While viewability and fraud continue to be two of the industry’s most contentious issues, occasionally the industry’s ad tech solutions have only fuelled tensions between buyers and sellers, as the advertiser gets one set of results, while the publisher gets another. Here Niall Hogan, UK Managing Director, Integral Ad Science, explains why these discrepancies can sometimes occur, even when both sides are using the same vendor. Hogan also discusses the blind spots for tech tools and the possibility of fraud on programmatic TV.
Publishers are increasingly using viewability products, but many find they get different results to advertisers, even when both sides are using the same tools. Could you explain why this happens?
A couple of years ago, the main reason for viewability discrepancies was due to different methodologies of viewability tech vendors. However, through the work of the MRC and the ABC, the technologies of leading vendors are now more or less aligned. Discrepancies between these vendors are today very small at 1 percent or 2 percent variance.
However, discrepancies do still exist, and there are two main factors. Firstly, when a particular technology vendor, such as Integral Ad Science, is being used by both the advertiser and the publisher, we can still see discrepancies, at times as large as 20 percent. Because we are counting two different things. Publishers measure using the publisher pixel, which is within the ad container, and this fires up as soon as the container is available on a loaded web page. For the advertiser, they measure from the ad call, this only fires once the creative is in the ad container. There may only be hundredths of a millisecond between them (more if the ad is a creative heavy file) and this drives the discrepancy.
Secondly, discrepancies can appear due to fraudulent activity. Technology vendors’ ability to detect fraud and then removal (or not) of these impressions from the count is also a major determining factor.
A very basic fraud product will use a black list, a list of blocked sites, to try and identify a bot, and block impressions only once it is seen. At Integral Ad Science our fraud models are sophisticated and advanced so we identify more fraud and remove more impressions from the count, again this can drive discrepancies.
To reach true viewability measurement, there must be reconciliation of the two counting methodologies. At Integral Ad Science we have worked hard to deliver a product, Publisher Expert, that can measure both and reconcile, meaning a product that can work for both the advertiser and publisher.
Where are the blind spots for viewability/fraud detection tools?
By definition a blind spot is something we don’t know about. Fraud detection tools are not blind-sided as it is difficult to define due to “unknown unknowns.” Online fraudsters are continuously adapting their techniques, and technology vendors are aware of situations where their fraud detection may be faltering, or of types of fraud that they don’t address.
Beyond that, I think domain spoofing is the next big fraud topic to come to the forefront of the fraud challenge.
Viewability measurement presents many technical challenges and can produce limitations. Certain situations where vendors cannot fully measure viewability includes:
- Cross domain iFrames with Safari as the browser
- Browser where there is cross domain iFrame and flash is not enabled
- Certain iFrame busting rich media ads and custom formats.
A recent Exchangewire Research study found that the majority of respondents reported that they considered up to five percent fraudulent traffic to be acceptable. Is this a realistic expectation in video advertising?
With maturing detection and prevention tools, five percent fraud is a reasonable goal for any category in any channel. It is not necessarily reasonable to expect an industry to achieve that across the board however, because not every advertiser will be interested in paying to avoid fraud and will continue to be victimised by it.
However, anyone in the advertising industry who is keen to avoid fraud in their campaigns will have the tools necessary to achieve 5 percent – it may not be reasonable right now, but soon it will be.
Do you have plans to launch services for programmatic TV advertising? Will the TV industry encounter problems with bot fraud as it moves online?
It is likely that all online channels will be susceptible to fraud, as fraudsters will operate wherever there is financial gain. A greater numbers of vendors will look to provide targeting segments for programmatic TV campaigns and certainly brand safety, viewability and fraud will be major considerations in programmatic advertising.