis
Guides, per-network breakdowns and benchmarks — a curated on-ramp; the footer has the full map.