Layer 1 · before the spend
🛡️ Fraud doesn't steal your clicks. It steals your learning.
Money spent on fake traffic hurts, but it comes back. What doesn't come back is Smart Bidding having learned that this traffic converts and spending the following months looking for more just like it. Ninja Shield cuts the problem where it starts: at the inventory that produces it.
What a Google report won't show you
A website built only to receive ads, an app opening itself in the background and a video nobody watches produce exactly the same rows as a real publisher: impression, click and, sometimes, conversion.
There is no column in Google Ads that says "this one was a farm".
The enemy
Fraud in 2026 is not a repeated click
Ten years ago fraud was someone clicking your ad. Today it is an industry with its own inventory — which is why blocking the odd IP no longer does anything.
Websites made for ads
Pages with no real content, bought traffic and as many ad slots as will fit. Their only customer is you, without knowing it.
Apps that open themselves
Applications loading ads in the background or behind another screen. Impressions no human eye ever saw.
Video with no viewer
Autoplayed, muted and off-screen — and it still counts, and it still gets paid.
Farms
Dozens of different sites that, underneath, share a server, an owner or an advertising identifier. Block them one by one and five more appear.
Automated forms
Programs filling in your form with invented data. To your account they are conversions, and the algorithm chases them.
And human-looking traffic
Visits that arrive, make one gesture and leave without touching anything. Neither an obvious robot nor a customer: paid filler.
How it defends you
Six layers between fraud and your budget
None of them is enough on its own, which is why they work together: each one watches a different signal, and what slips past one is caught by the next.
🎯 Performance scenarios
Rules over what has already been spent: where money goes with no result, which placement repeats the same pattern, which site has never converted. The trawl net: cheap and effective.
🔬 X-ray of the site
Before judging it by its numbers, the site is visited and analysed: what content it has, how much advertising, who owns it and which other sites share that owner. And this is the part almost nobody can do — see below.
👁️ Visit monitoring
What each paid visit does on arrival: whether it moves the mouse, whether it reads, how long it stays, whether it comes back. A behaviour score that tells a person from a program.
🧲 Quality by placement
What happened to the leads that came from each site. Not "how many", but how many were any good: the most honest signal there is, and Lead Rating provides it.
🕸️ Farm detector
Cross-checks server, network, owner and advertising identifiers across suspicious sites. When several match, they aren't sites: it's a farm, and the whole thing goes.
🌐 IP shield
Addresses that repeat the pattern are excluded from the account. The best-known layer and the least decisive one: it comes last on purpose.
The detail that changes everything
To analyse a farm website, it has to let you in first
Websites that live off fraud are not stupid: they know who is looking at them. And the first thing they block are the addresses of Google's datacentres, because that is where the systems that could expose them come from.
A script running inside Google Ads comes out of precisely those addresses. The result: the most suspicious sites are exactly the ones that refuse to be analysed, and end up filed as "couldn't be checked".
Our analyser doesn't live inside Google. It comes in from an ordinary Spanish IP, like any other visitor — and then the page answers and you see what's inside.
This is why the engine was taken out of Google: with a script inside the account, this layer simply cannot exist.
What you see once you're in
Real content
Whether there is original text or generated filler.
Ad density
How many ad slots per screen.
Who gets paid
The advertising identifiers it declares.
Where it lives
Server, network and domain registration.
And a risk score
With the reason written down, not just a number.
Network effect
A farm caught in one account protects all the others
Advertisers don't share customers, but they do share enemies: the same junk inventory shows up in accounts from industries that have nothing to do with each other.
Shared registry
When a site is identified as fraudulent it enters a common registry. It holds nobody's data: domains, networks and owners.
You start protected
A new account doesn't start from zero: it inherits inventory that has already been scored, so protection doesn't take weeks to be worth anything.
And it gets better
Every account that joins widens the registry. It is the one part of the system that grows stronger the more of us there are.
What is NOT shared: your leads. Nor your campaigns, your customers or your results. Only the registry of sites and farms — which contains no personal data.
Every day, in a loop
Watch, score, decide, block — and review what it blocked
Watch
It collects what happened in your campaigns and what visitors did on your website. Every signal, every day.
Score
Each placement, app, channel and IP gets a risk score with its reason: which signal fired and with what figure.
Decide, at the level you choose
You set how hard it pushes: from conservative — only the obvious — to strict. It isn't an on/off switch: it's a dial.
Exclude, and write it down
The exclusion is applied in your account and logged with the date and the reason. Nothing is blocked silently.
And review itself
A blocked site may stop deserving it. A periodic review pardons what no longer fits the pattern, so the list doesn't grow forever.
What you get to see
A panel you can audit
A system that blocks things in your account without showing why is an act of faith. This one shows everything it does.
Dashboard
What has been blocked, how much suspicious traffic comes in and how it evolves month to month.
Exclusion log
Every decision with its date, its reason and its figure. Reversible.
Score by site
The file on each analysed domain, with what was found inside.
Visit log
What each paid visit did, with its behaviour score.
What we won't tell you
Two promises that should make you suspicious
"We'll save you X % of your budget"
Nobody can know in advance how much fraud is in your account, because it depends on which networks you show on, your industry and your bid. What can be done is measure it in your account and show you.
"We block millions of IPs"
The number of blocked IPs is an indicator of nothing — Google only accepts a limited number of IP exclusions, and modern fraud doesn't live there. What moves the needle is the inventory.
Next step
How much of yours is traffic that never existed?
That question is answered by looking, not by estimating. Tell us how your account is set up and we'll tell you what can be measured and how long it takes.