Busting the myth of CPS, by mFilterIt’s Dhiraj Gupta

Partner Content: It is not necessarily rewarding to change the engagement model for e-commerce selling, says mFilterIt’s CTO & co-founder Dhiraj Gupta.

Dhiraj Gupta, CTO and co-founder, mFilterIt

E-commerce is booming in the GCC, and the Covid-19 lockdown only boosted this machine. As brands look to increase their share of a more-than $20bn opportunity in 2020 (Bain & Company estimates), e-commerce is also inviting the prying eyes of the fraudsters. To stay safe, advertisers have been gambling with cost-per-download, cost-per-install and cost-per-action in the past, and now they are making it look difficult for fraudsters with cost-per-sale (CPS) as the KPI.

But, just like changing passwords, changing the performance campaign model does not guarantee freedom from fake engagement in terms of clicks, views, actions or sales. Many digital leaders believe that by refining KPIs and making them harder to achieve, fraudulent activities get filtered out automatically. That is never the case. While the old saying of ‘when the going gets tough, the tough gets going’ often has positive connotations, unfortunately in this case fraudsters also get tough with changing metrics.

mFilterIt has several examples where it was deterministically proven that changing the metrics model did not yield the results expected, and the advertiser could not achieve pure results. Instead, it made ad fraud more lucrative as the incentives also went up.

The fundamental question here is: If the pay-out is on actual sale in the CPS model, how can there be any fraud? Sale is tangible, and only after the advertiser is convinced would the publisher be rewarded with the commission. It’s a straightforward, success-based fee model, right?

Well, there is no denial that the publisher earns only after someone buys a product or service. But the issue is about paying twice, as well as attributing the credit wrongly. In most CPS ad-fraud tricks, fraudulent publishers use methods such as cookie stuffing and click injection to steal the transaction and take credit for the sale. Even when the user has organically discovered the e-commerce platform and initiated a transaction to buy a product or service.

To get more technical, a user engages with an advertisement put up by an affiliate and ends up with a cookie getting stored on the device, which hijacks the attribution at the time of the user actually visiting the shopping site. The net result is that in the eyes of an advertiser the sale takes place due to the efforts of a particular affiliate. The affiliate gets paid for the trick rather than putting in the genuine effort, and the advertiser bleeds doubly – first through their spend on organic advertising and then by paying the affiliate for the sale, which would have happened anyway. The cost per sale automatically swells and also gives an incorrect picture of organic digital marketing efforts, which are mostly driven by the in-house team of an advertiser.

The proportion of money spent by an advertiser on organic digital marketing and the commission paid to an affiliate varies substantially from one brand to another. It is also determined by the product and the industry sector. However, mostly the commission paid is higher than the spend on organic advertising. mFilterIt’s analysis pegs the average additional spend that an advertiser must incur at 34 per cent. This simply means that instead of paying $1 for a sale to happen, the advertiser spends $1.34.

Other ramifications of this issue include:

a) Many associates of a brand bid on the brand keywords of their principals, so that their links rank high on the Google search results and increase the probability of users clicking their links to visit the principal’s e-commerce platform. This further increases the bid cost for the advertiser and also triggers organic stealing, which pushes the gross CPS upwards.

b) Another tactic employed is clickbait, which is sheer misrepresentation of creatives across walled-garden websites. This is a blatant abuse of brand guidelines and poses issues around brand infringement and financial abuse.

Recent research from mFilterIt confirmed the levels of bot activity and fraud in retargeting, re-marketing and partner-led marketing campaigns. These campaigns are integral to e-commerce. While the average fraud levels are in the range of 40-50 per cent, there are cases registering 99 per cent fraud regularly. The bots used in such campaigns are fake devices that get a lot of impressions. CPMs for retargeted ads are relatively higher than vanilla display ads. Hence the bots fire a lot of clicks. For instance, we often see ads getting more than 100 clicks from one ‘user’. But humans would rarely click on an ad more than a handful of times.

Now, even if the marketer tries to be smart by using CPS as a KPI, the bots go one step further and manipulate the attribution to get credit for the sale. This is done to sales that happen voluntarily, and hence organically. This organic stealing costs marketers millions of dollars.

E-commerce is not just an alternative to normal modes of purchasing. It is a behavioural change that is brought about over a period by e-commerce champions. For them, it’s critical to get the right picture and understand how buying behaviour is changing. While organic and inorganic push must co-exist, marketers will always be interested in discovering about the users who are habitually buying online so they can engage with them and even nurture communities. The real measure for this engagement is the organic route, and with CPS ad fraud this is misrepresented from the beginning.