IVT Detection & Blocking System for Connected TV

AdTech & Invalid Traffic Protection

2-week implementation

Appgrade (part of Embi Media) is an AdTech company specialized in buying and selling Connected TV (CTV) inventory. They purchase Smart TV inventory with Server-to-Server (S2S) connections and sell it to demand platforms that pay by CPM. Their main client is Xandr, a Microsoft platform.

Their main challenge was detecting and blocking in real time the Invalid Traffic (IVT) that some partners were sending, especially during the night. They already had 2 warnings from Xandr and one more warning meant their account would be shut down.

The Challenge

Partners sending IVT (Invalid Traffic) in Connected TV inventory during nighttime
Manual fraud detection the next day, when the damage was already done
Risk of Xandr account closure after 2 warnings for invalid traffic
Slow manual process: extracting Xandr reports and blocking bundles in LimeLight
Inability to monitor 24/7 the percentage of filtered requests per bundle

The Solution

Hourly Automatic Reports in Xandr

A robot that extracts Sell Fill and Delivery reports from Xandr every hour, analyzing Total Ad Requests, Filtered Requests and Site Domains to identify problematic bundles.

Real-Time IVT Detection

Algorithm that calculates the percentage of Filtered Requests per bundle. If it exceeds the configured threshold, it automatically identifies the fraudulent bundle.

Automatic Blocking in LimeLight

When a bundle with high IVT% is detected, the system connects to LimeLight via API and automatically adds it to the Global Block List without human intervention.

24/7 Monitoring

System running 24 hours a day, especially during nighttime when historically more fraudulent traffic arrived from partners.

Alerts & Notifications

Alert system via Slack and email when a bundle is detected and blocked, keeping the team informed in real time about actions taken.

Technologies Used

N8N (Workflow Automation)Python (Data Processing)Xandr APILimeLight APISlack API (Alerts)Webhooks & REST APIs

Project Timeline

Week 1

Analysis of current manual process, Xandr and LimeLight access, automation flow design

Week 2

Development with N8N: connection to Xandr and LimeLight APIs, detection and blocking logic implementation, testing and go-live

“

This system saved our operation. We went from detecting fraud the next day to blocking it in less than an hour automatically. We no longer wake up to IVT surprises during the night. The AdOps team can focus on optimizing campaigns instead of reviewing reports manually. Since implementation, we haven't received a single additional warning from Xandr and the quality of our inventory has notably improved.

Alan Resnik

CEO

Appgrade

Results Achieved

Warnings received after implementation

90%

Reduction in average IVT%

<1h

Bundle detection and blocking time

24/7

Automatic monitoring without intervention

100+