Advertisers Take Back Control with FouAnalytics
Over the last decade, advertisers were held back by the verification tools they were using to monitor their digital ads. These vendors repeatedly told advertisers that fraud was 1%, don't worry, keep spending. Advertisers now realize that the 1% these vendors reported was all the fraud they caught, not all the fraud there is. Advertisers also realize these vendors were measuring between 1 - 10% of the ads and just assuming the rest were "fraud-free," audible, viewable, and brand safe. It's not good to assume, especially when the adversaries we are facing (the fraudsters) are clever and armed with advanced bots.
That said, when advertisers deploy FouAnalytics in-ad tags to see where their ads are going and what forms of fraud are causing the ads to load, they "can DO better because they can SEE better." The left side example in the slide below is one of those large campaigns that legacy fraud vendors reported as 1% fraud. It's actually 90% dark red (fraud/bots). Now that the advertiser can see clearly, they can take action to reduce the fraud and improve their campaigns. The example on the right shows a campaign that started at 48% dark red. By adding fraudulent sites and apps to a block list after 5 days of measurement, this advertiser cut the dark red in half, to 23%.
What follows are the "go do's" for advertisers of different maturity levels with respect to digital media buying.
Advertisers using block lists
Many of the largest advertisers continue to use media agencies of large holding companies to buy their media. In most of these cases, the media agencies are using a block list approach, even if they claim that they use inclusion lists. The reason advertisers now know this is because FouAnalytics shows new fraudulent sites and bad mobile apps showing up in the data constantly. If the media agency were actually using an inclusion list, how do these new bad sites and apps appear in their campaigns? It doesn't matter either way, because when advertisers have FouAnalytics measurement in place, they can see these bad sites and apps and have the agency block them.
To make this task easier, we created the Domain App Report tab. There may be hundreds of thousands of sites and apps in any programmatic campaign. But we only need to focus on the top 10 - 20 most "prevalent" ones first, to assess them for quality. In the Domain App report, you will see a section called "worst offenders" where we summarize the top 10 domains or apps, highest volume ones first. The "count" divided by the "visits in date range" is the "prevalence" percentage. Each of the subsequent columns shows a different type of fraud, problem, or risk, along with a percentage. In the example below, you can see some domains have high percentages of stacked ads, others have high popunders, and some have large percentages of "no GPU" (fake browsers in data centers). Once you have reviewed these summaries, if you choose to block the site or app, click the checkbox to the right of each domain or app. This adds them to the lists at the top of the page. Once you have compiled these lists, copy and paste them and send to your agency to add to the blocklist.
Go Do:
review the domain app report tab once a month, or more often as you like. Review the top domains and apps for fraud risk. Select the ones to block and copy and paste the list to send to your agency. I recommend adding sites and apps to a block list, waiting a week or so to see the change in the charts, and then adding another group of sites and apps to the block list. This way, you can see the changes. If you add to the blocklist continuously, you won't see the improvements as easily.
Advertisers using inclusion lists, PMPs
Some advertisers have already moved to inclusion lists. This is a good thing because there are simply too many bad guys to block in a blocklist. The large media agencies that tell their clients their block lists have millions of domains in it literally don't know what they are doing and are doing a disservice to their clients. That's because most DSPs can't process blocklists more than 10,000 rows long. Everything after the 10,000th row is simply ignored; that's why having block lists of millions of domains is completely useless.
Inclusion lists of 10,000 rows are also the maximum that most DSPs can handle. For advertisers using inclusion lists, ask yourself how many sites and apps do humans actually use, regularly. A simple thought exercise to do is to ask yourself to name the top 10 sites you visit every day, and the top 10 mobile apps you use every day. For most people, they can name 5 - 7 and then struggle to reach even 10 sites and apps they use every day. So while humans DO visit long tail sites and apps sometimes, there's not a huge number of humans. So the concept of the long tail with "at-scale" audiences is a myth that has misled advertisers' digital spending for the last decade. Again, you don't have to believe me. Just review the sites and apps detected by FouAnalytics for fraud and risk, and turn them off by removing them from your inclusion lists. There are many, many MFA sites still chomping on budgets in programmatic campaigns.
Some advertisers think that buying from PMPs ("private marketplaces") will help them avoid ad fraud. That's not the case if the publishers and sellers in the PMP are cheaters -- i.e. they buy bot traffic, do audience extension, etc. to juice their numbers.
Finally, a note on what you see in placement reports and log level data. Some advertisers assume that because breitbart, porn sites, and other bad sites and apps don't appear in their placement reports, they are safe from these. That might be the case. But, placement reports and log level data only report the domain or app that was declared in the bid request. Sites like breitbart[.]com know their domain is blocked, so they simply lie and put some other domain in the bid request, to get around the block. Ads still go to breitbart, and breitbart still gets paid, because they use their own sellerID in the bid request. Breitbart doesn't appear in the placement report or log level data. That's why you need a postbid javascript tag, like the in-ad tag from FouAnalytics, to detect where your ad actually went. If you see breitbart in the FouAnalytics data, but your campaign set up had it blocked, you know they lied and got around those blocks. If you see breitbart in the FouAnalytics data, but your campaign set up did not have it in your inclusion lists, you know breitbart lied and pretended to be a domain that was in your inclusion list.
