It’s common for publishers to use a metric like CPMs to compare the performance of ad partners on their site. However, CPMs may not account for other factors that can impact a site’s overall revenue, and can cause confusion when higher CPMs don’t necessarily result in increased revenue.
What is CPM?
Cost Per Mille (CPM) is the amount advertisers are willing to pay for one thousand impressions on a given website. For example, a $2.00 CPM means that 1000 impressions at that rate would earn a payment of $2.00. There are many factors that play a role in determining CPM, like content, advertiser, geography, audience demographics, and intent. As a result, CPMs can vary widely by site and by advertising partner.
While CPM has its origins in the cost an advertiser must pay, the term is also commonly used to refer to the revenue that a publisher will receive for each 1000 impressions rendered on their website. If this value has already accounted for any fees or rev share from the advertising partner, it is commonly referred to as “net CPM”.
What does CPM not consider?
While useful, a metric like CPM doesn’t give a publisher a full view of the true revenue potential of their site, since it does not take into account important factors such as lost or unfilled impressions. Publishers can take these factors into account by dividing their revenue by all opportunities to serve an ad, a metric sometimes called effective CPM (eCPM):
Impressions can be considered lost or unfilled due to one of many reasons.
Setting floors on segments of inventory can be a useful way to push buyers to pay higher prices, but if the floors are set too high, buyers may not be willing to pay and impressions can go unfilled. Publishers need to account for fill rate and monitor their overall revenue, with an eCPM as an average of all impressions, even the ones that earn $0.00 because they did not meet the floor.
For example, if 1000 impressions are requested and 853 ads are served to those impressions, then the fill rate would be 85.3%.
Prior to header bidding, many publishers used a “waterfall” method to pass back unfilled impressions from one demand partner to another. This system added latency and could more often result in a failure to serve an ad to some users. The introduction of header bidding helped to reduce loss, and ensure more equal opportunity for buyers to bid on an impression in real time.
Since much of the ad serving technology publishers use relies on communication of data between multiple companies, there are occasions where timing or other technical issues can cause ads to not fill, or fail to be measured properly. This can result in discrepancies where, for example, a publisher’s ad server may record 1000 impressions at a $2.00 CPM, but if the advertiser’s ad server records only 900 impressions at the same rate, the publisher’s payment will be just $1.80. In this case, the eCPM would be calculated as revenue ($1.80) divided by requested impressions (1000), for an eCPM of $1.80.
CPMs and A/B testing
Accurately measuring performance to include factors like fill rate becomes incredibly important when testing optimizations to your site’s monetization. Even if a test is not split at an even 50/50, it’s still possible to calculate an eCPM or “normalize” test results to compare across equivalent volumes of traffic.
1% share of traffic earns $200. Normalized to 100%: $200/(1%) = $20,000
99% share of traffic earns $18,500. Normalized to 100%: $18,500/(99%) = $18,687
This approach shows which strategy will yield the highest overall revenue, where a CPM metric may fail to account for lost or unfilled impressions on one side of the test.
Can you trust eCPM?
While we’ve used the term eCPM above as a metric that includes a consideration for fill rate, it’s important to note that there are various definitions for this term across different platforms in the industry. In fact, Google Ad Manager does not factor in unfilled impressions with the “eCPM” in their reports.
When working with multiple vendors and reporting platforms, it’s always best to find out how each metric is calculated before relying on its value. Remember, the revenue you actually earn is the metric that matters most of all.