Ninja Scripts Academy
Google Ads from top to bottom: basic, intermediate and advanced. With real examples, best practices and the gaps the scripts fill. 🔄 Syllabus reviewed and updated every month with the latest Google Ads news · 🎧 Every lesson also in audio · 🖍️ Highlight and save your favourite lessons to come back to them.
🌐 Lessons are in Spanish for now; the English version is in preparation.
Ninja Scripts Suite subscribers get 100 % of the three levels, every new lesson each month and the study tools: 🎧 audio for every lesson, ⭐ favourites and 🖍️ highlights. See the subscription →
Showing only 🛡️ Fraud lessons at the Advanced level.
Advanced training
8 lessons · 1 h 18 minPutting limits on PMax: brand, negatives, placements, URLs and new customers 🔒 Subscribers
Every control lever PMax admits in 2026 — brand exclusions, account-level and campaign-level negatives, account placement exclusion lists, URL exclusions, the new customer acquisition goal, content and audience exclusions — what each one does, its limits and the order in which you set them up before the first euro.
The ad fraud ecosystem: who gets paid for your impressions, and why Google does not stop it entirely 🔑 With a free account
The players in ad fraud (site farms built for ads, apps living off accidental clicks, bots and non-human traffic, competitor clicks, proxies), how they make money out of your budget, what Google detects and refunds as «invalid traffic» and what it does not, and why the defence is the advertiser's responsibility.
Spotting farms and networks: the signals of a rubbish domain and how to group domains by their fingerprint 🔒 Subscribers
How a site is analysed (ad density, copied or generated content, no author and no contact details, domain age, ads.txt), the trust signals that clear a site (CMP, trust seal, company number, a real blog, social profiles, schema, an app, longevity, MX, DNSSEC), and the clustering by IP, name servers, RDAP record and publisher ID that turns a list of domains into whole networks.
Apps, IPs and non-human traffic: the other three doors into fraud 🔒 Subscribers
How apps living off accidental clicks are detected (anomalous CTR, zero-second sessions, filler categories) and how they are excluded by category or wholesale; account-level IP exclusions, their limits and how to feed them from your visitors' fingerprints; and the signals of non-human traffic on your site (browsing speed, no events, data centres, proxies) that a visitor script can measure.
Safe Sites, aggressiveness and calibration: protecting without losing reach 🔒 Subscribers
The Safe Sites whitelist and the «harvest campaign» that discovers them, the shield's aggressiveness levels (from conservative to maximum) and which thresholds each one sets in one go, the retroactive effect when you relax it (reinstating exclusions), the impact forecast before changing level, and how you measure whether the shield is improving your CPA or merely cutting reach.
Fraud in PMax, video and Shopping: where the rubbish gets into automated campaigns and how to shut it out 🔒 Subscribers
Why PMax is the farms' favourite ground (Display and video inventory with no control per campaign), Google video partners and bought views, apps inside PMax, fraud in Shopping (comparison shopping clicks and price bots) and the one defence they all respect: account-level exclusions maintained every night.
Ninja Shield in practice: installation, scenarios, the Dashboard and the false positives review 🔒 Subscribers
How the shield is deployed (installed by the team from the MCC, with its own spreadsheet per client and the visitor snippet on the website), its modules (Site Analyzer, Farms, Visitor/IP/App/Lead/Video Shield), the scenarios and Smart Mode, the Dashboard showing fraud blocked in euros, the false positives tab with its checkbox, and the client's routine.
The lead's fingerprint (layer A): what your site knows about every contact before anyone has called them 🔒 Subscribers
The signals a snippet captures at the moment of the form — the fraud score inverted, time on page and real scrolling, pages viewed, returning visitor, typing cadence, time of day, device, field coherence (a valid phone number, the email domain, a plausible name), honeypot — how they are weighted inside the ~40 behaviour points, and how the GCLID is captured without breaking privacy.