Advertisers Replace Moat, DoubleVerify, IAS with FouAnalytics

Not just me saying this ... advertisers are questioning whether the legacy verification and brand safety vendors have even been doing their job -- i.e. detecting fraud and bots.

I won't lament their dashboards, excel spreadsheets, or PDF reports. Advertisers have all seen these legacy verification vendors' outputs which mostly say 1% IVT or 99 - 100% "fraud free." Ahem.

I also won't talk about whether their tech "catches all the fraud out there" or whether the 1% is "all the fraud they can catch." Ahem.

And I won't speculate on when was the last time their algorithms were updated or whether their data scientists and back-end developers ever reviewed campaigns with actual clients. Ahem.

But I WILL show you (below) what dashboards, charts, and supporting data grids look like in FouAnalytics and how advertisers use FouAnalytics to see, understand, and optimize their digital campaigns better. This way, you can compare to what your legacy vendors were giving you for years and "decide Fou yourself" whether you want to keep using those old tools.

G-IVT ("general invalid traffic")

This is where the bots tell you they are bots or are in the IAB list of bots and spiders. You won't see the following details in Google Analytics, DoubleVerify, IAS, or Moat reports. The following is an example of G-IVT:1 ("is G-IVT") and iab-bot:1 ("is in IAB bot list"). Note the supporting details like platform:Linux x86_64 (which is a server operating system) and HTTP_USER_AGENT:HeadlessChrome or SnapchatAds crawler. This way you can understand why these are color coded orange in FouAnalytics ("declared bots" where we can see their name in the HTTP_USER_AGENT).

S-IVT ("sophisticated invalid traffic")

These are the bots that don't declare themselves and, in fact, do everything they can to disguise themselves and avoid getting caught. Note the HTTP_USER_AGENTS are mostly correct -- i.e. the latest versions of Chrome like 121, 120, 119, etc. These bots also rotate screen resolutions, window sizes, etc. They can even fake the referrers so the traffic appears to be coming from whatever the botmaker wants to show (see the referrer grid below). Again, you won't see the supporting data grids below in Google Analytics, DoubleVerify, IAS, or Moat reports or dashboards. More advanced botnets also bounce their traffic through residential proxy networks to make their traffic appear to be coming from homes. That means IP address based fraud detection is NOT going to catch those bots (because their IP addresses are not from obvious data centers). And verification vendors that use HTTP_USER_AGENTS are easily fooled because bots can declare accurate, latest version user agent strings to trick those legacy vendors.

Programmatic ads verification (measuring where the ads went, and if bots/fraud caused it)

Advertisers copy and paste FouAnalytics in-ad tags into their DSP or ad server to measure where their ads went and if some form of fraud caused the ad to load. See: How to place FouAnalytics tags

By quickly identifying the most egregious bad guys -- bad sites and bad apps -- advertisers can block them by adding them to a block list. In the slide above, you can see that after 5 days of measurement, the advertiser added bad sites and apps to their block list and immediately cut the obvious ad fraud in half -- reduced the dark red from 48% to 23%. See: How to use the Domain App report in FouAnalytics

Click verification (measuring the clicks that come from paid media)

In some cases, like Google search ads, Facebook display ads, TikTok, etc. we can't measure the ads because those platforms don't allow a third party javascript tag. So we place the FouAnalytics on-site tag on the websites and landing pages where the clicks go. This way, we can see the relative quality of clicks arriving from those various channels. The action that advertisers can take is re-allocate budget across channels -- towards sources that are sending more dark blue (human) clicks and away from sources that are sending more dark red (bot) clicks.

You can also compare sources directly in the FouAnalytics dashboard. If your click through urls have standard UTM_SOURCE, UTM_CAMPAIGN, and UTM_MEDIUM query strings, FouAnalytics automatically parses them for you and populates the data grids like the ones below. You will see that 28.8% of this site's traffic have UTM_SOURCE codes in the url, which implies that about 29% of the traffic to the site is from paid media. You will also see in the UTM_SOURCE data grid that facebook and gclid ("google click ID") represent 37.5% and 32.8% of those urls. The color coding in each row shows you they are already doing a great job of minimizing the dark red. But notice the rows that have a lot more dark red. Those are the ones that need further attention.

Viewability

If you've looked at viewability reports from legacy vendors you will have seen viewability reported in the 80 - 96% range. But how can ads be marked as "viewable" when legacy vendors didn't measure them with javascript tags (for data collection). Note in the excel below, looking at the numbers: 1) delivery site, app bundle, and app name were all unknown, 2) only 2% of the impressions out of 353 million were measured with a javascript tag, and 3) 84% viewable rate was reported.

In FouAnalytics, you don't have to take my word for it. You can interrogate the supporting data if you want. The "intersection" data grid shows you what percentage of the ad unit was in the viewport of the browser; "visibilitystate" tells you whether the browser was minimized or not and whether the tab was active or not. Those are ingredients in the measurement of viewability. There are other ingredients that we use and other ways FouAnalytics verifies the accuracy of the measurement. Note on the left side of the chart above, FouAnalytics sees viewable rates in the 0.3% - 47.6% range, not 80 - 95% as reported by legacy vendors with questionable tech and data collection.

Carbon emissions and MFA sites

What about carbon emissions and MFA sites? How does FouAnalytics handle those? Simple. By showing you the data so you can judge for yourself. MFA sites are the ones that load ungodly numbers of ads per page and use 100% plagiarized content, both text and images (see example below). Once you have this detail you can understand why we marked sites as MFA and fraudulent. You decide whether you want to spending your budgets on these sites.

What about prebid blocking?

FouAnalytics does not do any proactive blocking during prebid nor does FouAnalytics block sites and apps automatically. Both of these are the opposite of best practices, even though various legacy vendors have sold those services to advertisers. The reason prebid blocking doesn't protect you from fraud is because everything in the bid request is declared, and therefore easily falsified. Breitbart .com won't put their own domain in the bid request because they know they will get blocked. So they and all fraudsters lie about the domain or app passed in the bid request. The IP address and USER AGENT in the bid request is also easily falsified by bad guys, so none of the data in the bid request should be used for fraud verification and prevention. Prebid blocking is a useless waste of computing power, bandwidth, and dollars (vendor fees).

The best way to monitor your campaigns is to use post-bid javascript tags to measure where your ads actually went. That will tell you if the prebid filtering did anything (hint: they miss 99.999% of the fraud, still.). The data grid below shows how FouAnalytics catches spoofing -- the discrepancy between the url declared in the bid request versus the url on which the ad actually ran. With FouAnalytics in-ad tags you can save money not having to pay for prebid filtering, which is useless (e.g. it didn't catch the simple discrepancies below).

So what?

You've probably seen enough already (even though there's 12 years worth of content more to go). Are you going to be one of those advertisers left behind, stuck using legacy verification vendors' inadequate reports, excel spreadsheets, and dashboards? Or are you going to join the throngs of advertisers upgrading to FouAnalytics, so you can "see Fou yourself" what you've been missing?

Reach out to me, if you want to use FouAnalytics for your ads and sites.

Further reading: https://www.dhirubhai.net/in/augustinefou/recent-activity/newsletter/