How to use the green volume bars in FouAnalytics

The bottom third of the time series chart in FouAnalytics is the hourly volume. As you can see from the example above, there's a 5+2 pattern. That means the 5 weekdays have higher volume of traffic than the 2 weekend days. That makes sense for a financial information site. You can also see the effect of Labor Day (Sept 2) when the stock market was closed (the traffic above remained low, like a weekend day).

How to use the green volume bars for fraud analysis

The above was measured on-site. You can also use the green volume bars in in-ad measurement to analyze for ad fraud, or to confirm that something is real -- i.e. makes sense. Have a look at the following. This is in-ad measurement. Which one shows a legit alarm clock app loading ad impressions, A or B?

Which one did you choose? Note the hourly volume bars. In B, the height of the green volume bars shows there's about the same amount of impressions hour by hour - from 1:00 am and then drops off by 18:00. In A, the largest volume happens at 6:00 am. If you isolate an app like an alarm clock app, A shows a pattern of hourly volume that makes sense for an alarm clock app, when humans are waking up and turning off the alarm. B, however, shows that the app was loading ad impressions throughout the day, which may make sense for other types of apps, but not an alarm clock app. With the above example, I can say I finally found a "legit" alarm clock app that is loading ads when humans are using it, rather than continuously throughout the day when the alarm clock app is not being used.

Ad impressions used up in overnight hours

In the chart below, you can see the green volume bars are the highest right after midnight. Most of the impressions were served between midnight and 2:00 am. And after 8:00 am, the green bars drop to zero. That means there's no more volume of ad impressions to be served during waking hours, when humans are online. This is not fraud, but this is also not optimal for the advertiser because their daily allocation of ads were used up in the overnight hours when not many humans are awake. Once you see this in the FouAnalytics data, you can go back into your campaign settings and add "day parting" to turn off the overnight hours and add "pacing" so your ads are more spread out across the waking hours of the day.

CTV ad fraud, ads served when no one is watching

In the following example, you can check the green volume bars to see CTV ads being served continuously throughout the night, when the TVs are not on and humans are not streaming connected TV content because they are sleeping. You can even see an interesting example of CTV App A, selling "legit" inventory direct, but selling impressions through exchanges that run overnight. No one's looking, right? If you have FouAnalytics in-ad tags or CTV tags you can monitor this yourself and resolve both fraud issues and also non-fraud issues like ads running in overnight hours which don't do anything for your marketing outcomes.

Green volume bars provide context too

In the chart below, you see the huge amount of orange and red? Should you be alarmed? Nope. Note the volume bars are nearly zero. That gives you the context to understand that the large amount of orange and red in the stacked percentage (middle) part of the chart is not significant. Even a few declared (orange) bots can amount to a large percentage when the overall volume is low or zero.

Similarly, the spikes in orange and red (below) correspond to hours with low to no volume (green bars are zero). So those orange and red spikes are not significant.

And when you see a huge spike in green (chart below), you can immediately look at the top third of the chart to see what caused the spike in volume. In this case it was a massive spike in red (bots). So this warrants further investigation. You isolate the data and the hour and look at the supporting data to understand why it was marked as red.

And one final note, the dashboard is displayed in your local time zone, even though the data is recorded in UTC. Note the example below is from a UK retailer, but the data is shown in my time zone EDT Eastern Standard Time. The traffic to the site drops off in the overnight hours (volume bars are lowest) when humans are sleeping (and the lowest volume overnight hours are shifted to the left by 6 hours - UK time versus Eastern time). Each vertical line with the date is midnight 00:00 (displayed in my time zone).

Further reading: 756 more articles on ad fraud with screen shots and examples https://www.dhirubhai.net/in/augustinefou/recent-activity/newsletter/

Happy Saturday Y'all!