How Advertisers use FouAnalytics to Optimize Digital Campaigns

I've been at this for a long time (26 years digital marketing, last 10 years hunting bad guys). I am finally encouraged; I am seeing good momentum now. Advertisers have been in-housing for a few years now, so it is more mainstream and they are more comfortable with it. It's "rational in-housing" too which means they are taking the pieces in-house that make sense to take in-house. Advertisers will leave the creative messaging and making of ads to their creative agencies. But when it comes to digital media, the whole point of "programmatic" was automation and efficiency (as opposed to letting agencies feast on undisclosed markups, illegal kickback schemes, and principal trading). It also makes zero sense to have over-worked, under-paid 22-yr olds buying your digital media with zero planning, zero strategy, and zero experience; no wonder the only levers they know to pull are "lower cost" and "higher clicks" kinda like playing a video game. Oh right, they ARE 20 year olds. That doesn't help you do better digital marketing; in fact it's making it worse, way worse. But sadly, that's how the majority of the $189 billion in the U.S. was spent in digital in 2021.

In-Housing is Mainstreaming, Especially with Analytics

Advertisers are getting smarter, and in-housing some key pieces, specifically buying and analytics. This way they can better control what they buy and they can see more clearly how a campaign is actually going, compared to the monthly spreadsheets they get from their agency. They can see if the agency is delivering on the basic specifications of the campaigns. From what I have seen, most are not. They don't even know most of the ads are used up in the overnight hours or go to fake or undesirable sites, for example. I don't blame the folks who work at the agency; they are learning on the fly, because all of this is new to them. What can an advertiser do if the reports they are given are not accurate and not detailed enough? They can't make good business decisions.

Advertisers previously paid for tools that didn't work well, and those tools certainly didn't provide sufficient details to act on and troubleshoot campaigns. One fraud verification vendor said 5% IVT ("invalid traffic") and some other vendor said 1% IVT for the same exact campaign. No one can explain the discrepancy, despite both of them being MRC accredited and TAG certified.

Research shown in the screenshot shows that these technologies can't catch even the simplest things -- e.g. domain mismatches. From today's WSJ article: Ad-Tech Firms Didn’t Sound Alarm on False Information in Gannett’s Ad Auctions, "Firms that facilitate online-ad transactions had enough information to detect error that affected Gannett’s systems for over nine months." The IAS pixel was on the page, but it failed to detect the bid url was different from the page it came from. Duh. If they can't catch this, how can they catch determined bad guys that deliberately falsify -- i.e. mis-declare -- domains?

They don't do basic brand safety detection correctly either -- they blocked ads on the homepages of NYTimes, WSJ.com, and Washington Post because they contained the keyword "covid-19." Major advertisers and agencies have made the decision to let their fraud vendor contracts run out, and not renew. That's a LOT of savings, considering DoubleVerify and IAS together make an annualized $800 million in revenues.

"FouAnalytics is analytics for your digital media, just like Google Analytics is analytics for your website."

You were held back by previous blunt instruments; now you can upgrade your tools, and your digital marketing

What can an advertiser do if they had better tools and better data about their digital media and programmatic campaigns? Well, a lot. For example, in FouAnalytics, there's hourly detail so you can check if your ads and budget were used up between midnight and 4 am (see slide below). Monthly spreadsheets don't show you this level of detail. Even Google analytics doesn't give you hourly details in default reports. But if you can see this detail yourself, you can see whether day parting (setting which hours) and pacing (delivering impressions evenly) were set, or set correctly. Perhaps the agency forgot, or were too busy.

With FouAnalytics, you can also see where your ads went. That's important, right? Especially if you want to keep your ads away from breitbart.com or sanctioned Russian, Iranian, Syrian websites, and other sites on the OFAC sanctions list. On the left side of the slide below, an existing fraud vendor has large buckets named mobile in-app. They can't tell what apps the ads ran in, but yet they marked those as "99.992% fraud-free." How? On the right side is data from FouAnalytics on the same campaign. We can see the ads went to gambling, casino, lotto, scratch off, and other not optimal apps. The advertiser did not know where their ads went, so they could not act. If you can't see anything, you can't do anything. When you upgrade your tools, you know where your ads went. Then you can make business decisions to block them or continue to buy from them. I'd recommend blocking the fraudulent ones and crappy ones too, obviously.

