The Era of Fraud Verification is Ending; the Era of Analytics is Here

When advertisers shifted budgets into programmatic media buying, and away from buying media from real publishers, ad fraud went mainstream, circa 2012. Large advertiser budgets in the billions could be diverted to millions of fake websites and no one would notice. Everyone wanted the large quantities of ads that were not possible on sites that had real human audiences, because human audiences don't grow very fast, as any real publisher can tell you. But sites with fake traffic can scale infinitely to absorb all of that "spend" that has to get spent. Marketers got what they wanted -- large quantities, low prices, and high clicks -- and most importantly "campaign delivered in full" (i.e. they spent all their money). Everyone knew shady stuff was going on, and the numbers involved didn't even pass the sniff test for the last 10 years, but no one wanted to ask harder questions. Everyone was gorging themselves on that river of money.

The rise of verification

A cottage industry of fraud verification vendors sprung up to take advantage of the situation, so they could gorge on some of that money too. They would pretend to measure for ad fraud; but they could not report too much. Otherwise, they'd be accused of fear mongering to drum up business. They also wanted ad fraud to continue so they can keep making money detecting it. And they served an even greater need, the need for CYA and CTB -- "cover your ass" and "check the box" respectively. Marketers needed to be able to continue buying large amounts of ads; if it were uncovered that large amounts were fake, they'd have far less to buy. The fraud verification vendors provided them the necessary cover. If anyone asked about ad fraud, they would have checked the box -- yep, we use this vendor or that vendor. If anyone pressed them further with concrete evidence of fraud, they could blame the vendor -- THEY said it was "free from fraud" so we kept buying; WE didn't know. These verification companies did very well for themselves. For example, Moat sold to Oracle for $800 million in 2017 after a short existence detecting viewability of ads. Now Double Verify and Integral Ad Science are both public companies, making $800 million in annualized revenues and sporting $3 billion and $1.7 billion valuations respectively. They made a lot of money, but did they provide any protection from ad fraud?

The complete, utter, documented failure of verification

If you've read any of my work over the last 10 years, you will remember the many cases and examples where these verification vendors failed to detect the fraud that was there, or detected IVT ("invalid traffic") when there were no bots. It was almost like they were trying to detect the wrong things. As recently as last month, two researchers exposed the fact that this verification tech failed to detect or notify clients about simple domain mismatches; and that had gone on for at least 9 months -- i.e. still, no one noticed. Billions of ads were transacted with incorrect domains and no one noticed or did anything. It was the job of these verification vendors to catch things like that -- domain mismatches -- and they couldn't even do that most basic thing. How would they fare against determined fraudsters and bot makers purposely evading or defeating their detection. Not well, at all. See: WSJ: Gannett Gave Advertisers Inaccurate Information for Nine Months and WSJ: Ad-Tech Firms Didn’t Sound Alarm Even Though They Had the Data, Why?

The complete, utter, uselessness of accreditation

When advertisers started questioning the effectiveness of fraud verification and the numbers they produced -- like 1% IVT -- the vendors all hid behind "trust us, we're MRC accredited and TAG certified." That would shut the advertiser up for a while longer so the vendor could continue to make money off of them. But advertisers are realizing that accreditation is like a protection racket for the protection. Accreditation and certification are just as useless as the fraud verification itself, because the MRC and TAG have no tech, no data, no experience, no feedback loop, and no "answer key" to know if the vendors actually measured for the bots correctly. The MRC gives accreditation after an Ernst & Young accountant interviewed the fraud vendor and went through a checklist of what they would measure. Yeah, they have no tech and no way to know if the vendor actually measured the bots correctly. TAG is even worse, giving out "certified against fraud" plaques for self-attested paperwork and fees paid. Sounds a whole lot like "pay to play" doesn't it? And did you notice that none of the verification vendors caught or acted on the Gannett mess-up; that means they don't even meet the minimum standards of IVT detection needed for MRC accreditation (i.e. detecting basic IVT). So far, two months after that news broke, the MRC and TAG have not revoked their certifications for their utter failure, or even taken any action at all, like putting them on notice for their failures. Sounds like they are all in it together, right?

Rightly, advertisers no longer trust fraud verification

The fraud verification vendors use black box tech; that means no one knows how it works. Even the vendor itself doesn't know how it works. Try asking your account manager any question, like "why was this marked as IVT" or "why was this not marked as IVT?" They have no clue. Bob Hoffman wrote, "The exasperating part of this is that there is virtually no way for the recipients of the reports to verify the accuracy of the reports."?That is true of the reports that the ad systems and agencies give to their clients. It is also true of black box fraud detection tech too, because they have no way to verify the accuracy of those reports either.?How can you trust those reports? And the verifiers of the verifiers are also opaque, not transparent. How do you trust that MRC accreditation and TAG certification does anything, other than being yet another tax on your dollar?

