Perfecting programmatic: Integral Ad Science's Niall Hogan on brand risk and viewability
Integral Ad Science UK managing director Niall Hogan asks, can programmatic and high media quality coexist?
It has been suggested that programmatic media is the fastest growing phenomenon in the global media and advertising industry. A recent IAB/PwC report found that programmatic trading in the US reached $10.1bn in 2014, making up 20 per cent of the near $50bn total US internet ad revenue for 2014.
This brave new world may bring a whole host of benefits in time-saving, targeting and revenues, but can create a whole new world of complications for media quality when you see that rates of ad fraud, and the risk associated with brand safety, almost doubles when working in programmatic channels rather than publisher direct. Viewability rates also take a nosedive.
According to the Q2 2015 Integral Ad Science UK Media Quality report, brand safety risk more than doubled through programmatic networks and exchanges versus publisher direct, from 6.3 per cent to 13.6 per cent. 13.6 per cent of impressions within programmatic channels were found to be fraudulent, compared with 6.6 per cent for publisher direct buys. Meanwhile, viewability decreased from 63.4 per cent to 52.1 per cent.
Before stepping into the vast world of programmatic, the first step is to understand why there is greater media quality risk seen in programmatic trading.
Firstly, programmatic brings greater automation and reduces the human interaction in the process of buying and selling of media. This absence of the human touch has allowed for an environment that can be gamed by bad actors; they place poor quality or fraudulent inventory within programmatic exchanges and networks. They take advantage of the fact that we are now using platforms to purchase and execute media deals, sometimes bidding on websites we have never heard of, and no longer calling a known and trusted sales executive at Yahoo, Mail Online, or AOL.
Secondly, the programmatic exchanges pool huge amounts of inventory; they are working with hundreds, sometimes tens of thousands of websites. The scale of inventory is difficult to continuously monitor for quality, and again, this leaves the door open to bad actors that are able to game the trading system. Exchanges typically include inventory from the entire supply ecosystem. Premium inventory can be found, but mostly supply is from the medium-long tail. There is no guarantee that this inventory is professionally produced. It is here that we find the largest risk of fraudulent activity and of ads appearing next to inappropriate content, and the lowest opportunity that ads will be seen.
So, is it possible for high quality media and programmatic trading to coexist? To address this challenge, we have seen a trend of buyers and sellers coming together to set up private marketplaces. While they can achieve higher viewability rates, they do still fall prey to media quality issues such as brand risk and fraudulent activity (our Q2 2015 UK Media Quality report found that with publisher direct deals, six per cent of impressions posed a medium to very high risk to brands, while seven per cent of impressions were fraudulent) and more importantly they negate the real potential of programmatic buying; the combination of using first and third-party data to target consumers over a large-scale reach. Private marketplaces are a work-around and do not hold the answer.
So if private marketplaces are not the answer, and we have concerns over the quality within an open marketplace, what can be done to mitigate risk? The answer to finding harmony between programmatic and media quality is working with impression-level data to inform the programmatic buy and ensure that only quality impressions are bid on.
Media quality data provides historical information about a webpage, such as information from the meta-tag that describes the images on the page. This data is recorded, collated and stored so that we understand the page content and can inform any potential buyer before they make the bid. To avoid targeting inappropriate content, buyers can set their brand safety risk threshold from medium to very high risk.
We can also monitor how many times the ads on each page were viewed, and for how long. This historical data is used to predict future viewability and helps the savvy buyer to find ad environments with the highest percentage chance of being viewed amongst the billions of impressions available.
Finally, we can also monitor impression-level activity for signals of fraudulent behaviour, such as fake browsers requesting impressions that will not be seen by humans, or masked URLs pretending to be well-known sites. When we see this suspicious activity, we know not to serve an advert to this browser or website and through data segments integrated within DSPs we inform the buyer not to bid on these impressions. Programmatic buyers can improve the quality of media by using data segments to ensure that only quality impressions are bid upon.
Although the quality risks are real, the industry continues to increase its investment in programmatic advertising because it can deliver great results and help improve eCPMs. This was highlighted in a recent IAB Europe report, which found “over 90 per cent of advertising executives plan to increase their programmatic investment”.
It is the responsibility of everyone in the digital industry to protect and nurture this investment, and we can achieve this when buying programmatically by using media quality data alongside audience targeting data; ensuring that we find the best audience, and that this audience is able to consume the message within an environment that is appropriate to the brand or service.
This was first published in The Drum's 16 September issue.