Programmatic has been problematic from the start

For those of you who have read my articles and seen my data over the last 10 years, virtually none of what follows will be new to you. But for those who have been misled by the Association of National Advertisers and many adtech companies that have been trying to preserve the status quo (and their revenue streams) from the truth, what follows may still be new to them.

What happened this month (July 2023) -- from the Canadian government stopping ad spending on Meta/Facebook to the EU Parliament going back to audit three years of YouTube ad buys -- tells you that after a decade, fraud is still as rampant as 10 years ago; but we are finally starting to acknowledge that ad fraud is a far bigger problem than we were led to believe. Keep in mind that ad fraud is more than just bots, even through the primary form of ad fraud 10 years ago was bots.

A look back

Let me take a step back and explain why programmatic has been problematic since its inception.

In the early days of digital advertising, advertisers bought ads from publishers like New York Times, Gannett, etc. They could meet with the salespersons of the publishers and negotiate media buys. When programmatic ad exchanges came along, ad buyers were dissociated from ad sellers by the ad exchange. Ad buyers now paid the ad exchange to place ads across hundreds of thousands of sites that they never visited or inspected. This is what enabled ad fraud to proliferate in a way that was simply not possible before. Never before was it possible for websites that were less than 10 days old to sell ads to large advertisers. Programmatic ad exchanges enabled exactly this. Fraudsters could now set up tens of thousands of fake sites using stolen content and wordpress templates, copy and paste some code, and start selling ads in a matter of days. It was like the money spigot just got opened, wide.

Brian O'Kelly, widely recognized as the "father of programmatic," should also be known as the naive kid, who stupidly opened Pandora's box of fraud that took digital advertising off-course for the last decade (2013 - 2023). Put yourself in the shoes of opportunistic fraudsters who were gifted this golden goose. Who wouldn't set up as many websites as possible to make as much money as possible while it was possible. If you didn't have any traffic coming to your websites, you'd create it yourself, by repeatedly loading the pages or using python scripts to load webpages or buy the traffic from traffic sellers. Ad revenue was driven by numbers of impressions, because everything was sold on a CPM (cost per thousand impressions) basis. So you did everything you could to make more money by inflating impressions and pageviews. This included the use of "bots" -- software programs that could repeatedly load webpages. It was like magic, the money just rolled in.

By 2015, everyone knew ad fraud was so rampant that better controls and verification were needed to help reduce or prevent it. Thanks again to BOK and AppNexus for stupidly/naively publishing a datapoint to claim credit for themselves -- that 92% of the ads they sold were so obviously fraud, they themselves could tell, without using any other specialised fraud detection tech. None of the advertisers that bought this crap from them got any refunds, so AppNexus doesn't get credit for doing the cleanup. Note that as far back as 2015, we already had data showing that programmatic was greater than 90% fraud. My data corroborated this.

What happened next I didn't predict correctly; but in hindsight, I should have 100% seen coming. At the time, I was optimistic that with the rise of "fraud verification" vendors, we would actually see a decrease in ad fraud. But alas, the exact opposite happened.

The ANA had already published their first "bot fraud" study with WhiteOps in 2014. I was helping to draft the 2015 report. Without getting into the gory details, the ANA forced us to rewrite the draft 8 times, in essence cherry-picking the data so it would appear that the overall average fraud rate stayed the same, instead of publishing the truth (that fraud went up). The chart (and conclusion) we ended up publishing in the second bot fraud study was one that showed some campaigns went up in fraud, while other campaigns went down in fraud, so the average stayed roughly the same, from 2014 to 2015. I should have seen this as a sign of things to come.

In 2016, I sat in Bob Liodice's office with two other gentlemen ("witnesses") and described what I saw in my data -- that ad fraud was rampant, and existing verification vendors' tech was not catching most of it. I also discussed other forms of fraud that were appearing that were not just fake bot traffic. Some publishers added other techniques to further multiply their own revenues, "because they could." For example, when a bot loads a single webpage, why stop at 5 ads on the page. Back then, we documented sites with 89 ad units on the same page. Why stop there? Why not refresh the ad slots every 1 second or stack 15 ads on top of each other in the same page? These fraudulent techniques inflated the key number that drove revenue -- numbers of ad impressions -- for the website owner. While the ANA was stupidly/naively focused on viewability, fraudsters were technically selling 100% viewable ads, because they refreshed the ad slots at exactly 1 second, which met the standard definition of "viewable ad" exactly.

