Categories: Analysis

When does a $2 CPM beat a $4 CPM?

online ads conversion number analytic website traffic visitor statistic chart session report vector

Publishers are used to dealing in CPMs. They’re a convenient way to compare rates and an easy metric to work with.  However, they are only useful when making true like for like comparisons – something that is easy to exploit.

We love transparency here at OKO. We know that our approach works and encourage publishers to uncover the true value of every ad they run – even providing the tools to make that easier. Not everyone takes this approach and we often see CPMs used a tools to deceive rather than inform publishers.   That often goes something like this:

Shady ad guy: Hey publisher, awesome website you have there. Do you know that my network  “Mega Awesome Ads” can make you loads more money? What CPMs are you getting?

Publisher : I average about $2 CPM I guess

Shady ad guy: Two dollars?? Ha, that’s pathetic. Mega Awesome Ads  will pay you four! You can’t lose can you? Let me help you switch your tags today.

(some time later)

Publisher : Hello? Hello? I’m trying to find out why my earnings haven’t doubled. Anyone there?

How could this go wrong?

The problem is fill rate. Comparing your average impression CPM against a source that isn’t filling at 100% is folly. If a source is promising to double your CPMs and match your current fill rate without any other down-side then jump at it.  If however they are promising a higher CPM for a proportion of those impressions then you need to look at the numbers a bit closer.

You are meant to think “I might as well get paid more for SOME impressions”, but that is how the ruse works: You are already getting paid more for SOME impressions and you might get paid less for them if you switch. To understand that better, let’s look at some real impression CPMs.

What makes an average CPM?

It’s an obvious point, but all advertisers are not paying the average. To get your average impression CPM there are advertisers paying more and less than that amount and you might be surprised by how much the amounts vary.

AdSense provides the revenue profile report to give you some insight into how much variance there is.  It isn’t the most beautiful report, but let’s take a look at results for a publisher averaging just under our mythical $2 impression CPM figure:

This publisher is averaging $1.82, but has ad requests earning CPMs of up to $19.29. If you zoom in on the image, you will see that 45% of ad requests achieve an average CPM of over $3.81. A network offering $4 CPM might seem attractive, but it needs to fill at a high rate to compete.

How do you know if it is a good deal?

The best way is to test. Run any new demand partner of a proportion of impressions and compare the revenue per request. I’ve written before on how to do this accurately.

If you work with OKO and use OKO Lens, this can be done easily within your dashboard. If not then it is a little time-consuming, but relatively straightforward using an ad-server like DFP.   If you don’t use an ad server and don’t use OKO Lens, then your best bet is to closely monitor the revenue you are receiving for each pageview and try to adjust for changes in traffic patterns.  If you  are limited to that approach and trying to assess different demand sources, that might be an indication that it is time to add an ad server  into your mix.

Mat Bennett :