Digital ad fraud has already surpassed credit card fraud, and it’s expected to reach $100 billion in 2023.
And where there’s money, there are scammers.
While ad fraud was a problem in pre-pandemic times, the pandemic caused a surge in OTT viewership that proved irresistible to fraudsters. More hackers are migrating to programmatic CTV advertising and OTT advertising (tempted by premium CPMs). Schemes with ominous-sounding names like Streamscam, Monarch, and DiCaprio ate up millions of dollars last year.
Most fraud prevention solutions are built to be binary — an impression is immediately deemed to be fraudulent if a certain signal is or is not present. While this is definitely a start, today’s scammers are more sophisticated, and avoiding CTV ad fraud will take more smarts. We’ve put together this primer on how to prevent ad fraud based on our own best practices to help you avoid even the most intricate fraud schemes.
CTV/OTT fraud almost always originates via ad requests (whenever a website requests ads to be displayed).
At Madhive, we validate these requests using advanced statistical modeling. We track the ad impressions served via server-side ad insertion, and then match those to the number of ad impressions that actually play on a user’s end device.
This means we’re able to identify discrepancies early on — and it prevents us from bidding on inventory that seems suspect.
The statistical modeling that detects fraud attempts requires speed and scale.
Most ad tech companies analyze a sample of one out of every 10 or 20 requests — that isn’t nearly enough. We ingest the entire live bidstream and analyze billions of fraud signals in real time, using machine learning to understand data patterns.
In most cases, third-party verifiers aim to weed out fraud by looking at bot traffic and behavior, but they don’t have access to the supply chain on the transaction. Madhive’s infrastructure is able to gain a robust view of billions and billions of signals that we use to train our machine learning algorithms.
We’ve simplified our ad tech stack by eliminating the need for extra vendors and their fees.
We are able to view the entire interaction from bid to impression because of our direct relationships with publishers, including VAST tags. When you use a point solution to try to block fraud, that point solution usually can’t say where the request came from — we can.
What do Publisher IDs, Quartile Pixels, VCR, Price Floors, and Supply Paths have in common?
They’re all fingerprinted by Madhive. Yes, it takes a lot of data and machine learning, but it’s worth it.
Fingerprinting each supply path not only optimizes price, it also identifies invalid supply. By assigning each supply path its own unique “fingerprint,” it becomes clear who the real media owner is. A fraudster cannot duplicate all of the intricacies of a real provider, so the subtle abnormalities immediately alert us that it’s fake.
To take it one step further, we keep a vast library of fingerprints and regularly compare them with supply path data from several billion OTT ad buys. We determine these known supply paths through various methods, like:
With this wealth of data, we can compute — in real time, pre-bid — an accurate identity score that defeats both simple fraud attempts and sophisticated, lucrative attacks.
We are passionate about transparency and about participating in open source standards that make digital media a more transparent and accountable place. We employ Double Verify, Human and Maxmind prebid filters on 100% of inventory, meaning we filter out geolocation and fraud instances before any impression requests are exposed to potential bids.
In addition, we are Certified Against Fraud by TAG - the Trustworthy Accountability Group for the media industry.While most ad tech vendors opt for the easier self-attestation route, we decided to invest the time and effort to achieve this accreditation through independent validation. We had an independent auditor certify that we successfully implemented the Media Ratings Council’s General Invalid Traffic (GIVT) and SIVT guidelines into our product.
As streaming continues to gain popularity and advertisers increase their spend, there has never been a better time to shore up your OTT campaigns against ad fraud.
We certainly don’t want the rampant fraud in digital advertising to show up in programmatic television, too (just think of all the types of ad fraud happening with display ads!).
Thankfully, the ad tech industry is getting smarter to better combat fraudsters. We can overcome past hurdles by using more advanced tools like AI and cryptography, as well as improving transparency across the tech stack.
The outcome is a programmatic advertising industry that works better for everyone — advertisers, publishers, and viewers alike.
