The digital ad industry knows there is a huge online advertising fraud problem: The Association of National Advertisers this week said global advertisers could lose $US6.3 billion next year in wasted money, buying ads that are never seen by humans, if current fraud rates continue at the same pace.
But while there has been extensive amounts of research conducted on the area of ad fraud what was unclear was which advertising sellers were responsible for poor quality inventory being sold online — and which ones had clean noses.
Today, Pixalate, a programmatic advertising analytics platform, has released its first monthly Global Seller Trust Index, which ranks programmatic advertising sellers (ad fraud is particularly prevalent in the area of programmatic advertising, because advertisers use software — rather than speaking to human salespeople — to buy their ads.)
Pixalate has looked at more than 100 billion impressions across the net, assessing each seller on areas such as viewability, fraud, engagement and domain masking (read the full methodology below.) And big names including Yahoo, AOL and Millennial Media are among the surprises sitting outside of the top 20 rankings.
Here’s Pixalate’s Global Seller Trust Index top 20 performers for November. Google AdExchange sits atop the rankings of 400 sellers, and there are some solid performances from big players like OpenX and Rubicon. There are also some lesser-known names in there ad execs might not expect”
But, Pixalate CEO Jalal Nasir told Business Insider the “biggest surprise” was seeing a bunch of well-known companies with top comScore rankings (based on unique visitors and % of the internet population reached) performing so poorly:
Yahoo — which was the second biggest ad network/buy-side network in October according to comScore, behind Google — ranked at just No.38. It received an average standing on most metrics but received poor grades in the fraud and domain masking categories. (Domain masking occurs where ads are served on web sites that are disguising the actual web address of the site.)
AOL — which owns the seventh-biggest network, Advertising.com — was ranked at No.29. It received a good standing overall, but the data reveals that a lot of long-tail sites with lower quality inventory were selling on the exchanges by the AOL Network, Pixalate told us.
Millennial Media — Nasir says the “biggest surprise” among the low rankings was publicly listed company Millennial Media, which reported $US69.8 million in revenue in its last quarter. It ranked at No.300 of the 400 companies analysed, due to a “high ratio of fraudulent traffic.”
Business Insider contacted Yahoo, AOL and Millennial Media for comment but didn’t hear back.
Other notable top ad sellers ommitted from Pixalate’s top 20 include Conversant (formerly ValueClick Networks) — which is ranked No.3 by comScore but just No. 122 on Pixalate’s trust rankings, due to a low score on fraud — and Burst — No.8 on comScore but just No.165 here, due to a low score on domain masking. (Both Conversant and Burst were contacted for comment by BI.)
Nasir told Business Insider it is hoped the rankings will encourage agencies and brand marketers to review their “archaic” domain whitelists. Not least, because Pixalate says 46% of ad inventory is masked, so advertisers are often not buying space on what they believe to be safe and reputable websites.
Any new and growing marketplace — including Uber now with all its challenges, Amazon, and other disruptive market technologies — have found their path to success by creating trust with an actionable rating system.
A rating system that successfully serves a marketplace must benchmark its suppliers on quality, not quantity. Trust is currently lacking in the advertising industry and it is being distorted by the classic reach metric. Big brand networks don’t necessarily mean high quality ad inventory.
RTB is under attack from a few bad or careless players and this is exacerbated by archaic measurement methodologies. Sellers won’t be able to game the system by showing classic reach or uniques metrics. This is a supply chain management problem and it will encourage good sellers to be ranked high rand lower ranking sellers to invest in their supply quality management departments.
Over the past couple of months, in direct response to the growing number of headlines around the advertising fraud problem, number of ad tech companies including AppNexus, DataXu have been offering clients “fraud free” guarantees (at a premium price). Rocket Fuel, on the other hand, has taken a different approach by launching a free tool called Traffic Scanner that allows marketers to track the quality of ad inventory their campaigns are being served against.
Pixalate’s Global Trust Index is primarily focused on companies whose main line of business is selling ad inventory. It plans on releasing two further indices, which will focus on website quality (which will no doubt shine a spotlight on popular platforms like Facebook, YouTube and Twitter) and a separate ranking looking at DSPs and buy-side platforms.
Here is Pixalate’s full methodology:
Global quality ratings are based upon a weighted analysis of overall effectiveness assessing viewability, fraud, engagement, domain masking, and network and inventory scores in compliance with recognised industry standards.
1. Network score: Based on domain quality represented by sellers on open exchanges. For instance, a seller with high quality domains rated among the top sellers for that domain will get a high score.
2. Inventory score: Determined using how much of the inventory sold is legitimate based on price, trust graph and multiple other factors. A seller that is correctly representing domains and getting inventory from legitimate sources will rate higher on this scale.
3. Fraud score: Pixalate’s proprietary fraud engine classifies each impression as being viewed by a human versus a bot. A seller with the lowest fraud score will earn the most points on this scale.
4. Viewability score: Based on Media Rating Council (MRC) standards, each impression is classified for true viewability. A seller with more viewable impressions is given a higher score.
5. Engagement score: For this scale, every ad served from the seller is evaluated based on user interaction as determined by mouse hover time, time spent on page and click. Seller with high engagement score is likely to have better performance.
6. Masking score: The masking score is based on how much of the inventory sold by a given seller is mislabeled intentionally. Fraudulent sellers use domain identity theft to hide the true identity of low quality inventory. A seller with a high masking score has the least amount of masked inventory in their supply.
Business Insider Emails & Alerts
Site highlights each day to your inbox.