First Direct Evidence that Bot Activity Drives Lower CPMs

Advertisers are addicted to large quantities of programmatic ad impressions, low CPM prices, and high click rates. All of this is possible due to bot activity. Human time online, on social, and on mobile are all maxed out since 2012-13. Even with parallel consumption (watching TV, using laptop, and using mobile all at the same time) there aren’t enough humans on earth and time in a day to account for the 10s of trillions of impressions bought and sold in programmatic.

Bot activity on long tail sites that few to no humans know about is what allows those sites to sell enormous quantities of impressions. They have no cost of content; and they are simple “arbitrage plays.” If they can buy traffic for $1 CPMs and sell ads for $10 CPMs, they pocket a nice $9 of pure profit.

The data grid below shows the first direct evidence that bot activity enables a 40-50% decrease in CPMs, compared to when bot activity is excluded; said another way, when bot activity is excluded, the CPMs appear to double (in the cases where there is the most bot activity). The red lines below show the largest deltas between the CPMs in “bot excluded” data set versus the CPMs in the “with bots” set. For example, when bots are excluded from Android app CPMs the CPM is $6.87 versus $3.95 (when bot activity is mixed in). For Windows desktop, the CPM difference is also large — $12.81 versus $6.99. The largest details indicate where bot activity is most pervasive. Small deltas mean relatively low bot activity is present. This corroborates previous observations that bots focus on the highest margin categories — like Android app and Windows desktop (Chrome) because they can earn more money that way. Bots don’t pretend to be iOS or Safari as much because those earn lower yields for every unit of activity.

Marketers should realize that when they buy low CPM ads through programmatic channels, most of their ads are shown to bots. That is why the prices are so low. If you excluded the bot activity, or bought enough ads to be comparable to if you bought direct from good publishers in the first place, you end up not saving any money. You may have paid even more. If you buy from good publishers, the unit prices (CPMs) are higher, but you don’t need to buy as many ad impressions because those ads are shown to humans in the first place.

If you are currently using an IVT detection tech solution to detect invalid traffic, note that they are missing most of the invalid traffic and fraud. So you are deluding yourself into thinking your campaigns are working. You also need FouAnalytics to monitor for humans. That way you have the analytics to tell you what portion of ad impressions are shown to humans, not just what portion is shown to bots.

If you don’t understand the above, email me, and I can explain further.