Go Do:
To further reduce these instances of spoofing, even if you are using inclusion lists, consider using inclusion lists of sellerIDs, publisherIDs, or deal IDs, instead of domains and app names. This cuts down on the opportunity for breitbart or other fraudulent sites from easily spoofing, just by lying about the domain or app name in the bid request.
Bonus, if you want to do SPO ("supply path optimization") for free yourself, and not pay some other vendor useless fees for SPO, you can just tell your agency to UNcheck 42 of the 45 exchanges active in your media buy. You don't need 45 exchanges since they all sell and re-sell the same inventory anyway. You just need 3 exchanges to be able to buy every possible real impression. This also significantly cuts down on the "leakage" (your ads not going to the places you want) and increases the accuracy of your inclusion lists (see example in the slide above).
Optimize towards humans, not just away from fraud
For those advertisers already using inclusion lists for their digital media, they can also install FouAnalytics on-site tags on their websites and landing pages. This way, they can see the quality of the clicks coming from various paid media sources, including the ones like paid search and paid social where we can't measure the ads themselves.
In the slide above, you can see 3 paid display sources. You can see that A has more dark red (bots/fraud) and less blue, while C has more dark blue (humans) and less red. The simple action you can take to optimize your campaigns is to allocate more budget to C and reduce budget to A. This way you show more ads to humans and fewer ads to bots. With FouAnalytics, you can optimize towards humans (blue), not just away from fraud (red). And in some cases you can optimize across channels. For example, both paid search sources, D and E, and more dark blue than all of the paid display sources. So if you are able to re-allocate budget across channels or ad types you can also optimize towards more humans by spending more on those channels that deliver more humans to your landing pages. Paid social F has the highest dark blue of all the sources. You can invest more in that.
领英推荐
Go Do:
Add FouAnalytics on-site tags to your landing pages, so you can see the relative quality of the clicks arriving from various paid media sources. Since FouAnalytics reports on humans (other legacy vendors don't), you can not only optimize away from fraud, you can also optimize towards humans (dark blue). You can also re-allocate budgets towards channels that have the most dark blue.
Advertisers at a very good steady state
What happens if you have already optimized your media so well, your in-ad FouAnalytics charts look like the following. This is an example of a campaign managed by Goodway Group and practitioners Andrea Kwiatek Laura Taylor Allison F. Dark red is minimal, dark blue is very good.
Do you still need FouAnalytics? Yes! Just like you would not remove Google Analytics from your site when they are no bot attacks, you won't remove FouAnalytics from measuring your ads or your landing pages. You can use FouAnalytics to continue to monitor for problems. But you can also you FouAnalytics on-site tags to further optimize your digital marketing with something called attentiveness.
Not to be confused with "attention," attentiveness means how active humans are on your site/landing pages. Humans that deliberately clicked on your ads want to get more information, so they will do something else when they get to your site -- like move the mouse, scroll the page, click something, etc. More attentive humans means more conversions. More attentive humans come from better ad creatives, where the creative message inspired the human user to click through. Advertisers using FouAnalytics on-site have even used this "attentiveness" measure to gauge the effectiveness of other forms of media like CTV, and offline media such as billboards. See: How to do cookieless attribution with FouAnalytics for the methodology and examples.
Why is attentiveness better than attention or viewability? Viewability has to do with the ad itself, and whether 50% of the pixels were in the browser viewport for 1 second. It does not tell you if anyone saw the ad. Attention has to do with whether someone was "paying attention" to the ad. And of course larger ads that cover the screen get more attention than smaller ads that people may not even notice. But the problem is with measurement. There's no javascript code that can actually detect if someone was looking at the screen. So attention again has to do with characteristics of the ad -- e.g. larger ads get more attention, generally. Attentiveness, on the other hand, IS directly measured on the landing pages of advertisers. Attentiveness has to do with the percentage of human users that "did something else" on the advertisers' sites.
If you compare UTM_SOURCE=1 to UTM_SOURCE=2 in the slide above, you can differences in "attentiveness." For example Source 1 shows that 27 - 36% of the visitors clicked something on the landing page, while Source 2 shows that 70 - 79% of the visitors clicked something. Visitors from the latter are more "attentive" when they arrived on the landing page. More attentive users lead to more outcomes, even if the outcomes are not visible on the site (like ecommerce transactions). With attentiveness, we've shown that visitors that come to the site to look up more information after seeing TV ads or billboards lead to more completed purchases (except from 3 large advertiser campaigns below -- automotive, financial services, and tourism). See https://www.dhirubhai.net/pulse/how-do-cookieless-attribution-fouanalytics-dr-augustine-fou-ykwue for more details.
Go Do:
If you're an advertiser that has optimized your media/ads as much as possible already, you can start to optimize for greater attentiveness. Note that since your ads are mostly shown to humans, your ad creative has an opportunity to work. Better ad creatives can now lead to more attentive humans on your landing pages. Your creatives simply didn't have an opportunity to drive brand lift, recall, and loyalty, because previously your ads were shown to bots and not humans, even if the legacy fraud vendor said it was 1% bots.
This is how advertisers are taking back control of their digital media ("after a decade wandering in the wilderness"). With the right analytics in place -- FouAnalytics -- advertisers can not only monitor their own campaigns for compliance and transparency, advertisers can also better optimize for attentiveness and real outcomes.
Happy Saturday Y'all!