The topic of fraud in digital advertising has accumulated a large percentage of attention over the course of the past couple of years. As mobile ads continue to score highest engagement rates, it is increasingly important for the marketing ecosystem to tackle the challenges bots and uncertified inventory represent. For mobile programmatic, location traffic is at the essence of successful campaigns, yet adsquare’s internal analysis shows that only 29% of this traffic is usable in hyperlocal campaigns. 

In our aim to transform audience data modelling, we focus all of our efforts on building a transparent product that puts programmatic buyers in control. This is why fraud prevention and quality assurance are at the centre of adsquare’s product. From day one, we’re building a solution that helps marketers successfully navigate through an ecosystem that is unfavourably exposed to fraudulent inventory.

At this point it is important to note the difference between inaccurate and imprecise traffic, and the traffic that is considered fraudulent. Precise traffic contains data that can be used for hyperlocal campaigns, such as GPS data. Accurate traffic, on the other hand, is the fraud-free traffic that advertisers can use to effectively target customers. This challenge becomes even tougher as a growing number of publishers add even inaccurate or imprecise lat/long to the bid stream due to the fact that inventory with location correlates to a much higher eCPM.

At the core of this fraud prevention lies our Location Qualifier tool, which separates precise geolocation traffic from imprecise and inaccurate traffic. To ensure we are on top of managing fraudulent activity, our data scientists developed three-layered algorithms:

1. As we work with hyperlocal information, each lat/long corresponds to a nearly unique numerical definition. By looking at the pool of locations and detecting numerical anomalies, we can sort out the ones that are not valuable for hyperlocal campaigns.
2. The next step to ensure we prevent inaccurate bid requests is to deploy a blacklist of known hotspots (city centres, telco hotspots, etc). This blacklist is constantly monitored and expanding.  
3. Machine-learning techniques – Our intelligent pattern recognition system observes our databases to detect unusual behavior and prevents bots from penetrating into data streams.

In addition to this, we continuously add new techniques to tackle fraudulent activity. One example of this is our frequency-based solution, which filters users via the frequency of their appearance on our platform, providing an additional layer of trustworthiness for our customers.

Going forward, we will be actively expanding our set of anti-fraud measures to rule out fraudulent behaviour and help customers run effective, highly precise and accurate campaigns. Quality, transparency and control are vital in bringing more marketers to the mobile programmatic space, and adsquare is at the forefront of the goal for delivering high-end content.