Bots form 80% of ad fraud for travel industry, overall loss for APAC to hit US$56bn

Bots make up 80% of all invalid traffic for travel advertisers, a stark contrast to the 15% to 30% affecting other industries, according to internal research by Trafficguard. This comes as the total loss to mobile ad fraud in APAC is expected to hit US$56 billion this year, from US$19 billion in 2018, separate research by Trafficguard and Juniper Research revealed. This amount is expected to increase due to the return of travel in the region.

A series of factors could be attributed to this trend of ad fraud in the travel scene, the first being the travel industry having a proliferation of third-party online travel agencies (OTAs). Some of these OTAs are authorised and some aren't and they scrape the data of airlines and hotel databases before selling them on behalf of airlines and rentals.

The travel industry also has many high-value loyalty programmes that fraudsters will try to assume control of, Trafficguard said. It added that many travel vendors use apps and mobile experiences to reach their target customers and by trying to drive installations of their apps, they are more susceptible to receiving bad quality traffic and misattribution.

As airline websites and OTA travel apps are home to a host of data such as flight, pricing, booking, and discounts, bad bots can result in higher fees for OTAs as they make it appear as though far more people are viewing than booking flights.

Trafficguard's research found that many travel companies will spend large sums of money to drive consumers to download their travel applications. They will often use pay per install performance networks that are paid for each installation they drive.

Performance networks that are being paid to drive these installations are delivering upwards of 50% invalid traffic on average.

This results in many fraudulent installs and misattribution of post-install events such as flight and hotel bookings.

These numbers might be a concern for Southeast Asian advertisers since online travel spend is expected to rise 36% from US$13 billion last year to US$43 billion in 2025, according to the "e-Conomy SEA 2021, Roaring 20s: The SEA Digital Decade" study by Google, Temasek, and Bain & Company in 2021. At the same time, advertisers in the region would also have to be more cautious about ad fraud given that countries in Southeast Asia have eased travel restrictions and are gradually reopening borders. 

Malaysia, for example, will reopen borders on 1 April while Singapore announced last week that it will allow all fully vaccinated travellers from any country or region to enter quarantine free, as long as they have not visited any countries or regions on the Ministry of Health's Restricted Category. 

(Read also: How are the different markets in SEA luring back tourists?)

Ways to curb mobile ad fraud

Aside from the astounding number expected to be lost to ad fraud in APAC, SDK Spoofing and Install App Farms will continue to be a major concern for mobile app advertisers. According to Juniper Research, this issue is expected to increase, with one in 10 app installs predicted to be fraudulent by 2022. Juniper Research also estimated that an advertiser displaying a million ads in a 24 hour period will pay for more than 100,000 fraudulent ads on average before detection.

Machine learning is crucial in detecting and mitigating ad fraud at several levels. According to the research, only machine learning is capable of analysing the volumes of data required to predict the likelihood of fraud in real-time. Through a combination of unsupervised, semi-supervised, and supervised machine learning, fraud tools have the ability to detect anomalies, perform predictive modelling and find clusters to identify new and earlier indicators of fraud.

To curb ad fraud, here are some areas brands should consider when implementing fraud mitigation solutions as listed by Juniper Research.

1. Stage in the journey

This is the stage in the ad journey where ad fraud detection or prevention takes place, including impression, click and attribution. As anti-fraud capabilities tend to be an add-on to other products, Juniper Research said many solutions available today have a restricted focus on a small proportion of the advertising journey dictated by what their technology was initially developed to do.

2. Detection capabilities

Automated rule engines filter ad fraud based on a relatively static set of rules. More sophisticated solutions employ some level of statistical algorithms. According to Juniper Research, these approaches have a limited ability to detect fraud that has not been encountered before. With ad fraud constantly evolving to avoid detection, this leaves ad spend exposed.

3. Blocking capabilities

The ability to block in real-time is crucial to minimising the loss to fraud that can occur at multiple levels of the ad chain. Hence, it is in the advertisers' best interest to review the extent of their chosen platform's abilities.

According to Juniper Research, the best practice is to block ad fraud as early as reliably detected and to enable blocking at various stages. Early stage prevention can be evaded more easily as fraud tactics become increasingly sophisticated, making later stage detection at the click or attribution level vital in defending against sophisticated ad fraud.

4. Reporting capabilities

Sophisticated reporting offers clear reasoning for ad fraud diagnosis and tools which extend reporting to intermediaries, as well as advertisers, offer the whole supply chain a chance to mitigate ad fraud instead of just giving refunds when ad fraud is reported.

Photo courtesy: 123RF