Without getting into technical details, let's just look at what these vendors' reports say. Any advertiser who already pays for these services should ask to see their reporting (oh, you didn't get any reporting before? you should do so immediately). Examples from these vendors' reports (screen shots above) show 1% "ad fraud rate" from the vendor on the left and 0.01% fraud rate from the other vendor on the right. Looking at those numbers, ad fraud seems very low, until you realize that the 1% is all they could catch; and the other 99% represents their failure to detect anything wrong or their tags being blocked or stripped out. You heard that right -- the 1% fraud is all they could detect. How is that possible?

Rightly, advertisers are ditching fraud verification

You can easily surmise that bots will do everything they can to avoid getting caught. They can falsify the measurements, like Newsweek did with viewability. But it's even easier just to block the detection tags entirely. This means the fraud verification vendors are not detecting the fraud and bots get through undetected. The fact that these vendors are labeling the other 99.992% "fraud free" is absurd. How can they label a large bucket called "mobile in-app" as 99.992% fraud free when they don't know which apps were in the bucket? Further, how can they label a row called "N/A" to be 99.991% fraud free when they don't even know what site the ads went to? "No data" does not mean "no fraud." Duh. Anyone paying for these services should realize that "99.99% fraud free" is not true. You should stop wasting money on fraud detection that doesn't even work because they have a blind spot of up to 99.99%. I've been saying this for years, and no one would listen to me.

Advertisers I know are waiting for their fraud verification contracts to expire and don't plan to renew. They are taking the savings and shifting it to better and more complete analytics, to see for themselves. They are no longer stuck with blunt instruments; and they no longer have to trust black box tech where they can't know if the tech worked or if the numbers are correct. They knew for years there was something wrong because different MRC accredited vendors gave different fraud rates on the same campaign, and none of them could explain those discrepancies.

Rightly, advertisers are shifting to analytics

Now, ask yourself, are you paying those vendors for "cover your ass" purposes or are you hoping to get sufficient details about your digital media so you can see where your ads went, how much fraud and bot activity impacted your campaigns, and then take action based on what you can see? Those fraud vendors' spreadsheets and reports pictured above don't help you with any of that. Their giving you a number like 1% IVT or 5% IVT doesn't help either, because the campaign is over and your only recourse is trying to ask for a refund. Look where that got Uber. They are suing 100 mobile exchanges for outright fraud; the suit was filed in 2017. It's 2022 now and even if they win the lawsuit, most of those mobile exchanges no longer exist; they made off with the money and got away with it.

Specifically the era of digital media analytics, where advertisers have their own teams monitor and optimize campaigns, even if agencies continue to buy the media for them. You don't have to use FouAnalytics, you can use any other analytics. But FouAnalytics was purpose-built to measure digital media. I spent the last 10 years building it and I continue to tune the algorithms used for detecting fraud. Just like Google Analytics is analytics for websites; FouAnalytics is analytics for digital media.

The era of analytics is here

It has taken ten long years. But we are on the precipice of change now, catalyzed in some ways by the pandemic of 2020. Advertisers paused their digital ad spend. As they started turning digital media campaigns back on in 2021, they were more cautious and deliberate. They did more audits and asked more questions. They also continued the in-housing trend started in 2017-18. By having analytics as a core function and skill in-house (e.g. 1 FTE ("full time equivalent") the advertiser has the ability to see what is actually happening in their own campaigns. Most will have their media agencies continue to buy the media for them; but they will no longer be stuck with inferior tools -- blunt instruments -- like monthly spreadsheets and dashboards that just tell them the percentage fraud. They will have better tools -- specifically analytics -- versus black box fraud detection. With analytics in their digital media they can see better, so they can do better.

Here are a few examples of what advertisers and marketers have been able to do themselves with FouAnalytics, monitoring their digital media -- https://www.dhirubhai.net/pulse/fouanalytics-just-fraud-detection-its-better-digital

As you will see, the era of fraud detection is coming to an end. The era of advanced analytics in the hands of ad buyers is here. Some practitioners have been with me and have been using FouAnalytics to help their own companies or their clients for years now.

This Noah's Ark is setting sail. Are you on board? Come along.