"If you can see something, you can do something."

Make your campaign better in-flight; so you don't have to ask for refund afterwards

The following is a classic slide from 2015 where we can see the dark red (bots) at the start of the campaign. By removing the problematic sites, or even turning off entire networks that are problematic, you can see dramatic decreases in the fraud (red area). By week 3, you can see most of the dark red was reduced (fraudulent sites were removed). Today, it is more subtle than this because bots are more advanced, and much of the ad impressions have moved to mobile and in-app. The days of easy reductions of red bot traffic are in the rear-view mirror. But with FouAnalytics you can spot the bad sites and apps and add them to block lists, to make your campaigns better, while they are still running.

Budget-allocate your way to better outcomes

You can also judge the relative quality of your different paid channels. By having the FouAnalytics tags on landing pages, we can see the clicks arriving from all the different paid sources. For example, when you compare the red (bots) versus blue (humans) of the three paid display sources -- A, B, C -- you can easily understand it is better to reduce budget to A and increase budget to C -- i.e. optimize away from red and towards blue. You can also see that paid search is categorically better than paid display. Paid social (in this example, Facebook with FAN turned off) is the best of all, delivering the most dark blue (humans) clicks. And native -- Taboola -- is as crappy as we all know. Hopefully you will turn them off now. By allocating more dollars to better sources of media, you are showing more ads to humans (blue) and less ads to bots (red); you can thus budget-allocate your way to better outcomes.

As far as catching bots, why is FouAnalytics better than the other fraud detection tech? Those are all rules based, threshold based and parameters based. That means most bots are able to fake all browser parameters perfectly. The other fraud vendors wont have any discrepancies left to detect. FouAnalytics uses entropy analysis, which is based on looking at timings. It's hard for bots to fake multiple timings correctly to evade detection. For a more technical explanation, see: Why It’s Hard for Bots To Avoid FouAnalytics Detection.

Case examples from marketers

Over the years, I have helped many marketers optimize their digital marketing. In recent years, this is about buying better, and reducing the ad fraud. In the slide below, a marketer ran an experiment where they created a new campaign line specifying a very short inclusion list of good publishers. Remember, there are only a very small number of real publishers, with real content and real human audiences. Even though the CPM went up, the efficiency went up even more. This is where I also remind marketers that "cost efficiency" is a misnomer agencies have been foisting on you. In years past, you paid $30 CPMs. That's a "price" not a "cost." Today, you are buying ads at $3 CPM prices, but buying 10X the quantity. So your cost is still $30. You did not save any costs, compared to when you were buying $30 CPMs. But buying $3 inventory today is exposing you to rampant fraud; and you're buying ten times more. Go back to paying higher CPM prices, but buy far fewer impressions; that is how you can actually bring down your overall costs.

How would you like to double conversion rates?

Sounds magical? It's not magical, it's just practical. Marketers have real conversions on their sites -- e.g. ecommerce purchases. But those are finite. They calculate a conversion rate by dividing the number of conversions with the traffic they got from paid media. If the denominator is too large (due to bot activity) the conversion rate will be artificially low (left side of slide below). You can't stop the bots from coming to your site; but you can make your conversion rate calculations more accurate. How? By subtracting the portion that is bot traffic. In the slide below, when the amount of bot traffic was identified and accounted for, the conversion rate "doubled" from 7% to 13%. The latter is the correct conversion rate based on real conversions and real traffic. Some of you marketers have been suffering with low conversion rates, not because they were actually low, but because the analytics were not accurate. Get yourself better tools and better analytics.

So What?

Small and medium agencies have been using FouAnalytics behind the scenes to help clients for years. It may be too late for large agencies to get on board now that this Noah's Ark is leaving shore. In-housing is hitting it's stride.?Clients are ditching black box fraud detection and replacing it with FouAnalytics so they have their own digital media analytics in-house, just like they use Google Analytics for their own website. They don't even need to find new budget. They already have a line item for fraud verification; they are using a portion of that to "upgrade their tools;" better tools and more analytical details enable them to do better digital marketing.

"your best digital marketing lives are coming up shortly; take the first step"

Further reading: https://www.dhirubhai.net/today/author/augustinefou