How to spot a Fraud Prevention Specialist

September 4, 2018
Posted by

If you work in digital advertising, you likely get bombarded daily with scary fraud forecasts, pushed by measurement companies or brand safety verification vendors that have recently decided they want to get a piece of the ad fraud pie. Rather than specialising in fraud prevention, these companies do something else and also offer fraud prevention. These Also-Offerers use words like “leading”, “machine-learning”, “click injection” to try to convince you that they know what they are doing.
But how can you tell a company that knows its stuff, from an
Also-Offerer that is just full of BS? Read on to get an understanding of what a real fraud prevention specialist looks like!

Multi-Point Analysis

An also-offerer just looks at a single stage of the user journey ie the impression, click or install attribution. This is because their technology or primary service is something other than fraud prevention, and that other function deals with that specific single transaction point. For example, mobile measurement platforms that also offer fraud protection, only detect fraud at the install attribution stage – why wait for the install if you can spot the fraud earlier? Brand safety vendors that also offer fraud protection, only detect fraud at the impression, because that is the level they determine brand safety at. The problem here is that sophisticated invalid traffic can easily evade impression level detection.

Fraud prevention specialists block invalid traffic at multiple levels and block it as early as possible. As fraud tactics become more sophisticated, the more data points your fraud protection tool is processing, the faster sophisticated invalid traffic (SIVT) can be identified and blocked. If the tactic is so sophisticated that it evades detection until the install attribution, a fraud prevention specialist will work towards finding earlier indicators to block it sooner. How does it do that?

Sophisticated Machine Learning

Machine Learning (ML) is a term that appears on many an Also-Offerer’s website, but is probably not genuinely employed by many. So, how can you tell who is bluffing?

The purpose of machine learning in fraud protection is to build a model of what normal behaviour looks like, and then use it to determine anomalous behaviour. In this way, effective machine learning helps to identify traffic that is not valid.
using machine learning as a buzzword don’t talk about valid traffic – they only talk about the bad, invalid traffic. The traffic that is easy to explain, and has names invented by marketing teams. They give you big, scary numbers on the amount of ad stacking, click flooding, or other fraud du jour they have captured.

Focusing on the bad is a flawed approach. If you are only looking for tactics #1 and #2, you can’t see when tactic #3 is introduced. This is also why the effectiveness of blacklisting is so limited – it is a great first line of defense, but a blacklist can only block what it knows is invalid. By contrast, a fraud prevention specialist focuses on what valid looks like in order to determine what is invalid, blocking fraud tactics #1, #2, #3, #4, #n.

Score-based validation

Fraud mitigation is not as easy as blocking everything outside an install time window as many non-specialist Also-Offerers might have you believe. Signature-based fraud identification on its own leads to high numbers of false positives – that is, valid traffic mistakenly blocked as invalid traffic. False positives can be as detrimental to your advertising effectiveness as fraud is.

Fraud prevention specialists take a score-based approach to validation. The score is something that builds from the very first impression through all transactions on the user journey but it doesn’t stop there. It is impacted by traffic suppliers, sub-sources, IP level, and machine level over every transaction across multiple advertising campaigns and apps over time.

This means when traffic is invalidated, it is not based on one indicator but on as many as 200. Score-based approaches are common in leading cyber and network security technologies, and now also employed in specialised mobile ad fraud prevention solutions, like TrafficGuard, to effectively mitigate both ad fraud and false positives.

One who concentrates primarily on a particular subject or activity; highly skilled in a specific and restricted field

In every other industry, businesses enlist specialists to combat fraud and safeguard security but for some reason in digital advertising that hasn’t been the case. The main options for fraud mitigation to date have been companies that perform some other related digital advertising function, extending their offering to also offer ad fraud protection.
don’t specialise in fraud prevention, they have some other area of expertise and have adapted their tech rather than purpose-building. They also often have some sort of conflict of interest or misalignment of incentives between their primary function and fraud prevention.

Too much money and time are wasted on ad fraud to continue using fraud mitigation from Also-offerers. If they aren’t specialists, your fraud coverage is being compromised.



Subscribe now to get all the latest news and insights on digital advertising, machine learning and ad fraud.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

Related posts

Get clued up on invalid traffic (IVT) and the ways our ad fraud protection is helping marketers fight back.

September 19, 2023

How to use TrafficGuard’s analytics to improve Performance Max

Make game-changing optimisations using valuable insights
Learn more
August 25, 2023

Six tips to get the most from your Google Performance Max campaigns

Aaron Chambers lets you in on all the PMax secrets...
Learn more