2019 Awards Preview: Most Effective Anti-fraud Solution

Ahead of our 2019 Effective Mobile Marketing Awards, staged in partnership with our headline partner Dynata, and our partners DAX and TabMo, well be previewing the nominees in each category, giving you a glimpse at the high quality of entries weve seen this year. In today’s preview, we look at the best anti-fraud solutions out there.

Captiv8: Verified Reporting
With around 11 per cent of engagement on sponsored influencer Instagram content coming from fraud/bot accounts, Captiv8 wanted to develop reporting tools that detect fraudulent behaviour and provide transparency around campaigns.

It developed Verified Reporting, a machine learning-powered solution that guarantees content is viewed and interacted with by real people. The solution views follower counts to analyse any suspicious changes in follower activity and engagement quality.

Verified Reporting also verifies traffic sources by platform to identify fraud. It does this by identifying changes in geographic distribution of engagement of audience on a per post basis and compares this to posts using regression analysis, identifying odd patterns or fraud indicator behaviours.

Captiv8 provides the solution for every creator. Earlier this year, it teamed up with Horizon Media, becoming the ‘first’ company to provide independent measurement and reporting for agency influencer marketing campaigns.

FraudScore: FraudScore.Action
FraudScore is an independent anti-fraud solution for both mobile and web traffic. Its solution is being used by Zorka.Mobi, a creative mobile performance agency, to check the traffic that it gets for its partners’ apps.

Zorka.Mobi had to find a strong ad fraud detection solution for Russia, with the region being known as one of the most fraud prone traffic sources around the world. It had already faced difficulties from affiliates mixing a share of fraudulent traffic but was unable to provide detailed reporting on the issue, as most fraud detection solutions only provide an evaluation of fraudulent behaviour and don’t drill down into the data. This is where FraudScore stepped in.

FraudScore.Action is used by Zorka.Mobi for CPA campaigns post-analysis for in-app ads. The product is developed for CPI (cost per impression), CPA (cost per acquisition), CPL (cost per lead) traffic analysis and is used by advertising, CPA networks, and advertising agencies.

From September 2018 to December 2018, FraudScore reduced high index fraud from 26 per cent to 2.3 per cent. Zorka.Mobi used reports from FraudScore.Action to track traffic daily and warn affiliates about the need to stop using fraudulent sources.

Impact: Forensiq
Forensiq is a full-funnel ad protection suite aimed at helping clients to gain trust and transparency along the mobile and desktop media buying chain. Acquired by Impact in 2016, it has received Trustworthy Accountability Group (TAG) and Media Ratings Council (MRC) accreditations in the last couple of years.

Forensiq’s solution does not rely on SDK integration and is built to detect in-app specific fraud tactics. It also combines big data analysis with browser-level insights and server-side filtration, providing full funnel fraud detection with the help of AI and machine learning algorithms.

It uses three essential types of detection in the form of browser and device analysis, behavioural and network analysis, and data mining, AI, and machine learning techniques. Within its platform, Forensiq has a traffic quality dashboard, an ad viewability dashboard, attribution risk dashboards, mobile attribution risk protection, and its reason codes.

One of Forensiq’s clients, StackAdapt, demonstrated a 91 per cent fraud risk improvement over TAG’s 2018 benchmarks and a 41 per cent risk improvement over TAG’s 2018 Certified Channel benchmarks.

Lemmonet wanted to deliver 100 per cent fraud-free traffic to their global performance clients. With Forensiq, they score 200m clicks per month, and have seen an 86 per cent reduction in click fraud and an 87 per cent reduction in install fraud.

Integral Ad Science: pre-bid optimisation
In a bid to make a stand against the $8.2bn lost to ad fraud each year in the US alone, according to IAB figures, Integral Ad Science (IAS) developed tools and established a dedicated FraudLab to make sure ads are served by real publishers, show to real people, and are reach the right target audiences.

It was this kind of solution that one of IAS’s customers was looking for. The FMCG giant with brands across food and petcare wanted to increase the value of its campaigns but was being affected by buying impressions that reach non-human audiences. The FMCG brand was seeing 9.1 per cent of programmatic display and 9.8 per cent of programmatic video impressions succumbing to illegal bot traffic and other forms of fraud.

IAS offered a solution to the brand through its pre-bid ad fraud protection, which eliminates fraudulent inventory before it can be bid on. It worked with the brand, as well as programmatic buyers, to create risk thresholds that achieved the quality and quantity of ad space required by the brand. IAS worked with DSP platforms to provide protection from both GIVT and SIVT fraud.

