Supply of What?

Why are online CPMs so low?

Why is the reader of an article on worth a third of the exact same person reading the exact same article on paper?

Dumb question. Everyone know that CPMs are low because there is an excess of advertising inventory: supply and demand.

This can’t be true.

Let’s say you had a newspaper–like in the old days–and you sold a quarter page ad for a $25 CPM. And let’s say that advertisers thought that was a fair price and–despite not being able to measure it very well–thought it provided the right ROI. You had 20,000 readers in your little city, so you made $500 off that ad space every day. Now let’s say a competitor comes into your city with another newspaper, one that remedies your woeful sports coverage. Your competitor takes half your readership and sells ads just like you.

What has happened to inventory here? There are twice as many newspapers, so has it doubled?

No. Even though there are now twice as many ad spaces, they each get half the number of impressions. So the inventory, the total number of impressions, stays the same. Of course it does! Unless the readers are reading more news in total, the ad inventory has to stay the same: the inventory is the consumers’ attention.

(Of course, if CPMs stay the same, your newspaper now makes half as much because it has lost half its readership, but that’s a completely different problem.)

The same is true online. There is not appreciably more inventory in the world than there was ten years ago. In fact, in a world where we collectively spend less time with media, inventory is contracting. Online inventory is increasing because the amount of time spent online is increasing. But the supply of inventory per user is constant, and demand per user should be constant, so supply and demand should stay matched. Plummeting CPMs are not a supply and demand problem.

So, then, why are online CPMs so low? If it’s not supply and demand, what is it? The answer has to be either: (1) the market for online ad inventory is fubar, or (2) online ads just don’t work very well. I suspect it’s a bit of both, but this whole supply and demand argument has just got to go.


  1. Jerry, I’d have to completely disagree, primarily because the concepts of communication and media have crossed over.

    Take sports:

    You might have spent X time reading about sports in your local paper and Y time (where Y is at least 3-5 times larger than X) talking with your friends in a cafe, phoning them up, writing them letters and sending them faxes about sports.

    Before the Internet communications was not ad supported at all and now with the Internet it is.

    Also, even with the media component, demand that wasn’t met before (because of the cost to publish and distribute) is now being met, so people are reading more about sports than the one or two articles they did read in the local papers.

    So the number 1 factor in all of this is that communications is now trying to be ad-supported (i count web e-mail, forums and social networks in this). And paying a lesser role is that more demand for media time is able to be met as well.

    The net-net is a huge rise in ad supported page impressions. For instance from 2005-2007 I think ad impressions on the Internet tripled (source:adrelevance). The audience and usage didn’t triple but the number of ad supported pages did. And they tripled because of just a few sites (MySpace, Facebook and Youtube). The other sites grew too, suggesting that more ad inventory *was* created in aggregate.

    p.s. How’s that theory on mortgage lead prices holding up? :)

  2. Interesting hypothesis… I wonder if the report showing our media consumption being flat to down includes substitutes, like IM substituting for phone calls.

    I think your point is that if there are more impressions and we aren’t spending more time with media, then that means we are spending less time with each ad. But isn’t that the same as a newspaper selling 16 ads per page instead of 4? Of course the CPM goes down.

    I still think my basic question–why is there a difference between the CPM for a reader of an online NY Times article and the CPM for a reader of the same article offline?–stands. If your answer is that they’re selling ten ads next to the online article versus two offline, then I buy that, but revenue per user per article should still be the same… unless the reason is something other than supply and demand.

    What have mortgage lead prices done? I haven’t seen recent data. I am willing to admit that I was wrong that financial services companies would continue to advertise. I did not, 18 months ago, foresee the historic collapse of the international financial system. I am willing–if I was wrong–to take a back seat in prognostication ability to those who saw that coming, all three of them.

  3. This is very interesting, Jerry.

    Fundamentally, I agree with you that an impression is in essence an impression and should be valued the same as all other impressions. However…

    …media outlets have done a very good sales job over the years of convincing buyers that some impressions are better than others and hence deserving of higher value (for which the manifestation here is higher CPM).

    Generally the “better” impression argument comes down to context (i.e., I have a better environment in which to grab the end user’s attention). I suspect that media buyers inherently, and perhaps unconsciously, still value traditional media higher than electronic media, maybe because of the argument made by the content owners, maybe because of the perception of relative scarcity (it cost money to print the nth paper).

    Broadcast tv has used another variant of this argument — to wit: since we can deliver more eyeballs at one sitting than any of our cable competitors, you the advertiser should pay us a higher CPM for that instantaneous audience aggregation…hence broadcast tv CPM’s have always outpaced cable CPM’s, even these days as the ratings move towards one another.

    It will be interesting to compare CPMs on Hulu and other over-the-top television plays, as those services scale to see whether these CPM’s move towards the CPM’s for the same content delivered over the air.

Comments are closed.