I propose a new online ad metric: seconds per impression

Online ad sales reps enjoy using the term “engagement” in their sales pitches. “OurSite.com’s users are highly engaged with our content, and so shall they be with your ads!”

What does “engagement” really mean? Some part is puffery — it’s an abstract term. Anyone can claim they have engaging content, but it’s hard to measure. But, despite my deep rooted discontentment with ad sales pitches, “engagement” does attempt to define a real need from media buyers: I want my ads in placements that demand attention from the end-user.

What if there were a way to objectively measure engagement? Here’s a simple suggestion: measure the average number of seconds displayed per impression (SPI) for a given placement. You could backdoor a site-wide SPI estimate from existing analytics data by taking ‘average length per user session’ divided by ‘average page views per session’ and expressing this in seconds/page view units. (Both session length and page views per session data are available in common analytics packages.)

SPI stats could be available for one specific placement on a particular page of a site and also aggregated site-wide to give an overall average SPI for an entire web property.

What would this do? It would provide a quick and easy measurement to validate placement bids expressed in CPM (cost per thousand impression) when comparing against other sites of a potential buy. You could even derive a cost per thousand seconds (CPMS = CPM / SPI). By using CPMS or CPM and SPI in tandem, you would get a uniquely different tool to help compare the potential value of sites competing for a buy.

Let’s say MySpace.com offers a $3 CPM for run of network ads (run of network = non-section specific, anywhere they have free space), whereas NYTimes.com offers $10 CPM for run of network. Inventing some numbers from intuition, let’s say that MySpace users spend 5 seconds per page (5 SPI) whereas NYTimes.com users have an average of 15 SPI.

NYTimes CPMS @ 15 SPI = CPM / SPI = $10 CPM / 15 SPI = $0.67 CPMS (Cost per thousand seconds)
MySpace CPMS @ 5 SPI = CPM / SPI = $3 CPM / 5 SPI = $0.6 CPMS (Cost per thousand seconds)

Wow, these numbers seem about right. We approach a $0.6 cost per thousand seconds of user-engagement for both sites, with a slight premium placed on NYTimes.com from higher quality content. In fact, when considering CPMS instead of CPM, NYTimes.com seems like a bargain: it’s just a shave more expensive than MySpace when calculated in CPMS, but its content is unquestionably a more premium placement for most brands than MySpace.

This metric has neat implications for the monetization potential of social networks vs. online video. You might disagree with my 5 SPI for MySpace, but I contend that it will not exceed 10 SPI. Compare that to YouTube, which could have a site-wide average of 60 SPI (videos can range up to 10 minutes — 600 SPI!). Let’s say YouTube prices at $10 CPM (run) (note: it’s surely cheaper than $10) whereas MySpace is still $3 CPM (run).

MySpace CPMS @ 10 SPI = CPM / SPI = $3 CPM / 10 SPI = $0.3 CPMS (Cost per thousand seconds)
YouTube CPMS @ 60 SPI = CPM / SPI = $10 CPM / 60 SPI = $0.17 CPMS (Cost per thousand seconds)

Obviously, YouTube blows MySpace out of the water. (Of course, other performance metrics should help with planning media buys too too – historical CTR, APM, CPA, etc. This would be a nice standard addition to the media planning toolbox.)

“Oh no!” A social network ad salesman frets upon reading this post, “When calculated with CPMS it is now embarrassingly obvious that social networks are a bad buy for most display advertising! What ever am I to do?”

Not to worry, social network ad salesman, most clients and agencies don’t give a shit about this stuff, they just want to see their pretty ads on the pretty computer screen.

So, keep using the old ‘engagement’ pitch with the wide-eyed recent college grads munching away at free Starbucks muffins in conference rooms of their conglomerate parents, and they’ll keep throwing tens of thousands of dollars your way.

This entry was posted in marketing, media and tagged , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , . Bookmark the permalink.

4 Responses to I propose a new online ad metric: seconds per impression

  1. Julius says:

    I think this is a much more interesting way to objectively analyze the “bang for your buck” for an online ad buy (which I think are of dubious quality and effectiveness). At the least, it’s an interesting spin on existing metrics for someone that is getting involved in online media delivery.

    On a side note, I think you might run into some resistance touting your C PMS until you come up with a new acronym 🙂

  2. kfarr says:

    @Julius: Ha! C PMS is funny now that I look at it that way. Oh well, I’ll keep it for now. (It’s not like anyone will really use this anytime soon.)

  3. Greg says:


    good stuff, i like reading about this thing that has become such a driving force in my life.I find it very complex and interesting right down to the actual events of each ride.You will do well


  4. kfarr says:

    @Greg: Thanks for the comment. I enjoy the job for the people and experiences. It’s nice that is pays rent too.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s