Instead of going down, ad fraud continued to go UP due to the actions and inaction of the ANA and verification vendors. These vendors' tech was tuned for looking for bots, and was not well equipped to keep up with the rapid innovations of the bot-makers, attempting to avoid getting caught. Instead of being open to what they didn't know, the ANA doubled down and started sending out press releases that said fraud is 1% and well under control. In 2016, they launched TAG (Trustworthy Accountability Group) and asked Marc Pritchard to announce on stage that he would not buy from any vendor that is not TAG certified. This created an artificial rush to get TAG certified. But TAG was not even fully formed at the time -- not only did they not have any tech with which to check that fraud was measured correctly, even their paperwork was unfinished. TAG plagiarized feedback from vendors and used it to cobble together the first version of what went on to become the self-attested paperwork they use to "certify" vendors. This was a complete farce. TAG certification did not and does not reduce fraud. It quite literally has nothing to do with it, except provide the appearance that something is being done about ad fraud. The ANA sent out press releases year after year, saying that fraud is 1% and claiming credit that their programs like TAG caused fraud to be so low.

For years, this "transparency theatre" continued. And it continues to this day. Just this month, July 2023, Google cited DoubleVerify and Integral Ad Science numbers in an attempt to refute the 200 pages of damning evidence from Adalytics, which showed they sold billions of video ad views to over 1,100 advertisers over the last 3 years that were mis-represented as TrueView. None of the verification vendors detected these problems over the course of the 3 years, despite their being paid to do just that. One other important detail that is finally being reported and talked about is the fact that these vendors -- DV and IAS -- did not measure anything on YouTube or GVP (Google Video Partner sites and apps) directly with their own javascript tags. They were simply given data by YouTube and "performed calculations" on that data to provide reporting on viewability and fraud. Everyone now realizes this is not "independent verification" and is yet another example of "transparency threatre" that has covered up the truth for years.

Without getting into the weeds of their inability to catch most of the fraud, the point I will make here is that these verification vendors were not paid to detect fraud correctly. They were paid to be part of the cover-up -- i.e. the transparency theatre that misled advertisers into continuing to spend large sums of money on 100's of billions of ads that were not real, but not marked "invalid" by these legacy fraud verification vendors. The 1% IVT/fraud these vendors report is not all the fraud there is; it's all the fraud they can detect. These vendors's numbers are the numbers cited in the ANA's annual press releases and THAT has continued to mislead advertisers into thinking the problem of ad fraud was low and under control, when in reality ad fraud is at its highest point ever - both in dollars and in quantity of impressions. Everyone in the ecosystem benefited from the large volumes created by ad fraud, so everyone continued to look the other way. The annual press releases by the ANA made it OK to look the other way; and the inept technology of the verification vendors provided "cover" for the rampant fraud to continue, because their reports made it appear to be only 1%.

Hopefully now everyone realizes 1) it is not ok to trust the vendor selling you services to "grade their own homework," 2) it is equally useless to trust verification vendors that are given data to "perform calculations" and provide reporting, and 3) certification bodies like TAG, set up by trade bodies like the ANA, are equally useless in actual transparency, and in fact have further covered up the truth and misled advertisers into wasting tens of billions dollars more, over more years.

I'll end this article here by reiterating that "programmatic has been problematic" since the beginning. Fraud has been greater than 90% since the beginning and the actions and inaction of the ANA and other adtech vendors have covered up this truth for years. All of the dollars going to fake sites and fake impressions are dollars that flowed away from real mainstream publishers over the course of the last decade. That is why we are at the height of the current "crisis in journalism" with fake news sites and content overwhelming real news. Will advertisers finally take a stand and "vote with their dollars?" Will advertisers finally open their eyes to the reality of the situation and not be misled by trade associations and adtech vendors any further? I have been witness to the harms brought on by inaction and deliberate obfuscation over the last decade. Will you join me in turning around "digital?"

I will leave you with one final question -- have I EVER tried to sell you on my tech or my services? Or have I presented evidence and documentation of the phenomenon of ad fraud, over the last decade. You are welcome to use the tools that I developed for myself, entirely funded out of my own pocket over the last 10 years. In 2012, when I started the study of ad fraud, I decided not to become a fraud detection company, because I realized those companies rely on fraud to continue so they can keep making money. After years of using the tools myself to audit campaigns and deliver recommendations via powerpoint, I decided to open up the platform in 2020, so others can use it themselves and keep tabs on their own campaigns. I gave it a name -- FouAnalytics -- and set up "request an invite." FouAnalytics is free to use for small and medium businesses. For truly large clients, I can't afford to absorb the costs of servers, so I charge an annual subscription to use the analytics platform, just like Microsoft Office or Google Analytics Enterprise. I am outside the current ecosystem and have no incentive to preserve the status quo. In fact, I want to get digital advertising back on track.

Will you join me?

Aloha from Volcano National Park, Hawaii. It's not often that you get to walk across the entire crater of a volcano (path below). I am back in action the week of August 7th.