The pre-bid optimisation led to an 87.7 per cent decrease in fraud levels. This meant the FMCG brand was able to maintain a fraud level of around 1.3 per cent through continuous monitoring and fraud prevention, avoiding over 32.2m fraudulent impressions.

Machine is the ‘only’ mobile-specific full-stack solution, analysing data at pre-bid, impression, click, install, and post-install level. It boasts clients including Paddy Power, GrubHub, Rakuten, and Trainline, to name a few.

Since winning this category at last year’s awards, the company has grown 120 per cent, going from analysing over 4m app installs a month to analysing more than 10m app installs every month.

Machine analysed over 71.2m app installs from 115 networks in the first half of 2019. It detected that 51 per cent of these app installs were fraudulent. This meant it removed approximately $182m worth of fraudulent app installs. Of the 36.7m fraudulent installs, 63 per cent were farmed installs, 33 per cent was attribution fraud, and four per cent was incent fraud.

Machine learning-powered mobile ad fraud detection solution Scalarr was started in 2016 with the idea to develop an ‘efficient and highly’ accurate tool to detect all mobile app install ad fraud. It set out to do this because it found that existing anti-fraud solutions were only able to see ‘primitive’ fraud and unable to detect new patterns and modifications.

The Scalarr team is made up specialists in data science, machine learning, and mobile advertising. These specialists work to ensure that customers do not waste marketing budget on fraudulent non-human traffic and get ‘real’ LTV (lifetime value) and ROI (return on investment) metrics.

The company first introduced machine learning to its fraud detection in June 2017 and by September 2017 it had been joined by its first clients including Aviasales, Pixonic, and Goodgame Studios. A year later, it introduced a neural network model to uncover smart types of fraud.

In 2018, it investigated 170m installs and analysed 22bn events, uncovering 35.7m fraudulent installs. In this time, it also found 18 new fraud patterns.

Smaato: In-app fraud protection solution
Smaato’s programmatic exchange is designed to safeguard mobile investments from all external threats, including malware, click fraud, bots, and more. It uses a combination of the latest technology and dedicated experts to provide advertisers with access to ‘high-quality’ traffic and ‘authentic’ audiences for their campaigns, while working to keep publishers safe from inappropriate ads.

Its in-app fraud protection solution use a three-pronged approach: in-house technology, expert staff, and trusted third-party fraud protection vendors. It manages almost a trillion ad requests every month and uses machine learning to automatically detect suspicious behaviour.

Protected Media recently made use of Smaato’s solution to analyse 40bn impressions across 10 mobile DSPs and exchanges. It found that in-app fraud attempts occurred 40 per cent less often on the Smaato platform, compared to other DSPs and exchanges.

Smaato was also recently awarded the Certified Against Fraud seal from TAG.

TrafficGuard: Ad fraud prevention
Providing real-time ad fraud prevention for brands, apps, ad networks, and agencies, TrafficGuard aims to mitigate ad fraud to address all its associated costs. Its solution looks to make sure ad fraud never gets paid for by taking a proactive approach that helps everyone in the supply chain.

TrafficGuard sits within the advertising journey, analysing impressions, clicks, installs, and other conversion events as they occur. This means it can invalidate traffic at multiple stages in the journey, whether that’s before the attribution or at the click.

It looks to use a surgical approach to reduce false positives by only removing instances of invalid traffic, leading to improved valid traffic volumes for advertisers and overall better performance. It also shares IVT diagnosis with advertisers and supply partners to ensure transparency. And it’s been built to work alongside any mobile measurement platform (MMP) or with no MMP at all.

Since launching TrafficGuard SaaS in 2018, over 2.6bn transactions have been invalidated, representing estimated savings of $8.5m and average client ROI of 3770 per cent. It boasts customers including Rappi and Go-Jek.

Rappi has been able to mitigate an average of 25 per cent of their click traffic, peaking at 40 per cent on some days. It has also seen return on ad spend (ROAS) improve 25 per cent through reduced waste in ad spent and improved performance, as well as improving LTV and install to first order.

The Effective Mobile Marketing Awards Ceremony takes place in London on 14 November. To book your place at this celebration of the best in mobile marketing, click here.

And remember, there’s still time to vote in our special 10th Anniversary Awards for Most Effective Brand of the Decade; Agency of the Decade; and Disruptor of the Decade. Click here to vote now.