12 Oct 2020

Click fraud and viewability in 2020

By Daniel Schmidt, Ad Operations Lead at Ole!Connect

Viewability has been a buzzword in the advertising industry for a number of years now. Clients want their ads to be seen by prospective buyers and agencies have been selling them 'viewability solutions'. It's interesting to note that agencies use demand-side platforms (DSPs) as ad serving tools and they rely on their DSPs to not only ensure that ads are viewable, but also to ensure that viewability is reported accurately, which can be a tough task.

What is viewability and how is it affected?

Viewability is measured by how often an ad is viewed over the number of times it is served. If an ad was served 100 times and viewed 70 times, the viewability of that particular ad unit is 70%. Different DSPs have different criteria for what constitutes a 'view', so in order to remove disparity from any comparisons in this piece, we'll disregard all social and programmatic engines and focus on what is quantifiable in Google Ads.

Google uses its 'Active View' technology and supports the IAB and MRC definition of viewability: a minimum of 50% of the ad is in view for a minimum of one second for display ads, or two seconds for video ads.

So how is viewability affected? There are cases when the users scroll past an ad quickly, or abandons a tab in their browser and may even have seen the ad, but not long enough for it to be counted as viewable. There are probably also cases where the ad is served below the fold and is simply never scrolled down to - on a CPC bidding model you'll be safe from being charged for this 'impression'. While these are natural reductions in your campaign's viewability, there are also more sinister ways in which viewability can be diminished.

Whenever money is involved, fraudsters and opportunists will try to profit using dirty tactics, and unfortunately, ad serving is no different. When ads are served through exchanges, unscrupulous characters will reliably attempt to benefit by committing ad fraud, defined as any deliberate activity that impedes the regular delivery of ads to their intended audience. The 2 main methods of collecting ad revenue fraudulently are 'bot traffic' (posing as human traffic) or 'domain spoofing' (misrepresenting low-quality inventory as high quality). There are also other methods and that list is likely to grow as ad exchanges figure out how to curb fraudulent activity, and fraudsters invent new ways of gaming the system.

This fraud, measured by Google as 'invalid clicks', currently accounts for around 1% of ad revenue according to The Drum, while other sources estimate it to be even higher. Although the tech giant refunds advertisers for invalid traffic, this traffic still costs money at the outset and exhausts client budget, contributing to a loss of viewability. It should be noted that all stakeholders also run the risk of never getting full transparency into how Google calculates invalid traffic - whether their methods are accurate is a debate for another day.

The numbers from Google

During any given day of ad serving over Google's search and display networks, the largest ad networks of their kind, invalid clicks account for a percentage much larger than you may think. The following figures are extracted from a finance client running ads on Google:

For search ads, invalid traffic hovers between 9-23%. All impressions are counted as viewable on the search engine.

 For display ads, invalid traffic hovers between 28-35%. Viewable impressions are measured at around 75% (3 quarters of all ads served are not seen).

Finding workarounds

In conclusion, there are a few interesting takeaways that leave more questions than answers. On any given day, between 10-35% of clicks account for invalid traffic, and Google apparently refunds it but seems unable to clamp down on their chain of inventory supply - click fraud wouldn't be happening if fraudsters weren't benefiting.

If these are the invalid traffic numbers from Google, then it would be interesting to see the real numbers from the less vigilant and transparent mobile DSPs, whose supply chains are more intricate, convoluted and easily manipulated.

While we may not have the tools to completely eradicate click fraud in 2020, the big players will certainly be looking at ways of coding at a higher level of click validation standards. Invalid traffic is something that we have to find ways of working around for now, and in the meantime, there are certain strategies to improve viewability on Google and other DSPs, with slightly different tactical approaches for each targeting niche.